diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,49969 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 8324,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 2.894,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 3.0019,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 2.81,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 2.7314,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 2.783,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 3.076,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-07,
+      "loss": 2.8474,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 2.93,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 2.8097,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.0645,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-07,
+      "loss": 2.7881,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 2.8377,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-07,
+      "loss": 2.874,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8e-07,
+      "loss": 2.9909,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 2.7501,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2e-07,
+      "loss": 2.8209,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 2.8401,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 2.9457,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-07,
+      "loss": 2.6964,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 2.73,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 2.8011,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-07,
+      "loss": 2.8686,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 2.6424,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 2.9452,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 2.8218,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2e-07,
+      "loss": 2.674,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-07,
+      "loss": 2.7029,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6e-07,
+      "loss": 2.8422,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 2.8927,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 2.795,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 2.7053,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4e-07,
+      "loss": 2.9824,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.6e-07,
+      "loss": 2.787,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 2.7714,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 2.6498,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-07,
+      "loss": 2.7993,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4e-07,
+      "loss": 2.9047,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6e-07,
+      "loss": 2.8935,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8e-07,
+      "loss": 2.9095,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 2.8189,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 2.8549,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 2.6603,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.6e-07,
+      "loss": 2.7025,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8e-07,
+      "loss": 2.8764,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 2.862,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 3.036,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 2.6515,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 2.7183,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 2.9047,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.7998,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02e-06,
+      "loss": 2.7848,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.04e-06,
+      "loss": 2.7999,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06e-06,
+      "loss": 2.8986,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.08e-06,
+      "loss": 2.8251,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1e-06,
+      "loss": 2.9027,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.12e-06,
+      "loss": 2.8808,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.14e-06,
+      "loss": 2.91,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 2.7331,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 2.7588,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.7667,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 2.9032,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 2.7458,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26e-06,
+      "loss": 2.5394,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.28e-06,
+      "loss": 3.0551,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3e-06,
+      "loss": 2.7652,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.32e-06,
+      "loss": 2.8989,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.34e-06,
+      "loss": 2.813,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 2.975,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 2.5782,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 2.793,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.42e-06,
+      "loss": 2.8399,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-06,
+      "loss": 2.855,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.46e-06,
+      "loss": 2.8153,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.48e-06,
+      "loss": 2.861,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 3.0008,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.52e-06,
+      "loss": 2.6967,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.54e-06,
+      "loss": 2.7379,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.56e-06,
+      "loss": 2.7266,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 2.7166,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.7693,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 2.6919,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 2.7995,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 2.9599,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 2.7051,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 2.7666,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.72e-06,
+      "loss": 2.8841,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-06,
+      "loss": 2.7936,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.76e-06,
+      "loss": 2.7436,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 2.8614,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 2.975,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 2.8066,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 2.7571,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 2.6997,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 2.8759,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 3.0306,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 2.7032,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.94e-06,
+      "loss": 2.8026,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 2.6355,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-06,
+      "loss": 2.6331,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.7794,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.02e-06,
+      "loss": 2.7938,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.04e-06,
+      "loss": 3.0001,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.06e-06,
+      "loss": 2.7975,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.08e-06,
+      "loss": 2.7868,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 2.8132,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.12e-06,
+      "loss": 2.7565,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 2.7994,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-06,
+      "loss": 2.8481,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 2.6668,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2e-06,
+      "loss": 3.0088,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 3.011,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.24e-06,
+      "loss": 2.8221,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 2.7214,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.28e-06,
+      "loss": 2.792,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 2.6355,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 2.8261,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 2.7254,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 2.7874,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.38e-06,
+      "loss": 2.6549,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.6885,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.42e-06,
+      "loss": 2.7787,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 2.974,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.46e-06,
+      "loss": 2.6929,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 2.8407,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5e-06,
+      "loss": 2.911,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.52e-06,
+      "loss": 2.6967,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 2.9059,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.56e-06,
+      "loss": 2.834,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 2.6788,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6e-06,
+      "loss": 2.8869,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 2.7719,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.64e-06,
+      "loss": 2.8594,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 2.6831,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.68e-06,
+      "loss": 2.7681,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 2.605,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 2.8285,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 2.7449,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 2.7741,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 2.6963,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.8074,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.82e-06,
+      "loss": 2.9003,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.84e-06,
+      "loss": 2.6186,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.86e-06,
+      "loss": 2.689,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.88e-06,
+      "loss": 2.6509,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9e-06,
+      "loss": 2.5075,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.92e-06,
+      "loss": 2.8706,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 2.6303,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.96e-06,
+      "loss": 2.7687,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 2.6951,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-06,
+      "loss": 3.0126,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 2.7296,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.04e-06,
+      "loss": 2.6615,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 2.8486,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.08e-06,
+      "loss": 2.6163,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 2.5788,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.12e-06,
+      "loss": 2.851,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 2.7258,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 2.8291,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 2.7279,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.9104,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 2.6855,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 2.7472,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 2.8647,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 2.9263,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 2.6979,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 2.7195,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 2.8101,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 2.7477,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 2.8066,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 2.7861,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 2.8053,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.44e-06,
+      "loss": 2.7382,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.46e-06,
+      "loss": 2.6885,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.48e-06,
+      "loss": 2.7424,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-06,
+      "loss": 2.7374,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.52e-06,
+      "loss": 2.7216,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.54e-06,
+      "loss": 2.8457,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 2.7685,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.58e-06,
+      "loss": 2.8715,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 2.7292,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.62e-06,
+      "loss": 2.9648,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 2.7378,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.66e-06,
+      "loss": 2.9837,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 2.8348,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7e-06,
+      "loss": 2.8113,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 2.8178,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.74e-06,
+      "loss": 2.8594,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 2.7789,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 2.7139,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 2.7585,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 2.8334,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 2.6522,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.86e-06,
+      "loss": 2.6804,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88e-06,
+      "loss": 2.6061,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 2.6389,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 2.6688,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.94e-06,
+      "loss": 2.6974,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.96e-06,
+      "loss": 2.7438,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 2.818,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.5431,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 2.7117,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.04e-06,
+      "loss": 2.7474,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 2.6124,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.08e-06,
+      "loss": 2.8398,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1e-06,
+      "loss": 2.6281,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.12e-06,
+      "loss": 2.6267,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.14e-06,
+      "loss": 2.7716,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.16e-06,
+      "loss": 2.7426,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.18e-06,
+      "loss": 2.9914,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.7598,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.22e-06,
+      "loss": 2.6267,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.24e-06,
+      "loss": 2.6901,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.26e-06,
+      "loss": 2.9535,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 2.7835,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3e-06,
+      "loss": 2.641,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.32e-06,
+      "loss": 2.8062,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.34e-06,
+      "loss": 2.7327,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 2.6462,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.38e-06,
+      "loss": 2.7658,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4e-06,
+      "loss": 2.5678,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.42e-06,
+      "loss": 2.8125,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 2.7126,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 2.6183,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.48e-06,
+      "loss": 2.6963,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5e-06,
+      "loss": 2.6586,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 2.741,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 2.8272,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.56e-06,
+      "loss": 2.878,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.58e-06,
+      "loss": 2.8037,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 2.678,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 2.6924,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 2.739,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.66e-06,
+      "loss": 2.8372,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 2.7128,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7e-06,
+      "loss": 2.722,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 2.7681,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.74e-06,
+      "loss": 2.6312,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.76e-06,
+      "loss": 2.5889,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.78e-06,
+      "loss": 2.6731,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 2.6796,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 2.7552,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.84e-06,
+      "loss": 2.944,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86e-06,
+      "loss": 2.8513,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 2.6528,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 2.961,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.92e-06,
+      "loss": 2.5977,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94e-06,
+      "loss": 2.7888,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 2.5123,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 2.8207,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-06,
+      "loss": 2.8229,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.02e-06,
+      "loss": 2.8254,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.04e-06,
+      "loss": 2.7459,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 2.504,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 2.8549,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1e-06,
+      "loss": 2.9212,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.12e-06,
+      "loss": 2.7519,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 2.7328,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 2.9336,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.18e-06,
+      "loss": 2.8143,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2e-06,
+      "loss": 2.7595,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 2.8413,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 2.8455,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 2.8666,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.28e-06,
+      "loss": 2.706,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 2.7563,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 2.8858,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 2.7714,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.36e-06,
+      "loss": 2.6696,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 2.7685,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 2.8218,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 2.6513,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 2.5531,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 2.6738,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 2.6684,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 2.5771,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 2.7887,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 2.6957,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 2.7742,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 2.8841,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 2.6455,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 2.6645,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.64e-06,
+      "loss": 2.7753,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.66e-06,
+      "loss": 2.9213,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.68e-06,
+      "loss": 2.7834,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7e-06,
+      "loss": 2.6396,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.72e-06,
+      "loss": 2.8196,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.74e-06,
+      "loss": 2.661,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.76e-06,
+      "loss": 2.6563,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.78e-06,
+      "loss": 2.6019,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8e-06,
+      "loss": 2.7255,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.82e-06,
+      "loss": 2.7526,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.84e-06,
+      "loss": 2.7014,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.86e-06,
+      "loss": 2.8109,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 2.7619,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9e-06,
+      "loss": 2.7368,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.92e-06,
+      "loss": 2.8128,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.94e-06,
+      "loss": 2.7018,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 2.7541,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.98e-06,
+      "loss": 2.7203,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6e-06,
+      "loss": 2.7399,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.02e-06,
+      "loss": 2.7455,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 2.6941,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 2.9303,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.08e-06,
+      "loss": 2.7382,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.1e-06,
+      "loss": 2.7882,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 2.6581,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 2.7062,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.16e-06,
+      "loss": 2.602,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.18e-06,
+      "loss": 2.6892,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 2.5565,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 2.8248,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.24e-06,
+      "loss": 2.6882,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.26e-06,
+      "loss": 2.6782,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 2.6754,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 2.8579,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 2.6539,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.34e-06,
+      "loss": 2.5563,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 2.6639,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 2.6467,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 2.7478,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.42e-06,
+      "loss": 2.4947,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 2.5668,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 2.7168,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 2.8168,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 2.7267,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 2.6155,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 2.5881,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 2.5912,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 2.7586,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 2.7013,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 2.5506,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 2.9118,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 2.7971,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 2.7011,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 2.6823,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 2.6775,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 2.6323,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 2.6911,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 2.7132,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 2.7473,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 2.6943,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 2.7517,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 2.7072,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.88e-06,
+      "loss": 2.7432,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.9e-06,
+      "loss": 2.8832,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.92e-06,
+      "loss": 2.852,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 2.619,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.96e-06,
+      "loss": 2.8233,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.98e-06,
+      "loss": 2.8981,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7e-06,
+      "loss": 2.7773,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 2.7274,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.04e-06,
+      "loss": 2.8148,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.06e-06,
+      "loss": 2.6504,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.08e-06,
+      "loss": 2.5703,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 2.6483,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 2.6792,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.14e-06,
+      "loss": 2.679,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.16e-06,
+      "loss": 2.8833,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 2.6336,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 2.6633,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.22e-06,
+      "loss": 2.717,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.24e-06,
+      "loss": 2.554,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 2.6725,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 2.661,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.3e-06,
+      "loss": 2.5621,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.32e-06,
+      "loss": 2.5976,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 2.6586,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 2.7344,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 2.6487,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4e-06,
+      "loss": 2.6789,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 2.6848,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 2.6134,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 2.7112,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.48e-06,
+      "loss": 2.7807,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.7532,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 2.7242,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 2.6479,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 2.7481,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.58e-06,
+      "loss": 2.8313,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 2.607,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 2.6389,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 2.5013,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 2.6571,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 2.494,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.7e-06,
+      "loss": 2.5826,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.72e-06,
+      "loss": 2.9526,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.74e-06,
+      "loss": 2.756,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.76e-06,
+      "loss": 2.5758,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.78e-06,
+      "loss": 2.7195,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 2.6548,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 2.701,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 2.7511,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 2.6924,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.88e-06,
+      "loss": 2.7395,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.9e-06,
+      "loss": 2.6731,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.92e-06,
+      "loss": 2.6414,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.94e-06,
+      "loss": 2.8004,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 2.521,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 2.7233,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.7275,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 2.6185,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 2.57,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.06e-06,
+      "loss": 2.6627,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.08e-06,
+      "loss": 2.7651,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.1e-06,
+      "loss": 2.6516,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 2.8147,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.14e-06,
+      "loss": 2.8192,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.16e-06,
+      "loss": 2.671,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.18e-06,
+      "loss": 2.7703,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.2e-06,
+      "loss": 2.629,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 2.634,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.24e-06,
+      "loss": 2.8503,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.26e-06,
+      "loss": 2.6991,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.28e-06,
+      "loss": 2.714,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.3e-06,
+      "loss": 2.8076,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.32e-06,
+      "loss": 2.6083,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.34e-06,
+      "loss": 2.7288,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.36e-06,
+      "loss": 2.8534,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 2.6112,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 2.6004,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.42e-06,
+      "loss": 2.7698,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.44e-06,
+      "loss": 2.8104,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.46e-06,
+      "loss": 2.8756,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.48e-06,
+      "loss": 2.9634,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.5e-06,
+      "loss": 2.7704,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.52e-06,
+      "loss": 2.7487,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 2.8121,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 2.6654,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 2.6113,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.6e-06,
+      "loss": 2.5954,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.62e-06,
+      "loss": 2.7727,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.64e-06,
+      "loss": 2.6134,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.66e-06,
+      "loss": 2.7212,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.68e-06,
+      "loss": 2.7521,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 2.7893,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 2.6715,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 2.8775,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.76e-06,
+      "loss": 2.7883,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.78e-06,
+      "loss": 2.7315,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.8e-06,
+      "loss": 2.7721,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.82e-06,
+      "loss": 2.6575,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.84e-06,
+      "loss": 2.7204,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 2.6011,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 2.7836,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 2.6523,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 2.694,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.94e-06,
+      "loss": 2.7417,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.96e-06,
+      "loss": 2.7703,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.98e-06,
+      "loss": 2.865,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9e-06,
+      "loss": 2.7502,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 2.7532,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 2.6585,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 2.9593,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 2.6351,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 2.7307,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.12e-06,
+      "loss": 2.797,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.14e-06,
+      "loss": 2.7763,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.16e-06,
+      "loss": 2.7832,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 2.5781,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 2.7653,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 2.6696,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 2.6251,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 2.6316,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 2.7429,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.3e-06,
+      "loss": 2.6906,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.32e-06,
+      "loss": 2.6813,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 2.8493,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 2.715,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.38e-06,
+      "loss": 2.5253,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.4e-06,
+      "loss": 2.7604,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.42e-06,
+      "loss": 2.4687,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 2.7829,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 2.6582,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.48e-06,
+      "loss": 2.6355,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.5e-06,
+      "loss": 2.8583,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.52e-06,
+      "loss": 2.7574,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.54e-06,
+      "loss": 2.7816,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.56e-06,
+      "loss": 2.6692,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.58e-06,
+      "loss": 2.6472,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.6968,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 2.8404,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 2.6729,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.66e-06,
+      "loss": 2.9113,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.68e-06,
+      "loss": 2.6295,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.7e-06,
+      "loss": 2.7364,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.72e-06,
+      "loss": 2.7664,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.74e-06,
+      "loss": 2.7076,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 2.6171,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 2.7698,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 3.0325,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 2.7837,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.84e-06,
+      "loss": 2.5562,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.86e-06,
+      "loss": 2.765,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.88e-06,
+      "loss": 2.6804,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9e-06,
+      "loss": 2.7081,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 2.8065,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 2.7093,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 2.8892,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 2.7807,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 2.7471,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999999596928526e-06,
+      "loss": 2.8152,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999998387714165e-06,
+      "loss": 2.7673,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999996372357115e-06,
+      "loss": 2.7032,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999993550857701e-06,
+      "loss": 2.8087,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999989923216374e-06,
+      "loss": 2.8449,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999985489433723e-06,
+      "loss": 2.6203,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99998024951046e-06,
+      "loss": 2.4976,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999974203447434e-06,
+      "loss": 2.9274,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999967351245614e-06,
+      "loss": 2.5602,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99995969290611e-06,
+      "loss": 2.703,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999951228430154e-06,
+      "loss": 2.7599,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999941957819112e-06,
+      "loss": 2.6951,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999931881074477e-06,
+      "loss": 2.557,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999920998197874e-06,
+      "loss": 2.8728,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99990930919106e-06,
+      "loss": 2.7927,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999896814055916e-06,
+      "loss": 2.7068,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999883512794459e-06,
+      "loss": 2.6488,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999869405408833e-06,
+      "loss": 2.636,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999854491901312e-06,
+      "loss": 2.8244,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999838772274303e-06,
+      "loss": 2.6701,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999822246530337e-06,
+      "loss": 2.8465,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99980491467208e-06,
+      "loss": 2.6571,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999786776702327e-06,
+      "loss": 2.5901,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999767832624e-06,
+      "loss": 2.5644,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999748082440156e-06,
+      "loss": 2.7344,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999727526153979e-06,
+      "loss": 2.618,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99970616376878e-06,
+      "loss": 2.5785,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999683995288008e-06,
+      "loss": 2.6965,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999661020715234e-06,
+      "loss": 2.5738,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999637240054163e-06,
+      "loss": 2.8149,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999612653308631e-06,
+      "loss": 2.7302,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999587260482597e-06,
+      "loss": 2.8494,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999561061580161e-06,
+      "loss": 2.722,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999534056605543e-06,
+      "loss": 2.5865,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999506245563098e-06,
+      "loss": 2.6119,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99947762845731e-06,
+      "loss": 2.7701,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999448205292794e-06,
+      "loss": 2.5768,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999417976074294e-06,
+      "loss": 2.5837,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999386940806682e-06,
+      "loss": 2.5491,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999355099494961e-06,
+      "loss": 2.7923,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999322452144269e-06,
+      "loss": 2.7359,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999288998759865e-06,
+      "loss": 2.5339,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999254739347145e-06,
+      "loss": 2.6716,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999219673911633e-06,
+      "loss": 2.6797,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999183802458981e-06,
+      "loss": 2.596,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999147124994974e-06,
+      "loss": 2.6482,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999109641525524e-06,
+      "loss": 2.732,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999071352056676e-06,
+      "loss": 2.7887,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9990322565946e-06,
+      "loss": 2.7436,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998992355145604e-06,
+      "loss": 2.5541,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99895164771612e-06,
+      "loss": 2.7401,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998910134312709e-06,
+      "loss": 2.6659,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998867814942065e-06,
+      "loss": 2.5018,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998824689611012e-06,
+      "loss": 2.5397,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998780758326502e-06,
+      "loss": 2.6798,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998736021095621e-06,
+      "loss": 2.7053,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998690477925576e-06,
+      "loss": 2.628,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998644128823716e-06,
+      "loss": 2.7674,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99859697379751e-06,
+      "loss": 2.642,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998549012854562e-06,
+      "loss": 2.6768,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998500246002604e-06,
+      "loss": 2.8044,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998450673249499e-06,
+      "loss": 2.7895,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99840029460324e-06,
+      "loss": 2.8164,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99834911007195e-06,
+      "loss": 2.7666,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99829711966388e-06,
+      "loss": 2.8012,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998244323387412e-06,
+      "loss": 2.6359,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99819072125106e-06,
+      "loss": 2.4737,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998136313263465e-06,
+      "loss": 2.6361,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9980810994334e-06,
+      "loss": 2.6834,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998025079769766e-06,
+      "loss": 2.688,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997968254281595e-06,
+      "loss": 2.6501,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99791062297805e-06,
+      "loss": 2.7121,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997852185868424e-06,
+      "loss": 2.5346,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997792942962135e-06,
+      "loss": 2.6804,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997732894268737e-06,
+      "loss": 2.8493,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997672039797912e-06,
+      "loss": 2.6814,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99761037955947e-06,
+      "loss": 2.793,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997547913563352e-06,
+      "loss": 2.7212,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997484641819632e-06,
+      "loss": 2.7028,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99742056433851e-06,
+      "loss": 2.565,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997355681130313e-06,
+      "loss": 2.6943,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99728999220551e-06,
+      "loss": 2.829,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997223497574685e-06,
+      "loss": 2.6748,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997156197248564e-06,
+      "loss": 2.7285,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997088091237991e-06,
+      "loss": 2.5982,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997019179553956e-06,
+      "loss": 2.6312,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99694946220756e-06,
+      "loss": 2.6924,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99687893921005e-06,
+      "loss": 2.775,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996807610572793e-06,
+      "loss": 2.555,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996735476307292e-06,
+      "loss": 2.8272,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996662536425174e-06,
+      "loss": 2.8104,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996588790938201e-06,
+      "loss": 2.836,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996514239858263e-06,
+      "loss": 2.5552,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996438883197378e-06,
+      "loss": 2.5789,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996362720967698e-06,
+      "loss": 2.8688,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996285753181499e-06,
+      "loss": 2.6303,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996207979851195e-06,
+      "loss": 2.5387,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996129400989322e-06,
+      "loss": 2.6391,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99605001660855e-06,
+      "loss": 2.6031,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995969826721679e-06,
+      "loss": 2.7024,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995888831341636e-06,
+      "loss": 2.5714,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995807030481481e-06,
+      "loss": 2.6771,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995724424154403e-06,
+      "loss": 2.6377,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99564101237372e-06,
+      "loss": 2.7481,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99555679515288e-06,
+      "loss": 2.7416,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995471772505461e-06,
+      "loss": 2.7858,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995385944445173e-06,
+      "loss": 2.5898,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99529931098585e-06,
+      "loss": 2.6378,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995211872141466e-06,
+      "loss": 2.7951,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995123627926113e-06,
+      "loss": 2.6203,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995034578354022e-06,
+      "loss": 2.7581,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994944723439546e-06,
+      "loss": 2.7019,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994854063197176e-06,
+      "loss": 2.5173,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99476259764153e-06,
+      "loss": 2.7236,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994670326787351e-06,
+      "loss": 2.7883,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994577250649519e-06,
+      "loss": 2.6469,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994483369243038e-06,
+      "loss": 2.759,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994388682583045e-06,
+      "loss": 2.5253,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994293190684809e-06,
+      "loss": 2.6467,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994196893563722e-06,
+      "loss": 2.6278,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994099791235311e-06,
+      "loss": 2.7281,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994001883715234e-06,
+      "loss": 2.7038,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993903171019272e-06,
+      "loss": 2.7874,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993803653163346e-06,
+      "loss": 2.7049,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993703330163497e-06,
+      "loss": 2.5307,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993602202035902e-06,
+      "loss": 2.7684,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993500268796863e-06,
+      "loss": 2.6364,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993397530462818e-06,
+      "loss": 2.6076,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993293987050328e-06,
+      "loss": 2.7699,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99318963857609e-06,
+      "loss": 2.7259,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993084485056928e-06,
+      "loss": 2.5497,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992978526509794e-06,
+      "loss": 2.6458,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99287176295177e-06,
+      "loss": 2.6104,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992764194400075e-06,
+      "loss": 2.7969,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992655820872045e-06,
+      "loss": 2.5573,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99254664238516e-06,
+      "loss": 2.7115,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992436658957018e-06,
+      "loss": 2.7499,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992325870605353e-06,
+      "loss": 2.7235,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992214277348025e-06,
+      "loss": 2.7157,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99210187920303e-06,
+      "loss": 2.7748,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991988676188486e-06,
+      "loss": 2.6614,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991874668322648e-06,
+      "loss": 2.6307,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991759855623895e-06,
+      "loss": 2.6769,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991644238110741e-06,
+      "loss": 2.6164,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991527815801822e-06,
+      "loss": 2.6349,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991410588715911e-06,
+      "loss": 2.6492,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99129255687191e-06,
+      "loss": 2.7794,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991173720288847e-06,
+      "loss": 2.6926,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991054078985882e-06,
+      "loss": 2.6104,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990933632982307e-06,
+      "loss": 2.6673,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990812382297538e-06,
+      "loss": 2.8701,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990690326951126e-06,
+      "loss": 2.7313,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990567466962749e-06,
+      "loss": 2.7137,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990443802352216e-06,
+      "loss": 2.7408,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990319333139466e-06,
+      "loss": 2.758,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990194059344565e-06,
+      "loss": 2.7648,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990067980987712e-06,
+      "loss": 2.6133,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989941098089235e-06,
+      "loss": 2.634,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98981341066959e-06,
+      "loss": 2.7225,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989684918749365e-06,
+      "loss": 2.624,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989555622349276e-06,
+      "loss": 2.6906,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989425521490169e-06,
+      "loss": 2.581,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989294616193018e-06,
+      "loss": 2.6416,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989162906478933e-06,
+      "loss": 2.4916,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989030392369147e-06,
+      "loss": 2.7408,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988897073885024e-06,
+      "loss": 2.671,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988762951048062e-06,
+      "loss": 2.6259,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988628023879883e-06,
+      "loss": 2.561,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98849229240224e-06,
+      "loss": 2.5333,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98835575663702e-06,
+      "loss": 2.6159,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988218416606234e-06,
+      "loss": 2.7005,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988080272332028e-06,
+      "loss": 2.6932,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987941323836671e-06,
+      "loss": 2.7513,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987801571142569e-06,
+      "loss": 2.7071,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98766101427225e-06,
+      "loss": 2.6824,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98751965324838e-06,
+      "loss": 2.6819,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987377488093746e-06,
+      "loss": 2.6191,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987234518831273e-06,
+      "loss": 2.6233,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987090745484011e-06,
+      "loss": 2.5139,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98694616807514e-06,
+      "loss": 2.6586,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986800786627969e-06,
+      "loss": 2.6545,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98665460116594e-06,
+      "loss": 2.6103,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98650761171262e-06,
+      "loss": 2.6391,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986359818291706e-06,
+      "loss": 2.7295,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986211220927032e-06,
+      "loss": 2.6726,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986061819642554e-06,
+      "loss": 2.7735,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985911614462358e-06,
+      "loss": 2.6624,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985760605410662e-06,
+      "loss": 2.6959,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985608792511813e-06,
+      "loss": 2.742,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98545617579029e-06,
+      "loss": 2.5758,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985302755270695e-06,
+      "loss": 2.544,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985148530977767e-06,
+      "loss": 2.6583,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984993502936369e-06,
+      "loss": 2.5515,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984837671171498e-06,
+      "loss": 2.6473,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984681035708279e-06,
+      "loss": 2.6459,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984523596571964e-06,
+      "loss": 2.5454,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984365353787938e-06,
+      "loss": 2.5949,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984206307381714e-06,
+      "loss": 2.6478,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984046457378935e-06,
+      "loss": 2.6056,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983885803805373e-06,
+      "loss": 2.6816,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983724346686929e-06,
+      "loss": 2.5684,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983562086049636e-06,
+      "loss": 2.8159,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983399021919655e-06,
+      "loss": 2.6108,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983235154323276e-06,
+      "loss": 2.6073,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983070483286919e-06,
+      "loss": 2.7053,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982905008837133e-06,
+      "loss": 2.7898,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982738731000602e-06,
+      "loss": 2.6836,
+      "step": 707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982571649804126e-06,
+      "loss": 2.7115,
+      "step": 708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982403765274652e-06,
+      "loss": 2.7248,
+      "step": 709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982235077439242e-06,
+      "loss": 2.6297,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982065586325096e-06,
+      "loss": 2.7146,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981895291959541e-06,
+      "loss": 2.8611,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98172419437003e-06,
+      "loss": 2.656,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981552293584154e-06,
+      "loss": 2.6716,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981379589629624e-06,
+      "loss": 2.5776,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981206082534287e-06,
+      "loss": 2.5523,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981031772326118e-06,
+      "loss": 2.7048,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980856659033216e-06,
+      "loss": 2.7631,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980680742683822e-06,
+      "loss": 2.6329,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98050402330629e-06,
+      "loss": 2.6379,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980326500929121e-06,
+      "loss": 2.7127,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98014817558093e-06,
+      "loss": 2.6222,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979969047290472e-06,
+      "loss": 2.6534,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979789116086625e-06,
+      "loss": 2.6033,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9796083819984e-06,
+      "loss": 2.9023,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979426845054938e-06,
+      "loss": 2.6704,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979244505285505e-06,
+      "loss": 2.5729,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979061362719502e-06,
+      "loss": 2.7437,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978877417386456e-06,
+      "loss": 2.6636,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978692669316024e-06,
+      "loss": 2.8921,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978507118537994e-06,
+      "loss": 2.8422,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97832076508228e-06,
+      "loss": 2.8175,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978133608978927e-06,
+      "loss": 2.6225,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977945650258111e-06,
+      "loss": 2.6151,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977756888950139e-06,
+      "loss": 2.4018,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977567325085442e-06,
+      "loss": 2.6013,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977376958694583e-06,
+      "loss": 2.5516,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977185789808254e-06,
+      "loss": 2.6465,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97699381845728e-06,
+      "loss": 2.7325,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976801044672608e-06,
+      "loss": 2.7577,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97660746848532e-06,
+      "loss": 2.632,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976413089926629e-06,
+      "loss": 2.8808,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976217909027872e-06,
+      "loss": 2.8911,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976021925820516e-06,
+      "loss": 2.6939,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975825140336163e-06,
+      "loss": 2.6575,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975627552606538e-06,
+      "loss": 2.7469,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975429162663498e-06,
+      "loss": 2.623,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97522997053903e-06,
+      "loss": 2.6184,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975029976265248e-06,
+      "loss": 2.632,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974829179874398e-06,
+      "loss": 2.674,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974627581398853e-06,
+      "loss": 2.6871,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974425180871118e-06,
+      "loss": 2.678,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974221978323824e-06,
+      "loss": 2.7658,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974017973789735e-06,
+      "loss": 2.6934,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973813167301743e-06,
+      "loss": 2.6184,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973607558892864e-06,
+      "loss": 2.8569,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973401148596251e-06,
+      "loss": 2.6307,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973193936445185e-06,
+      "loss": 2.4401,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972985922473073e-06,
+      "loss": 2.744,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972777106713452e-06,
+      "loss": 2.8363,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972567489199987e-06,
+      "loss": 2.5208,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97235706996648e-06,
+      "loss": 2.572,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972145849046853e-06,
+      "loss": 2.7874,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971933826475162e-06,
+      "loss": 2.5977,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971721002285589e-06,
+      "loss": 2.4596,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971507376512449e-06,
+      "loss": 2.6819,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971292949190185e-06,
+      "loss": 2.7796,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971077720353368e-06,
+      "loss": 2.7317,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9708616900367e-06,
+      "loss": 2.6484,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970644858275008e-06,
+      "loss": 2.5201,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970427225103257e-06,
+      "loss": 2.6916,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970208790556531e-06,
+      "loss": 2.7888,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96998955467005e-06,
+      "loss": 2.7053,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969769517479162e-06,
+      "loss": 2.8093,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96954867901934e-06,
+      "loss": 2.7505,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969327039326193e-06,
+      "loss": 2.5734,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969104598435454e-06,
+      "loss": 2.5812,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968881356382985e-06,
+      "loss": 2.5958,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968657313204783e-06,
+      "loss": 2.6269,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968432468936967e-06,
+      "loss": 2.492,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96820682361579e-06,
+      "loss": 2.6069,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967980377277632e-06,
+      "loss": 2.704,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967753129959e-06,
+      "loss": 2.6147,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96752508169654e-06,
+      "loss": 2.8338,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967296232527013e-06,
+      "loss": 2.9013,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967066582487317e-06,
+      "loss": 2.5386,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96683613161448e-06,
+      "loss": 2.8145,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966604879945659e-06,
+      "loss": 2.4897,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966372827518133e-06,
+      "loss": 2.5608,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96613997436932e-06,
+      "loss": 2.6553,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96590632053676e-06,
+      "loss": 2.5605,
+      "step": 791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.965671866058126e-06,
+      "loss": 2.6024,
+      "step": 792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.965436610971219e-06,
+      "loss": 2.6897,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96520055531397e-06,
+      "loss": 2.634,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964963699124434e-06,
+      "loss": 2.6553,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964726042440802e-06,
+      "loss": 2.6558,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96448758530139e-06,
+      "loss": 2.681,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964248327744645e-06,
+      "loss": 2.6811,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964008269809142e-06,
+      "loss": 2.7352,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963767411533584e-06,
+      "loss": 2.5464,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963525752956805e-06,
+      "loss": 2.6996,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963283294117768e-06,
+      "loss": 2.6565,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963040035055562e-06,
+      "loss": 2.6059,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962795975809411e-06,
+      "loss": 2.773,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962551116418662e-06,
+      "loss": 2.792,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962305456922792e-06,
+      "loss": 2.6749,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962058997361413e-06,
+      "loss": 2.7405,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961811737774256e-06,
+      "loss": 2.6446,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961563678201187e-06,
+      "loss": 2.5141,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961314818682205e-06,
+      "loss": 2.7463,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961065159257428e-06,
+      "loss": 2.6984,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960814699967112e-06,
+      "loss": 2.6402,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960563440851636e-06,
+      "loss": 2.7279,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96031138195151e-06,
+      "loss": 2.7699,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960058523307375e-06,
+      "loss": 2.6444,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959804864959996e-06,
+      "loss": 2.5227,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959550406950273e-06,
+      "loss": 2.7408,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959295149319233e-06,
+      "loss": 2.7429,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959039092108025e-06,
+      "loss": 2.7074,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958782235357938e-06,
+      "loss": 2.6348,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958524579110384e-06,
+      "loss": 2.7218,
+      "step": 821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958266123406901e-06,
+      "loss": 2.7546,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958006868289164e-06,
+      "loss": 2.7894,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95774681379897e-06,
+      "loss": 2.6929,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.957485959978246e-06,
+      "loss": 2.8333,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.957224306869053e-06,
+      "loss": 2.7111,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956961854513574e-06,
+      "loss": 2.7532,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956698602954124e-06,
+      "loss": 2.6462,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956434552233147e-06,
+      "loss": 2.4514,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956169702393215e-06,
+      "loss": 2.6819,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95590405347703e-06,
+      "loss": 2.8479,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955637605527425e-06,
+      "loss": 2.5551,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955370358587354e-06,
+      "loss": 2.7663,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955102312699905e-06,
+      "loss": 2.7017,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9548334679083e-06,
+      "loss": 2.6073,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954563824255879e-06,
+      "loss": 2.85,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95429338178612e-06,
+      "loss": 2.6695,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954022140542622e-06,
+      "loss": 2.4893,
+      "step": 838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953750100569122e-06,
+      "loss": 2.7494,
+      "step": 839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953477261909478e-06,
+      "loss": 2.5879,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953203624607679e-06,
+      "loss": 2.8848,
+      "step": 841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952929188707843e-06,
+      "loss": 2.5864,
+      "step": 842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952653954254218e-06,
+      "loss": 2.6576,
+      "step": 843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952377921291179e-06,
+      "loss": 2.6337,
+      "step": 844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952101089863232e-06,
+      "loss": 2.7645,
+      "step": 845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951823460015004e-06,
+      "loss": 2.6757,
+      "step": 846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951545031791265e-06,
+      "loss": 2.7233,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951265805236903e-06,
+      "loss": 2.7793,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950985780396934e-06,
+      "loss": 2.754,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95070495731651e-06,
+      "loss": 2.7755,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950423336040904e-06,
+      "loss": 2.678,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950140916615526e-06,
+      "loss": 2.5232,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949857699085907e-06,
+      "loss": 2.5793,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94957368349771e-06,
+      "loss": 2.595,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949288869896727e-06,
+      "loss": 2.8473,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949003258328878e-06,
+      "loss": 2.5512,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948716848840211e-06,
+      "loss": 2.6794,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948429641476905e-06,
+      "loss": 2.6545,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948141636285264e-06,
+      "loss": 2.4357,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947852833311725e-06,
+      "loss": 2.6344,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94756323260285e-06,
+      "loss": 2.5807,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94727283420533e-06,
+      "loss": 2.575,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946981638165987e-06,
+      "loss": 2.7577,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94668964453177e-06,
+      "loss": 2.7748,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946396853349756e-06,
+      "loss": 2.7502,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946103264667151e-06,
+      "loss": 2.6322,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945808878531291e-06,
+      "loss": 2.5729,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94551369498964e-06,
+      "loss": 2.6847,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945217714089787e-06,
+      "loss": 2.7741,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944920935879456e-06,
+      "loss": 2.6681,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944623360406496e-06,
+      "loss": 2.773,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94432498771888e-06,
+      "loss": 2.8415,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944025817864722e-06,
+      "loss": 2.7011,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943725850892249e-06,
+      "loss": 2.7413,
+      "step": 874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.943425086849829e-06,
+      "loss": 2.6027,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.943123525785952e-06,
+      "loss": 2.5397,
+      "step": 876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942821167749239e-06,
+      "loss": 2.8096,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942518012788438e-06,
+      "loss": 2.5149,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942214060952426e-06,
+      "loss": 2.5931,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.94190931229021e-06,
+      "loss": 2.747,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941603766850924e-06,
+      "loss": 2.4034,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941297424683832e-06,
+      "loss": 2.5628,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.94099028583832e-06,
+      "loss": 2.6205,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940682350363913e-06,
+      "loss": 2.6363,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940373618310255e-06,
+      "loss": 2.5571,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940064089727127e-06,
+      "loss": 2.6705,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.939753764664428e-06,
+      "loss": 2.4719,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.939442643172197e-06,
+      "loss": 2.6,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.939130725300592e-06,
+      "loss": 2.5352,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938818011099906e-06,
+      "loss": 2.5904,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938504500620554e-06,
+      "loss": 2.4875,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938190193913084e-06,
+      "loss": 2.7945,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937875091028173e-06,
+      "loss": 2.6087,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937559192016624e-06,
+      "loss": 2.7428,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937242496929367e-06,
+      "loss": 2.5869,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936925005817464e-06,
+      "loss": 2.5228,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936606718732103e-06,
+      "loss": 2.569,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936287635724601e-06,
+      "loss": 2.5986,
+      "step": 898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935967756846406e-06,
+      "loss": 2.7891,
+      "step": 899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935647082149088e-06,
+      "loss": 2.7049,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935325611684348e-06,
+      "loss": 2.668,
+      "step": 901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93500334550402e-06,
+      "loss": 2.5601,
+      "step": 902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93468028366006e-06,
+      "loss": 2.5658,
+      "step": 903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934356426204557e-06,
+      "loss": 2.7243,
+      "step": 904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934031773189724e-06,
+      "loss": 2.7012,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933706324667906e-06,
+      "loss": 2.6517,
+      "step": 906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933380080691573e-06,
+      "loss": 2.7875,
+      "step": 907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933053041313325e-06,
+      "loss": 2.7827,
+      "step": 908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932725206585892e-06,
+      "loss": 2.6219,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932396576562128e-06,
+      "loss": 2.8049,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932067151295018e-06,
+      "loss": 2.7362,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931736930837676e-06,
+      "loss": 2.5993,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931405915243343e-06,
+      "loss": 2.7668,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931074104565387e-06,
+      "loss": 2.6353,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930741498857307e-06,
+      "loss": 2.4801,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930408098172725e-06,
+      "loss": 2.5892,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930073902565397e-06,
+      "loss": 2.7477,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929738912089207e-06,
+      "loss": 2.4668,
+      "step": 918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929403126798162e-06,
+      "loss": 2.6532,
+      "step": 919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929066546746403e-06,
+      "loss": 2.6674,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928729171988191e-06,
+      "loss": 2.5192,
+      "step": 921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928391002577926e-06,
+      "loss": 2.6031,
+      "step": 922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928052038570129e-06,
+      "loss": 2.7143,
+      "step": 923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92771228001945e-06,
+      "loss": 2.806,
+      "step": 924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927371726980669e-06,
+      "loss": 2.7175,
+      "step": 925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927030379508691e-06,
+      "loss": 2.5713,
+      "step": 926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926688237658551e-06,
+      "loss": 2.7587,
+      "step": 927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926345301485414e-06,
+      "loss": 2.6487,
+      "step": 928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92600157104457e-06,
+      "loss": 2.563,
+      "step": 929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925657046391438e-06,
+      "loss": 2.701,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925311727581567e-06,
+      "loss": 2.5976,
+      "step": 931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924965614670629e-06,
+      "loss": 2.661,
+      "step": 932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924618707714429e-06,
+      "loss": 2.605,
+      "step": 933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.9242710067689e-06,
+      "loss": 2.5763,
+      "step": 934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923922511890098e-06,
+      "loss": 2.7099,
+      "step": 935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923573223134213e-06,
+      "loss": 2.5051,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923223140557558e-06,
+      "loss": 2.5177,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922872264216581e-06,
+      "loss": 2.6984,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922520594167847e-06,
+      "loss": 2.8218,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92216813046806e-06,
+      "loss": 2.5356,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921814873174045e-06,
+      "loss": 2.8193,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921460822342757e-06,
+      "loss": 2.7087,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92110597803128e-06,
+      "loss": 2.5888,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920750340296823e-06,
+      "loss": 2.5517,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920393909196729e-06,
+      "loss": 2.6957,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920036684788464e-06,
+      "loss": 2.6603,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919678667129618e-06,
+      "loss": 2.6993,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919319856277921e-06,
+      "loss": 2.6966,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918960252291216e-06,
+      "loss": 2.5685,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918599855227487e-06,
+      "loss": 2.7165,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91823866514484e-06,
+      "loss": 2.523,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917876682101504e-06,
+      "loss": 2.5824,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917513906155847e-06,
+      "loss": 2.5774,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917150337366357e-06,
+      "loss": 2.587,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916785975791649e-06,
+      "loss": 2.7042,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916420821490474e-06,
+      "loss": 2.6069,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.9160548745217e-06,
+      "loss": 2.6973,
+      "step": 957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.915688134944328e-06,
+      "loss": 2.9187,
+      "step": 958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.915320602817493e-06,
+      "loss": 2.6963,
+      "step": 959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914952278200444e-06,
+      "loss": 2.6668,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914583161152574e-06,
+      "loss": 2.5568,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914213251733387e-06,
+      "loss": 2.7312,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913842550002526e-06,
+      "loss": 2.7678,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91347105601976e-06,
+      "loss": 2.806,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913098769844984e-06,
+      "loss": 2.7121,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91272569153822e-06,
+      "loss": 2.5571,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91235182115962e-06,
+      "loss": 2.652,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911977158769461e-06,
+      "loss": 2.6688,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911601704428153e-06,
+      "loss": 2.7113,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911225458196226e-06,
+      "loss": 2.741,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910848420134342e-06,
+      "loss": 2.4043,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910470590303294e-06,
+      "loss": 2.5095,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910091968763995e-06,
+      "loss": 2.5416,
+      "step": 973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909712555577491e-06,
+      "loss": 2.7067,
+      "step": 974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909332350804954e-06,
+      "loss": 2.7044,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908951354507683e-06,
+      "loss": 2.6406,
+      "step": 976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908569566747108e-06,
+      "loss": 2.7271,
+      "step": 977
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908186987584783e-06,
+      "loss": 2.8177,
+      "step": 978
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907803617082391e-06,
+      "loss": 2.6757,
+      "step": 979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90741945530174e-06,
+      "loss": 2.6966,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907034502304771e-06,
+      "loss": 2.6855,
+      "step": 981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906648758153547e-06,
+      "loss": 2.6183,
+      "step": 982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906262222910264e-06,
+      "loss": 2.6375,
+      "step": 983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905874896637238e-06,
+      "loss": 2.6201,
+      "step": 984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905486779396921e-06,
+      "loss": 2.8266,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905097871251887e-06,
+      "loss": 2.723,
+      "step": 986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90470817226484e-06,
+      "loss": 2.5363,
+      "step": 987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904317682498609e-06,
+      "loss": 2.6324,
+      "step": 988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903926402016153e-06,
+      "loss": 2.5363,
+      "step": 989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903534330880557e-06,
+      "loss": 2.7391,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903141469155036e-06,
+      "loss": 2.5352,
+      "step": 991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90274781690293e-06,
+      "loss": 2.6342,
+      "step": 992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.902353374187706e-06,
+      "loss": 2.7823,
+      "step": 993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901958141072957e-06,
+      "loss": 2.5623,
+      "step": 994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901562117622412e-06,
+      "loss": 2.7264,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901165303899916e-06,
+      "loss": 2.6299,
+      "step": 996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.900767699969449e-06,
+      "loss": 2.6094,
+      "step": 997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.900369305895115e-06,
+      "loss": 2.6387,
+      "step": 998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899970121741147e-06,
+      "loss": 2.6321,
+      "step": 999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899570147571907e-06,
+      "loss": 2.3323,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899169383451878e-06,
+      "loss": 2.5735,
+      "step": 1001
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89876782944568e-06,
+      "loss": 2.6439,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89836548561805e-06,
+      "loss": 2.7669,
+      "step": 1003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89796235203386e-06,
+      "loss": 2.8535,
+      "step": 1004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897558428758107e-06,
+      "loss": 2.7649,
+      "step": 1005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897153715855911e-06,
+      "loss": 2.6537,
+      "step": 1006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.896748213392529e-06,
+      "loss": 2.5848,
+      "step": 1007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.896341921433337e-06,
+      "loss": 2.6654,
+      "step": 1008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895934840043839e-06,
+      "loss": 2.7781,
+      "step": 1009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895526969289671e-06,
+      "loss": 2.7418,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895118309236591e-06,
+      "loss": 2.5787,
+      "step": 1011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89470885995049e-06,
+      "loss": 2.5821,
+      "step": 1012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89429862149738e-06,
+      "loss": 2.6385,
+      "step": 1013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.893887593943405e-06,
+      "loss": 2.7581,
+      "step": 1014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.893475777354832e-06,
+      "loss": 2.4466,
+      "step": 1015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89306317179806e-06,
+      "loss": 2.7203,
+      "step": 1016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.892649777339611e-06,
+      "loss": 2.6585,
+      "step": 1017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89223559404614e-06,
+      "loss": 2.6484,
+      "step": 1018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891820621984418e-06,
+      "loss": 2.5929,
+      "step": 1019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891404861221356e-06,
+      "loss": 2.5658,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890988311823986e-06,
+      "loss": 2.6685,
+      "step": 1021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890570973859468e-06,
+      "loss": 2.4192,
+      "step": 1022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890152847395085e-06,
+      "loss": 2.6418,
+      "step": 1023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.889733932498253e-06,
+      "loss": 2.8217,
+      "step": 1024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.889314229236515e-06,
+      "loss": 2.7264,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888893737677537e-06,
+      "loss": 2.8306,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888472457889115e-06,
+      "loss": 2.585,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888050389939172e-06,
+      "loss": 2.6385,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.887627533895757e-06,
+      "loss": 2.6988,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.887203889827045e-06,
+      "loss": 2.7786,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88677945780134e-06,
+      "loss": 2.5515,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.886354237887075e-06,
+      "loss": 2.6657,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885928230152804e-06,
+      "loss": 2.7088,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885501434667215e-06,
+      "loss": 2.599,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885073851499119e-06,
+      "loss": 2.692,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884645480717452e-06,
+      "loss": 2.8362,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884216322391282e-06,
+      "loss": 2.6754,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883786376589802e-06,
+      "loss": 2.7087,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883355643382328e-06,
+      "loss": 2.6454,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882924122838311e-06,
+      "loss": 2.7205,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.882491815027322e-06,
+      "loss": 2.5998,
+      "step": 1041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.882058720019062e-06,
+      "loss": 2.6998,
+      "step": 1042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.881624837883358e-06,
+      "loss": 2.5463,
+      "step": 1043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.881190168690164e-06,
+      "loss": 2.7631,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.880754712509563e-06,
+      "loss": 2.8184,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.88031846941176e-06,
+      "loss": 2.4644,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.879881439467091e-06,
+      "loss": 2.7637,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87944362274602e-06,
+      "loss": 2.6318,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.879005019319133e-06,
+      "loss": 2.7644,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.878565629257146e-06,
+      "loss": 2.7064,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.8781254526309e-06,
+      "loss": 2.4167,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.877684489511367e-06,
+      "loss": 2.6741,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87724273996964e-06,
+      "loss": 2.6613,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.876800204076944e-06,
+      "loss": 2.5728,
+      "step": 1054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.876356881904627e-06,
+      "loss": 2.6559,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.875912773524165e-06,
+      "loss": 2.6052,
+      "step": 1056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87546787900716e-06,
+      "loss": 2.6146,
+      "step": 1057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.875022198425347e-06,
+      "loss": 2.7509,
+      "step": 1058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.874575731850575e-06,
+      "loss": 2.5809,
+      "step": 1059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.874128479354833e-06,
+      "loss": 2.614,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.873680441010227e-06,
+      "loss": 2.6881,
+      "step": 1061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.873231616888997e-06,
+      "loss": 2.7237,
+      "step": 1062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.872782007063503e-06,
+      "loss": 2.7201,
+      "step": 1063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.872331611606238e-06,
+      "loss": 2.521,
+      "step": 1064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.871880430589817e-06,
+      "loss": 2.6008,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.871428464086983e-06,
+      "loss": 2.4882,
+      "step": 1066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.870975712170605e-06,
+      "loss": 2.4552,
+      "step": 1067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.870522174913683e-06,
+      "loss": 2.683,
+      "step": 1068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.870067852389338e-06,
+      "loss": 2.7308,
+      "step": 1069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.869612744670817e-06,
+      "loss": 2.6873,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.869156851831502e-06,
+      "loss": 2.8014,
+      "step": 1071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.868700173944895e-06,
+      "loss": 2.4885,
+      "step": 1072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86824271108462e-06,
+      "loss": 2.6166,
+      "step": 1073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.867784463324437e-06,
+      "loss": 2.552,
+      "step": 1074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86732543073823e-06,
+      "loss": 2.6872,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.866865613400008e-06,
+      "loss": 2.6828,
+      "step": 1076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.866405011383902e-06,
+      "loss": 2.536,
+      "step": 1077
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86594362476418e-06,
+      "loss": 2.6703,
+      "step": 1078
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.865481453615227e-06,
+      "loss": 2.3919,
+      "step": 1079
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86501849801156e-06,
+      "loss": 2.7571,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86455475802782e-06,
+      "loss": 2.5048,
+      "step": 1081
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.864090233738776e-06,
+      "loss": 2.5079,
+      "step": 1082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86362492521932e-06,
+      "loss": 2.6762,
+      "step": 1083
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.863158832544477e-06,
+      "loss": 2.5886,
+      "step": 1084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.862691955789393e-06,
+      "loss": 2.6237,
+      "step": 1085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.862224295029339e-06,
+      "loss": 2.5936,
+      "step": 1086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.861755850339718e-06,
+      "loss": 2.7221,
+      "step": 1087
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.861286621796056e-06,
+      "loss": 2.6428,
+      "step": 1088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.860816609474008e-06,
+      "loss": 2.509,
+      "step": 1089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86034581344935e-06,
+      "loss": 2.6801,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.859874233797991e-06,
+      "loss": 2.6718,
+      "step": 1091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.859401870595959e-06,
+      "loss": 2.6979,
+      "step": 1092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.858928723919419e-06,
+      "loss": 2.6733,
+      "step": 1093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.858454793844648e-06,
+      "loss": 2.5676,
+      "step": 1094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.857980080448064e-06,
+      "loss": 2.7128,
+      "step": 1095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.8575045838062e-06,
+      "loss": 2.7672,
+      "step": 1096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.857028303995719e-06,
+      "loss": 2.8331,
+      "step": 1097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.856551241093414e-06,
+      "loss": 2.8234,
+      "step": 1098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.856073395176203e-06,
+      "loss": 2.7091,
+      "step": 1099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.855594766321122e-06,
+      "loss": 2.578,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.855115354605345e-06,
+      "loss": 2.7099,
+      "step": 1101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854635160106166e-06,
+      "loss": 2.6883,
+      "step": 1102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854154182901e-06,
+      "loss": 2.584,
+      "step": 1103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.853672423067404e-06,
+      "loss": 2.4562,
+      "step": 1104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.853189880683046e-06,
+      "loss": 2.6069,
+      "step": 1105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.852706555825727e-06,
+      "loss": 2.7305,
+      "step": 1106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85222244857337e-06,
+      "loss": 2.7552,
+      "step": 1107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85173755900403e-06,
+      "loss": 2.7975,
+      "step": 1108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.851251887195886e-06,
+      "loss": 2.5809,
+      "step": 1109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.850765433227239e-06,
+      "loss": 2.6274,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.850278197176522e-06,
+      "loss": 2.6602,
+      "step": 1111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.849790179122287e-06,
+      "loss": 2.6088,
+      "step": 1112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.849301379143223e-06,
+      "loss": 2.5476,
+      "step": 1113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.848811797318132e-06,
+      "loss": 2.5605,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.848321433725954e-06,
+      "loss": 2.6549,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.847830288445745e-06,
+      "loss": 2.8376,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.847338361556697e-06,
+      "loss": 2.5053,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846845653138117e-06,
+      "loss": 2.4495,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846352163269448e-06,
+      "loss": 2.66,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845857892030252e-06,
+      "loss": 2.7009,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84536283950022e-06,
+      "loss": 2.6353,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.844867005759171e-06,
+      "loss": 2.6904,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.844370390887046e-06,
+      "loss": 2.568,
+      "step": 1123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.843872994963912e-06,
+      "loss": 2.6672,
+      "step": 1124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.843374818069964e-06,
+      "loss": 2.5252,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.842875860285525e-06,
+      "loss": 2.5919,
+      "step": 1126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.842376121691038e-06,
+      "loss": 2.5046,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.841875602367079e-06,
+      "loss": 2.5479,
+      "step": 1128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.841374302394341e-06,
+      "loss": 2.7249,
+      "step": 1129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.84087222185365e-06,
+      "loss": 2.7983,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.84036936082596e-06,
+      "loss": 2.6776,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83986571939234e-06,
+      "loss": 2.7237,
+      "step": 1132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.839361297633994e-06,
+      "loss": 2.7218,
+      "step": 1133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83885609563225e-06,
+      "loss": 2.3394,
+      "step": 1134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83835011346856e-06,
+      "loss": 2.5308,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.837843351224504e-06,
+      "loss": 2.6986,
+      "step": 1136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.837335808981785e-06,
+      "loss": 2.5363,
+      "step": 1137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.836827486822233e-06,
+      "loss": 2.6301,
+      "step": 1138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.836318384827807e-06,
+      "loss": 2.6516,
+      "step": 1139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.835808503080586e-06,
+      "loss": 2.7416,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83529784166278e-06,
+      "loss": 2.6613,
+      "step": 1141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.834786400656719e-06,
+      "loss": 2.6756,
+      "step": 1142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.834274180144865e-06,
+      "loss": 2.6653,
+      "step": 1143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.833761180209799e-06,
+      "loss": 2.6257,
+      "step": 1144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.833247400934236e-06,
+      "loss": 2.5206,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.832732842401008e-06,
+      "loss": 2.7033,
+      "step": 1146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.832217504693078e-06,
+      "loss": 2.5908,
+      "step": 1147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.831701387893533e-06,
+      "loss": 2.4812,
+      "step": 1148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.831184492085584e-06,
+      "loss": 2.522,
+      "step": 1149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.830666817352575e-06,
+      "loss": 2.6933,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.830148363777963e-06,
+      "loss": 2.5252,
+      "step": 1151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.829629131445342e-06,
+      "loss": 2.7246,
+      "step": 1152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.829109120438426e-06,
+      "loss": 2.6901,
+      "step": 1153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.828588330841055e-06,
+      "loss": 2.6394,
+      "step": 1154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.828066762737196e-06,
+      "loss": 2.6392,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82754441621094e-06,
+      "loss": 2.5612,
+      "step": 1156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.827021291346506e-06,
+      "loss": 2.6585,
+      "step": 1157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.826497388228234e-06,
+      "loss": 2.6451,
+      "step": 1158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.825972706940591e-06,
+      "loss": 2.6866,
+      "step": 1159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.825447247568177e-06,
+      "loss": 2.7848,
+      "step": 1160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.824921010195706e-06,
+      "loss": 2.6162,
+      "step": 1161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.824393994908021e-06,
+      "loss": 2.7045,
+      "step": 1162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.823866201790097e-06,
+      "loss": 2.832,
+      "step": 1163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.823337630927027e-06,
+      "loss": 2.7403,
+      "step": 1164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.822808282404031e-06,
+      "loss": 2.5951,
+      "step": 1165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.822278156306455e-06,
+      "loss": 2.6795,
+      "step": 1166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.821747252719772e-06,
+      "loss": 2.5628,
+      "step": 1167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.821215571729578e-06,
+      "loss": 2.7008,
+      "step": 1168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.820683113421595e-06,
+      "loss": 2.5576,
+      "step": 1169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82014987788167e-06,
+      "loss": 2.5441,
+      "step": 1170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.819615865195778e-06,
+      "loss": 2.5093,
+      "step": 1171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.819081075450014e-06,
+      "loss": 2.6538,
+      "step": 1172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.818545508730604e-06,
+      "loss": 2.4913,
+      "step": 1173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.818009165123895e-06,
+      "loss": 2.6728,
+      "step": 1174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.817472044716362e-06,
+      "loss": 2.5991,
+      "step": 1175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.816934147594606e-06,
+      "loss": 2.6714,
+      "step": 1176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.816395473845346e-06,
+      "loss": 2.5955,
+      "step": 1177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.815856023555437e-06,
+      "loss": 2.8624,
+      "step": 1178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.815315796811849e-06,
+      "loss": 2.3883,
+      "step": 1179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.814774793701686e-06,
+      "loss": 2.5678,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.814233014312174e-06,
+      "loss": 2.6132,
+      "step": 1181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.813690458730659e-06,
+      "loss": 2.6245,
+      "step": 1182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.813147127044618e-06,
+      "loss": 2.5358,
+      "step": 1183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.812603019341652e-06,
+      "loss": 2.7457,
+      "step": 1184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.812058135709488e-06,
+      "loss": 2.6248,
+      "step": 1185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.811512476235974e-06,
+      "loss": 2.3929,
+      "step": 1186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.810966041009088e-06,
+      "loss": 2.6402,
+      "step": 1187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.810418830116933e-06,
+      "loss": 2.5529,
+      "step": 1188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.809870843647731e-06,
+      "loss": 2.645,
+      "step": 1189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.809322081689834e-06,
+      "loss": 2.8,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.808772544331719e-06,
+      "loss": 2.72,
+      "step": 1191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.808222231661989e-06,
+      "loss": 2.632,
+      "step": 1192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.807671143769367e-06,
+      "loss": 2.6817,
+      "step": 1193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.807119280742702e-06,
+      "loss": 2.7025,
+      "step": 1194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.806566642670979e-06,
+      "loss": 2.6578,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80601322964329e-06,
+      "loss": 2.7352,
+      "step": 1196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.805459041748864e-06,
+      "loss": 2.4547,
+      "step": 1197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.804904079077054e-06,
+      "loss": 2.5056,
+      "step": 1198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.804348341717333e-06,
+      "loss": 2.7427,
+      "step": 1199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.803791829759302e-06,
+      "loss": 2.5503,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80323454329269e-06,
+      "loss": 2.7746,
+      "step": 1201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.802676482407343e-06,
+      "loss": 2.6463,
+      "step": 1202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80211764719324e-06,
+      "loss": 2.5803,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80155803774048e-06,
+      "loss": 2.6343,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.800997654139286e-06,
+      "loss": 2.5275,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.800436496480009e-06,
+      "loss": 2.572,
+      "step": 1206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.799874564853126e-06,
+      "loss": 2.7689,
+      "step": 1207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.799311859349235e-06,
+      "loss": 2.5281,
+      "step": 1208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.798748380059058e-06,
+      "loss": 2.5983,
+      "step": 1209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.798184127073446e-06,
+      "loss": 2.6778,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.797619100483374e-06,
+      "loss": 2.5932,
+      "step": 1211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.797053300379938e-06,
+      "loss": 2.5835,
+      "step": 1212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.796486726854361e-06,
+      "loss": 2.6636,
+      "step": 1213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.795919379997993e-06,
+      "loss": 2.5734,
+      "step": 1214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.795351259902308e-06,
+      "loss": 2.749,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.794782366658896e-06,
+      "loss": 2.5195,
+      "step": 1216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.794212700359488e-06,
+      "loss": 2.7522,
+      "step": 1217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.793642261095924e-06,
+      "loss": 2.5126,
+      "step": 1218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.793071048960177e-06,
+      "loss": 2.6713,
+      "step": 1219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.792499064044343e-06,
+      "loss": 2.4674,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.791926306440643e-06,
+      "loss": 2.7383,
+      "step": 1221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.791352776241421e-06,
+      "loss": 2.6019,
+      "step": 1222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.790778473539147e-06,
+      "loss": 2.5178,
+      "step": 1223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.790203398426414e-06,
+      "loss": 2.689,
+      "step": 1224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.789627550995941e-06,
+      "loss": 2.6425,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.789050931340572e-06,
+      "loss": 2.6346,
+      "step": 1226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.788473539553271e-06,
+      "loss": 2.7565,
+      "step": 1227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.787895375727137e-06,
+      "loss": 2.6264,
+      "step": 1228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.787316439955378e-06,
+      "loss": 2.5931,
+      "step": 1229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.786736732331343e-06,
+      "loss": 2.586,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.786156252948493e-06,
+      "loss": 2.7511,
+      "step": 1231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.785575001900418e-06,
+      "loss": 2.6356,
+      "step": 1232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.784992979280833e-06,
+      "loss": 2.6094,
+      "step": 1233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.784410185183579e-06,
+      "loss": 2.7318,
+      "step": 1234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.783826619702615e-06,
+      "loss": 2.6301,
+      "step": 1235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.783242282932028e-06,
+      "loss": 2.6861,
+      "step": 1236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.782657174966035e-06,
+      "loss": 2.5651,
+      "step": 1237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78207129589897e-06,
+      "loss": 2.6839,
+      "step": 1238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.781484645825289e-06,
+      "loss": 2.5388,
+      "step": 1239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.780897224839583e-06,
+      "loss": 2.7062,
+      "step": 1240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.780309033036556e-06,
+      "loss": 2.5085,
+      "step": 1241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.779720070511048e-06,
+      "loss": 2.7103,
+      "step": 1242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77913033735801e-06,
+      "loss": 2.6695,
+      "step": 1243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.778539833672525e-06,
+      "loss": 2.5719,
+      "step": 1244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.777948559549802e-06,
+      "loss": 2.6052,
+      "step": 1245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.777356515085167e-06,
+      "loss": 2.7043,
+      "step": 1246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77676370037408e-06,
+      "loss": 2.6633,
+      "step": 1247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.776170115512115e-06,
+      "loss": 2.5843,
+      "step": 1248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.775575760594978e-06,
+      "loss": 2.5889,
+      "step": 1249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.774980635718494e-06,
+      "loss": 2.5847,
+      "step": 1250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.774384740978614e-06,
+      "loss": 2.5711,
+      "step": 1251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.773788076471415e-06,
+      "loss": 2.7146,
+      "step": 1252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.773190642293094e-06,
+      "loss": 2.5059,
+      "step": 1253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.772592438539977e-06,
+      "loss": 2.8255,
+      "step": 1254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77199346530851e-06,
+      "loss": 2.6024,
+      "step": 1255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.771393722695267e-06,
+      "loss": 2.5862,
+      "step": 1256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.770793210796938e-06,
+      "loss": 2.6022,
+      "step": 1257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.770191929710348e-06,
+      "loss": 2.7668,
+      "step": 1258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.76958987953244e-06,
+      "loss": 2.8871,
+      "step": 1259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.76898706036028e-06,
+      "loss": 2.6784,
+      "step": 1260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.768383472291061e-06,
+      "loss": 2.7255,
+      "step": 1261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.767779115422097e-06,
+      "loss": 2.5715,
+      "step": 1262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.767173989850828e-06,
+      "loss": 2.5899,
+      "step": 1263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.766568095674819e-06,
+      "loss": 2.4836,
+      "step": 1264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.765961432991758e-06,
+      "loss": 2.5232,
+      "step": 1265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.765354001899455e-06,
+      "loss": 2.7133,
+      "step": 1266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.764745802495846e-06,
+      "loss": 2.6456,
+      "step": 1267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.764136834878987e-06,
+      "loss": 2.6134,
+      "step": 1268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.763527099147064e-06,
+      "loss": 2.6227,
+      "step": 1269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.762916595398384e-06,
+      "loss": 2.7248,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.762305323731376e-06,
+      "loss": 2.7307,
+      "step": 1271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.761693284244598e-06,
+      "loss": 2.6071,
+      "step": 1272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.761080477036722e-06,
+      "loss": 2.8569,
+      "step": 1273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.760466902206557e-06,
+      "loss": 2.5946,
+      "step": 1274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.759852559853025e-06,
+      "loss": 2.7093,
+      "step": 1275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.759237450075174e-06,
+      "loss": 2.6723,
+      "step": 1276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75862157297218e-06,
+      "loss": 2.4425,
+      "step": 1277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.758004928643342e-06,
+      "loss": 2.6542,
+      "step": 1278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.757387517188076e-06,
+      "loss": 2.5784,
+      "step": 1279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.756769338705927e-06,
+      "loss": 2.5291,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.756150393296566e-06,
+      "loss": 2.7432,
+      "step": 1281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.755530681059784e-06,
+      "loss": 2.5371,
+      "step": 1282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754910202095493e-06,
+      "loss": 2.5031,
+      "step": 1283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754288956503737e-06,
+      "loss": 2.6121,
+      "step": 1284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.753666944384675e-06,
+      "loss": 2.8312,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.753044165838594e-06,
+      "loss": 2.7478,
+      "step": 1286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.752420620965903e-06,
+      "loss": 2.845,
+      "step": 1287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.751796309867139e-06,
+      "loss": 2.5277,
+      "step": 1288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.751171232642954e-06,
+      "loss": 2.7744,
+      "step": 1289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75054538939413e-06,
+      "loss": 2.6891,
+      "step": 1290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.74991878022157e-06,
+      "loss": 2.6358,
+      "step": 1291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.749291405226304e-06,
+      "loss": 2.6578,
+      "step": 1292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.748663264509481e-06,
+      "loss": 2.5,
+      "step": 1293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.748034358172376e-06,
+      "loss": 2.6628,
+      "step": 1294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.747404686316386e-06,
+      "loss": 2.551,
+      "step": 1295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.74677424904303e-06,
+      "loss": 2.9117,
+      "step": 1296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.746143046453956e-06,
+      "loss": 2.6162,
+      "step": 1297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.745511078650931e-06,
+      "loss": 2.6874,
+      "step": 1298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.744878345735846e-06,
+      "loss": 2.54,
+      "step": 1299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.744244847810716e-06,
+      "loss": 2.6775,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.743610584977676e-06,
+      "loss": 2.6283,
+      "step": 1301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.742975557338992e-06,
+      "loss": 2.6796,
+      "step": 1302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.742339764997046e-06,
+      "loss": 2.5238,
+      "step": 1303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.741703208054347e-06,
+      "loss": 2.6554,
+      "step": 1304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.741065886613523e-06,
+      "loss": 2.7491,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.740427800777333e-06,
+      "loss": 2.822,
+      "step": 1306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.739788950648651e-06,
+      "loss": 2.5097,
+      "step": 1307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.739149336330482e-06,
+      "loss": 2.6376,
+      "step": 1308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.738508957925947e-06,
+      "loss": 2.6424,
+      "step": 1309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.737867815538292e-06,
+      "loss": 2.6645,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.737225909270891e-06,
+      "loss": 2.8391,
+      "step": 1311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.736583239227235e-06,
+      "loss": 2.4144,
+      "step": 1312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.735939805510943e-06,
+      "loss": 2.5847,
+      "step": 1313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.735295608225753e-06,
+      "loss": 2.5234,
+      "step": 1314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.73465064747553e-06,
+      "loss": 2.7315,
+      "step": 1315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.734004923364258e-06,
+      "loss": 2.5849,
+      "step": 1316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.733358435996045e-06,
+      "loss": 2.6387,
+      "step": 1317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.732711185475127e-06,
+      "loss": 2.6055,
+      "step": 1318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.732063171905858e-06,
+      "loss": 2.6118,
+      "step": 1319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.731414395392716e-06,
+      "loss": 2.6367,
+      "step": 1320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.730764856040302e-06,
+      "loss": 2.6154,
+      "step": 1321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.73011455395334e-06,
+      "loss": 2.613,
+      "step": 1322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.729463489236679e-06,
+      "loss": 2.5388,
+      "step": 1323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.728811661995287e-06,
+      "loss": 2.5568,
+      "step": 1324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72815907233426e-06,
+      "loss": 2.6106,
+      "step": 1325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72750572035881e-06,
+      "loss": 2.6312,
+      "step": 1326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.726851606174281e-06,
+      "loss": 2.7771,
+      "step": 1327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72619672988613e-06,
+      "loss": 2.6366,
+      "step": 1328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.725541091599946e-06,
+      "loss": 2.7585,
+      "step": 1329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.724884691421433e-06,
+      "loss": 2.5895,
+      "step": 1330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.724227529456424e-06,
+      "loss": 2.7193,
+      "step": 1331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72356960581087e-06,
+      "loss": 2.6253,
+      "step": 1332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72291092059085e-06,
+      "loss": 2.5116,
+      "step": 1333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.722251473902562e-06,
+      "loss": 2.6533,
+      "step": 1334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.721591265852326e-06,
+      "loss": 2.6217,
+      "step": 1335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72093029654659e-06,
+      "loss": 2.7431,
+      "step": 1336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.720268566091915e-06,
+      "loss": 2.5942,
+      "step": 1337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.719606074594999e-06,
+      "loss": 2.5996,
+      "step": 1338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.718942822162646e-06,
+      "loss": 2.5461,
+      "step": 1339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.718278808901797e-06,
+      "loss": 2.6276,
+      "step": 1340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.717614034919509e-06,
+      "loss": 2.6358,
+      "step": 1341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.71694850032296e-06,
+      "loss": 2.5685,
+      "step": 1342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.716282205219457e-06,
+      "loss": 2.7049,
+      "step": 1343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.715615149716424e-06,
+      "loss": 2.7119,
+      "step": 1344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.714947333921407e-06,
+      "loss": 2.4839,
+      "step": 1345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.714278757942082e-06,
+      "loss": 2.4821,
+      "step": 1346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.71360942188624e-06,
+      "loss": 2.7496,
+      "step": 1347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.712939325861794e-06,
+      "loss": 2.7042,
+      "step": 1348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.712268469976789e-06,
+      "loss": 2.4932,
+      "step": 1349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.71159685433938e-06,
+      "loss": 2.8221,
+      "step": 1350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.710924479057855e-06,
+      "loss": 2.5959,
+      "step": 1351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.710251344240619e-06,
+      "loss": 2.5959,
+      "step": 1352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.709577449996199e-06,
+      "loss": 2.6307,
+      "step": 1353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.708902796433249e-06,
+      "loss": 2.4944,
+      "step": 1354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.708227383660538e-06,
+      "loss": 2.5639,
+      "step": 1355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.707551211786966e-06,
+      "loss": 2.489,
+      "step": 1356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.706874280921549e-06,
+      "loss": 2.5778,
+      "step": 1357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70619659117343e-06,
+      "loss": 2.6372,
+      "step": 1358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.705518142651868e-06,
+      "loss": 2.6256,
+      "step": 1359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.704838935466251e-06,
+      "loss": 2.8043,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.704158969726086e-06,
+      "loss": 2.592,
+      "step": 1361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.703478245541005e-06,
+      "loss": 2.6031,
+      "step": 1362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.702796763020757e-06,
+      "loss": 2.6353,
+      "step": 1363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.702114522275216e-06,
+      "loss": 2.6636,
+      "step": 1364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.701431523414382e-06,
+      "loss": 2.371,
+      "step": 1365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.700747766548373e-06,
+      "loss": 2.7056,
+      "step": 1366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.700063251787431e-06,
+      "loss": 2.5786,
+      "step": 1367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.699377979241915e-06,
+      "loss": 2.5416,
+      "step": 1368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.698691949022314e-06,
+      "loss": 2.5366,
+      "step": 1369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.698005161239237e-06,
+      "loss": 2.5828,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.697317616003409e-06,
+      "loss": 2.6523,
+      "step": 1371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.696629313425688e-06,
+      "loss": 2.6129,
+      "step": 1372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.695940253617044e-06,
+      "loss": 2.4794,
+      "step": 1373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.695250436688573e-06,
+      "loss": 2.5515,
+      "step": 1374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.694559862751495e-06,
+      "loss": 2.6366,
+      "step": 1375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.69386853191715e-06,
+      "loss": 2.7393,
+      "step": 1376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.693176444296999e-06,
+      "loss": 2.5332,
+      "step": 1377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.692483600002626e-06,
+      "loss": 2.4741,
+      "step": 1378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.69178999914574e-06,
+      "loss": 2.5597,
+      "step": 1379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.691095641838168e-06,
+      "loss": 2.6521,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.690400528191861e-06,
+      "loss": 2.5744,
+      "step": 1381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.689704658318888e-06,
+      "loss": 2.6384,
+      "step": 1382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.689008032331446e-06,
+      "loss": 2.5182,
+      "step": 1383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.688310650341852e-06,
+      "loss": 2.6784,
+      "step": 1384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68761251246254e-06,
+      "loss": 2.6287,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.686913618806073e-06,
+      "loss": 2.7042,
+      "step": 1386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.686213969485132e-06,
+      "loss": 2.6248,
+      "step": 1387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.685513564612521e-06,
+      "loss": 2.7073,
+      "step": 1388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.684812404301162e-06,
+      "loss": 2.4783,
+      "step": 1389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.684110488664107e-06,
+      "loss": 2.4616,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68340781781452e-06,
+      "loss": 2.6964,
+      "step": 1391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.682704391865697e-06,
+      "loss": 2.6094,
+      "step": 1392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.682000210931046e-06,
+      "loss": 2.684,
+      "step": 1393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.681295275124104e-06,
+      "loss": 2.6348,
+      "step": 1394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.680589584558521e-06,
+      "loss": 2.6252,
+      "step": 1395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.679883139348082e-06,
+      "loss": 2.7184,
+      "step": 1396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.679175939606683e-06,
+      "loss": 2.5765,
+      "step": 1397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.678467985448345e-06,
+      "loss": 2.3808,
+      "step": 1398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.67775927698721e-06,
+      "loss": 2.6887,
+      "step": 1399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.67704981433754e-06,
+      "loss": 2.6965,
+      "step": 1400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.676339597613726e-06,
+      "loss": 2.5613,
+      "step": 1401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.675628626930271e-06,
+      "loss": 2.5554,
+      "step": 1402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.674916902401805e-06,
+      "loss": 2.5328,
+      "step": 1403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.674204424143079e-06,
+      "loss": 2.746,
+      "step": 1404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.673491192268964e-06,
+      "loss": 2.6315,
+      "step": 1405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.672777206894452e-06,
+      "loss": 2.3984,
+      "step": 1406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.672062468134662e-06,
+      "loss": 2.6085,
+      "step": 1407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.671346976104828e-06,
+      "loss": 2.7326,
+      "step": 1408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.670630730920304e-06,
+      "loss": 2.5899,
+      "step": 1409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.669913732696577e-06,
+      "loss": 2.6177,
+      "step": 1410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.66919598154924e-06,
+      "loss": 2.6628,
+      "step": 1411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.668477477594021e-06,
+      "loss": 2.7427,
+      "step": 1412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.66775822094676e-06,
+      "loss": 2.6795,
+      "step": 1413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.66703821172342e-06,
+      "loss": 2.623,
+      "step": 1414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.666317450040093e-06,
+      "loss": 2.4876,
+      "step": 1415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.665595936012982e-06,
+      "loss": 2.6778,
+      "step": 1416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.664873669758415e-06,
+      "loss": 2.588,
+      "step": 1417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.664150651392846e-06,
+      "loss": 2.5699,
+      "step": 1418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.66342688103284e-06,
+      "loss": 2.7762,
+      "step": 1419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.662702358795098e-06,
+      "loss": 2.3558,
+      "step": 1420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.661977084796426e-06,
+      "loss": 2.5766,
+      "step": 1421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.661251059153762e-06,
+      "loss": 2.6752,
+      "step": 1422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.660524281984164e-06,
+      "loss": 2.5678,
+      "step": 1423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.659796753404806e-06,
+      "loss": 2.6222,
+      "step": 1424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.659068473532988e-06,
+      "loss": 2.686,
+      "step": 1425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65833944248613e-06,
+      "loss": 2.759,
+      "step": 1426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65760966038177e-06,
+      "loss": 2.4977,
+      "step": 1427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.656879127337571e-06,
+      "loss": 2.4989,
+      "step": 1428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.656147843471319e-06,
+      "loss": 2.5743,
+      "step": 1429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.655415808900915e-06,
+      "loss": 2.6892,
+      "step": 1430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.654683023744383e-06,
+      "loss": 2.7186,
+      "step": 1431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65394948811987e-06,
+      "loss": 2.5371,
+      "step": 1432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.653215202145643e-06,
+      "loss": 2.7226,
+      "step": 1433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65248016594009e-06,
+      "loss": 2.4539,
+      "step": 1434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.651744379621722e-06,
+      "loss": 2.5593,
+      "step": 1435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.651007843309164e-06,
+      "loss": 2.5544,
+      "step": 1436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.650270557121168e-06,
+      "loss": 2.6385,
+      "step": 1437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64953252117661e-06,
+      "loss": 2.5453,
+      "step": 1438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.648793735594478e-06,
+      "loss": 2.5724,
+      "step": 1439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.648054200493887e-06,
+      "loss": 2.7389,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64731391599407e-06,
+      "loss": 2.6903,
+      "step": 1441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.646572882214385e-06,
+      "loss": 2.6983,
+      "step": 1442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.645831099274304e-06,
+      "loss": 2.6395,
+      "step": 1443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.645088567293426e-06,
+      "loss": 2.5018,
+      "step": 1444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.644345286391468e-06,
+      "loss": 2.7372,
+      "step": 1445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.643601256688267e-06,
+      "loss": 2.4862,
+      "step": 1446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.642856478303784e-06,
+      "loss": 2.5152,
+      "step": 1447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.642110951358097e-06,
+      "loss": 2.5342,
+      "step": 1448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.641364675971407e-06,
+      "loss": 2.6705,
+      "step": 1449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.640617652264033e-06,
+      "loss": 2.6344,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.63986988035642e-06,
+      "loss": 2.7042,
+      "step": 1451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.639121360369127e-06,
+      "loss": 2.6463,
+      "step": 1452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.638372092422839e-06,
+      "loss": 2.6987,
+      "step": 1453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.637622076638358e-06,
+      "loss": 2.6854,
+      "step": 1454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.636871313136608e-06,
+      "loss": 2.5133,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.636119802038634e-06,
+      "loss": 2.6984,
+      "step": 1456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.635367543465603e-06,
+      "loss": 2.616,
+      "step": 1457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.634614537538797e-06,
+      "loss": 2.4556,
+      "step": 1458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.633860784379624e-06,
+      "loss": 2.6933,
+      "step": 1459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.633106284109612e-06,
+      "loss": 2.6485,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.632351036850404e-06,
+      "loss": 2.6258,
+      "step": 1461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.631595042723772e-06,
+      "loss": 2.6719,
+      "step": 1462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.6308383018516e-06,
+      "loss": 2.4835,
+      "step": 1463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.6300808143559e-06,
+      "loss": 2.5818,
+      "step": 1464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.629322580358796e-06,
+      "loss": 2.6605,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.62856359998254e-06,
+      "loss": 2.5854,
+      "step": 1466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.627803873349503e-06,
+      "loss": 2.6581,
+      "step": 1467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.627043400582173e-06,
+      "loss": 2.7113,
+      "step": 1468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.626282181803158e-06,
+      "loss": 2.5465,
+      "step": 1469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.62552021713519e-06,
+      "loss": 2.5895,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.624757506701118e-06,
+      "loss": 2.5225,
+      "step": 1471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.623994050623917e-06,
+      "loss": 2.6917,
+      "step": 1472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.623229849026674e-06,
+      "loss": 2.6314,
+      "step": 1473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.622464902032599e-06,
+      "loss": 2.6384,
+      "step": 1474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.621699209765028e-06,
+      "loss": 2.6022,
+      "step": 1475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.620932772347408e-06,
+      "loss": 2.5904,
+      "step": 1476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.620165589903313e-06,
+      "loss": 2.6037,
+      "step": 1477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.619397662556434e-06,
+      "loss": 2.5978,
+      "step": 1478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.618628990430585e-06,
+      "loss": 2.6549,
+      "step": 1479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.617859573649695e-06,
+      "loss": 2.6927,
+      "step": 1480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.617089412337817e-06,
+      "loss": 2.5409,
+      "step": 1481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.616318506619122e-06,
+      "loss": 2.6058,
+      "step": 1482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.615546856617903e-06,
+      "loss": 2.503,
+      "step": 1483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.614774462458573e-06,
+      "loss": 2.6709,
+      "step": 1484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.614001324265665e-06,
+      "loss": 2.7324,
+      "step": 1485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.613227442163826e-06,
+      "loss": 2.4723,
+      "step": 1486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.612452816277833e-06,
+      "loss": 2.5892,
+      "step": 1487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.611677446732576e-06,
+      "loss": 2.6584,
+      "step": 1488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.610901333653065e-06,
+      "loss": 2.5884,
+      "step": 1489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.610124477164436e-06,
+      "loss": 2.4421,
+      "step": 1490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.609346877391935e-06,
+      "loss": 2.6581,
+      "step": 1491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.608568534460938e-06,
+      "loss": 2.518,
+      "step": 1492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.607789448496932e-06,
+      "loss": 2.7892,
+      "step": 1493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.607009619625532e-06,
+      "loss": 2.5812,
+      "step": 1494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.606229047972465e-06,
+      "loss": 2.7331,
+      "step": 1495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.605447733663586e-06,
+      "loss": 2.5899,
+      "step": 1496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.60466567682486e-06,
+      "loss": 2.6392,
+      "step": 1497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.603882877582382e-06,
+      "loss": 2.5432,
+      "step": 1498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.603099336062357e-06,
+      "loss": 2.5804,
+      "step": 1499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.602315052391116e-06,
+      "loss": 2.5907,
+      "step": 1500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.601530026695109e-06,
+      "loss": 2.5421,
+      "step": 1501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.600744259100905e-06,
+      "loss": 2.643,
+      "step": 1502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.59995774973519e-06,
+      "loss": 2.6593,
+      "step": 1503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.599170498724772e-06,
+      "loss": 2.5082,
+      "step": 1504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.598382506196581e-06,
+      "loss": 2.4918,
+      "step": 1505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.597593772277662e-06,
+      "loss": 2.7278,
+      "step": 1506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.59680429709518e-06,
+      "loss": 2.522,
+      "step": 1507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.596014080776424e-06,
+      "loss": 2.6042,
+      "step": 1508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.595223123448796e-06,
+      "loss": 2.5533,
+      "step": 1509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.594431425239825e-06,
+      "loss": 2.5438,
+      "step": 1510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.593638986277154e-06,
+      "loss": 2.4683,
+      "step": 1511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.592845806688545e-06,
+      "loss": 2.657,
+      "step": 1512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.592051886601883e-06,
+      "loss": 2.5805,
+      "step": 1513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.59125722614517e-06,
+      "loss": 2.7307,
+      "step": 1514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.590461825446527e-06,
+      "loss": 2.6306,
+      "step": 1515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.589665684634197e-06,
+      "loss": 2.562,
+      "step": 1516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58886880383654e-06,
+      "loss": 2.4641,
+      "step": 1517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.588071183182035e-06,
+      "loss": 2.4935,
+      "step": 1518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.587272822799283e-06,
+      "loss": 2.4716,
+      "step": 1519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.586473722817002e-06,
+      "loss": 2.6376,
+      "step": 1520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58567388336403e-06,
+      "loss": 2.7806,
+      "step": 1521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.584873304569322e-06,
+      "loss": 2.4493,
+      "step": 1522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.584071986561958e-06,
+      "loss": 2.7338,
+      "step": 1523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.583269929471129e-06,
+      "loss": 2.7047,
+      "step": 1524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.582467133426153e-06,
+      "loss": 2.7635,
+      "step": 1525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.581663598556461e-06,
+      "loss": 2.5728,
+      "step": 1526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.580859324991607e-06,
+      "loss": 2.6713,
+      "step": 1527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.580054312861264e-06,
+      "loss": 2.5567,
+      "step": 1528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.579248562295222e-06,
+      "loss": 2.7706,
+      "step": 1529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.57844207342339e-06,
+      "loss": 2.6388,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.577634846375798e-06,
+      "loss": 2.6246,
+      "step": 1531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.576826881282595e-06,
+      "loss": 2.7485,
+      "step": 1532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.576018178274047e-06,
+      "loss": 2.6885,
+      "step": 1533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.57520873748054e-06,
+      "loss": 2.585,
+      "step": 1534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.574398559032579e-06,
+      "loss": 2.5497,
+      "step": 1535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.573587643060788e-06,
+      "loss": 2.4683,
+      "step": 1536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.57277598969591e-06,
+      "loss": 2.623,
+      "step": 1537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.571963599068807e-06,
+      "loss": 2.6493,
+      "step": 1538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.571150471310459e-06,
+      "loss": 2.5531,
+      "step": 1539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.570336606551966e-06,
+      "loss": 2.5747,
+      "step": 1540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.569522004924547e-06,
+      "loss": 2.5166,
+      "step": 1541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.568706666559537e-06,
+      "loss": 2.5962,
+      "step": 1542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.567890591588393e-06,
+      "loss": 2.6826,
+      "step": 1543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.56707378014269e-06,
+      "loss": 2.5252,
+      "step": 1544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.56625623235412e-06,
+      "loss": 2.7067,
+      "step": 1545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.565437948354498e-06,
+      "loss": 2.7468,
+      "step": 1546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.56461892827575e-06,
+      "loss": 2.7205,
+      "step": 1547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.56379917224993e-06,
+      "loss": 2.5395,
+      "step": 1548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.562978680409204e-06,
+      "loss": 2.6991,
+      "step": 1549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.56215745288586e-06,
+      "loss": 2.7441,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.561335489812301e-06,
+      "loss": 2.6663,
+      "step": 1551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.560512791321053e-06,
+      "loss": 2.6246,
+      "step": 1552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.559689357544756e-06,
+      "loss": 2.6095,
+      "step": 1553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.558865188616177e-06,
+      "loss": 2.5742,
+      "step": 1554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.558040284668189e-06,
+      "loss": 2.6222,
+      "step": 1555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.557214645833792e-06,
+      "loss": 2.6949,
+      "step": 1556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.556388272246104e-06,
+      "loss": 2.6171,
+      "step": 1557
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.555561164038361e-06,
+      "loss": 2.7241,
+      "step": 1558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.554733321343913e-06,
+      "loss": 2.6961,
+      "step": 1559
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.553904744296232e-06,
+      "loss": 2.5286,
+      "step": 1560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.553075433028913e-06,
+      "loss": 2.5792,
+      "step": 1561
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55224538767566e-06,
+      "loss": 2.6807,
+      "step": 1562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.551414608370303e-06,
+      "loss": 2.8158,
+      "step": 1563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.550583095246786e-06,
+      "loss": 2.6429,
+      "step": 1564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.549750848439172e-06,
+      "loss": 2.6286,
+      "step": 1565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.548917868081644e-06,
+      "loss": 2.4178,
+      "step": 1566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.5480841543085e-06,
+      "loss": 2.5774,
+      "step": 1567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.547249707254166e-06,
+      "loss": 2.7037,
+      "step": 1568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.546414527053167e-06,
+      "loss": 2.5711,
+      "step": 1569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.545578613840168e-06,
+      "loss": 2.6101,
+      "step": 1570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.544741967749937e-06,
+      "loss": 2.5438,
+      "step": 1571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.543904588917366e-06,
+      "loss": 2.5924,
+      "step": 1572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.543066477477467e-06,
+      "loss": 2.6435,
+      "step": 1573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.542227633565364e-06,
+      "loss": 2.607,
+      "step": 1574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.541388057316302e-06,
+      "loss": 2.6511,
+      "step": 1575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.54054774886565e-06,
+      "loss": 2.6576,
+      "step": 1576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.539706708348886e-06,
+      "loss": 2.5765,
+      "step": 1577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53886493590161e-06,
+      "loss": 2.6457,
+      "step": 1578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.538022431659541e-06,
+      "loss": 2.7188,
+      "step": 1579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.537179195758513e-06,
+      "loss": 2.7571,
+      "step": 1580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53633522833448e-06,
+      "loss": 2.5843,
+      "step": 1581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.535490529523514e-06,
+      "loss": 2.5458,
+      "step": 1582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.534645099461805e-06,
+      "loss": 2.5623,
+      "step": 1583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.533798938285663e-06,
+      "loss": 2.5564,
+      "step": 1584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.532952046131507e-06,
+      "loss": 2.6347,
+      "step": 1585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.532104423135885e-06,
+      "loss": 2.6198,
+      "step": 1586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.531256069435459e-06,
+      "loss": 2.5483,
+      "step": 1587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.530406985167005e-06,
+      "loss": 2.5712,
+      "step": 1588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.52955717046742e-06,
+      "loss": 2.6821,
+      "step": 1589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.52870662547372e-06,
+      "loss": 2.4305,
+      "step": 1590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.527855350323035e-06,
+      "loss": 2.56,
+      "step": 1591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.527003345152616e-06,
+      "loss": 2.3647,
+      "step": 1592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.526150610099831e-06,
+      "loss": 2.711,
+      "step": 1593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.525297145302167e-06,
+      "loss": 2.6373,
+      "step": 1594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.524442950897221e-06,
+      "loss": 2.463,
+      "step": 1595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.523588027022721e-06,
+      "loss": 2.5171,
+      "step": 1596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.522732373816499e-06,
+      "loss": 2.5808,
+      "step": 1597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.521875991416514e-06,
+      "loss": 2.6144,
+      "step": 1598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.52101887996084e-06,
+      "loss": 2.778,
+      "step": 1599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.520161039587666e-06,
+      "loss": 2.7144,
+      "step": 1600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.519302470435298e-06,
+      "loss": 2.7252,
+      "step": 1601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.518443172642168e-06,
+      "loss": 2.4493,
+      "step": 1602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.517583146346816e-06,
+      "loss": 2.7483,
+      "step": 1603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.516722391687903e-06,
+      "loss": 2.5071,
+      "step": 1604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.515860908804206e-06,
+      "loss": 2.7041,
+      "step": 1605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.514998697834623e-06,
+      "loss": 2.6237,
+      "step": 1606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.514135758918166e-06,
+      "loss": 2.5838,
+      "step": 1607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.513272092193965e-06,
+      "loss": 2.4795,
+      "step": 1608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.512407697801268e-06,
+      "loss": 2.6168,
+      "step": 1609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.51154257587944e-06,
+      "loss": 2.5734,
+      "step": 1610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.510676726567965e-06,
+      "loss": 2.8275,
+      "step": 1611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.50981015000644e-06,
+      "loss": 2.4319,
+      "step": 1612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.508942846334585e-06,
+      "loss": 2.5992,
+      "step": 1613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.508074815692232e-06,
+      "loss": 2.5862,
+      "step": 1614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.507206058219332e-06,
+      "loss": 2.4351,
+      "step": 1615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.506336574055954e-06,
+      "loss": 2.6975,
+      "step": 1616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.505466363342283e-06,
+      "loss": 2.4826,
+      "step": 1617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.504595426218626e-06,
+      "loss": 2.7028,
+      "step": 1618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.503723762825397e-06,
+      "loss": 2.6407,
+      "step": 1619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.502851373303137e-06,
+      "loss": 2.6229,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.5019782577925e-06,
+      "loss": 2.5398,
+      "step": 1621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.501104416434255e-06,
+      "loss": 2.6138,
+      "step": 1622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.500229849369291e-06,
+      "loss": 2.5243,
+      "step": 1623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.499354556738615e-06,
+      "loss": 2.6005,
+      "step": 1624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.498478538683346e-06,
+      "loss": 2.4422,
+      "step": 1625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.497601795344728e-06,
+      "loss": 2.7324,
+      "step": 1626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.496724326864111e-06,
+      "loss": 2.6163,
+      "step": 1627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.495846133382973e-06,
+      "loss": 2.5485,
+      "step": 1628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.494967215042901e-06,
+      "loss": 2.6753,
+      "step": 1629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.494087571985604e-06,
+      "loss": 2.7141,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.493207204352902e-06,
+      "loss": 2.6547,
+      "step": 1631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.49232611228674e-06,
+      "loss": 2.5318,
+      "step": 1632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.49144429592917e-06,
+      "loss": 2.6309,
+      "step": 1633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.490561755422374e-06,
+      "loss": 2.6446,
+      "step": 1634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.489678490908634e-06,
+      "loss": 2.7225,
+      "step": 1635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.488794502530361e-06,
+      "loss": 2.5662,
+      "step": 1636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.487909790430081e-06,
+      "loss": 2.4752,
+      "step": 1637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.487024354750434e-06,
+      "loss": 2.5272,
+      "step": 1638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.486138195634176e-06,
+      "loss": 2.5857,
+      "step": 1639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.485251313224181e-06,
+      "loss": 2.5848,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.484363707663443e-06,
+      "loss": 2.4172,
+      "step": 1641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.483475379095066e-06,
+      "loss": 2.7155,
+      "step": 1642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.482586327662276e-06,
+      "loss": 2.5455,
+      "step": 1643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.481696553508411e-06,
+      "loss": 2.6111,
+      "step": 1644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.48080605677693e-06,
+      "loss": 2.4677,
+      "step": 1645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.479914837611409e-06,
+      "loss": 2.6357,
+      "step": 1646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.479022896155532e-06,
+      "loss": 2.5689,
+      "step": 1647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.478130232553111e-06,
+      "loss": 2.702,
+      "step": 1648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.477236846948068e-06,
+      "loss": 2.6484,
+      "step": 1649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.476342739484438e-06,
+      "loss": 2.4815,
+      "step": 1650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.47544791030638e-06,
+      "loss": 2.5919,
+      "step": 1651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.474552359558167e-06,
+      "loss": 2.6211,
+      "step": 1652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.473656087384186e-06,
+      "loss": 2.5475,
+      "step": 1653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.472759093928942e-06,
+      "loss": 2.4485,
+      "step": 1654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.471861379337054e-06,
+      "loss": 2.4919,
+      "step": 1655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.470962943753263e-06,
+      "loss": 2.6224,
+      "step": 1656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.470063787322418e-06,
+      "loss": 2.545,
+      "step": 1657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.469163910189494e-06,
+      "loss": 2.5292,
+      "step": 1658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.468263312499574e-06,
+      "loss": 2.6412,
+      "step": 1659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.46736199439786e-06,
+      "loss": 2.4453,
+      "step": 1660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.46645995602967e-06,
+      "loss": 2.554,
+      "step": 1661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.465557197540438e-06,
+      "loss": 2.7085,
+      "step": 1662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.464653719075718e-06,
+      "loss": 2.5804,
+      "step": 1663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.463749520781172e-06,
+      "loss": 2.5613,
+      "step": 1664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.462844602802585e-06,
+      "loss": 2.5723,
+      "step": 1665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.461938965285856e-06,
+      "loss": 2.662,
+      "step": 1666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.461032608376997e-06,
+      "loss": 2.7024,
+      "step": 1667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.460125532222142e-06,
+      "loss": 2.6135,
+      "step": 1668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.459217736967535e-06,
+      "loss": 2.579,
+      "step": 1669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.458309222759542e-06,
+      "loss": 2.5759,
+      "step": 1670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.457399989744635e-06,
+      "loss": 2.59,
+      "step": 1671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.456490038069417e-06,
+      "loss": 2.5522,
+      "step": 1672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45557936788059e-06,
+      "loss": 2.5628,
+      "step": 1673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.454667979324987e-06,
+      "loss": 2.6187,
+      "step": 1674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.453755872549545e-06,
+      "loss": 2.4746,
+      "step": 1675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.452843047701324e-06,
+      "loss": 2.6569,
+      "step": 1676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.451929504927496e-06,
+      "loss": 2.7459,
+      "step": 1677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.451015244375352e-06,
+      "loss": 2.5345,
+      "step": 1678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.450100266192295e-06,
+      "loss": 2.5932,
+      "step": 1679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.449184570525847e-06,
+      "loss": 2.6893,
+      "step": 1680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.448268157523645e-06,
+      "loss": 2.7194,
+      "step": 1681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44735102733344e-06,
+      "loss": 2.8082,
+      "step": 1682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.446433180103098e-06,
+      "loss": 2.4355,
+      "step": 1683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.445514615980604e-06,
+      "loss": 2.5808,
+      "step": 1684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.444595335114058e-06,
+      "loss": 2.5742,
+      "step": 1685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44367533765167e-06,
+      "loss": 2.478,
+      "step": 1686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.442754623741775e-06,
+      "loss": 2.6777,
+      "step": 1687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.441833193532817e-06,
+      "loss": 2.7856,
+      "step": 1688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.440911047173355e-06,
+      "loss": 2.7012,
+      "step": 1689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.439988184812065e-06,
+      "loss": 2.5071,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.439064606597743e-06,
+      "loss": 2.4894,
+      "step": 1691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.438140312679292e-06,
+      "loss": 2.561,
+      "step": 1692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.437215303205736e-06,
+      "loss": 2.85,
+      "step": 1693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.436289578326215e-06,
+      "loss": 2.5594,
+      "step": 1694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.435363138189976e-06,
+      "loss": 2.6036,
+      "step": 1695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.434435982946397e-06,
+      "loss": 2.5689,
+      "step": 1696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.433508112744953e-06,
+      "loss": 2.6231,
+      "step": 1697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.432579527735251e-06,
+      "loss": 2.4876,
+      "step": 1698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.431650228066999e-06,
+      "loss": 2.5626,
+      "step": 1699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.43072021389003e-06,
+      "loss": 2.6454,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.429789485354287e-06,
+      "loss": 2.5638,
+      "step": 1701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.428858042609832e-06,
+      "loss": 2.7173,
+      "step": 1702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.42792588580684e-06,
+      "loss": 2.7663,
+      "step": 1703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.4269930150956e-06,
+      "loss": 2.6596,
+      "step": 1704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.426059430626518e-06,
+      "loss": 2.6069,
+      "step": 1705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.425125132550115e-06,
+      "loss": 2.4881,
+      "step": 1706
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.424190121017027e-06,
+      "loss": 2.6046,
+      "step": 1707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.423254396178003e-06,
+      "loss": 2.5127,
+      "step": 1708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.422317958183908e-06,
+      "loss": 2.6742,
+      "step": 1709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.421380807185727e-06,
+      "loss": 2.5407,
+      "step": 1710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.420442943334551e-06,
+      "loss": 2.5994,
+      "step": 1711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.419504366781591e-06,
+      "loss": 2.6025,
+      "step": 1712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.418565077678174e-06,
+      "loss": 2.7089,
+      "step": 1713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.417625076175741e-06,
+      "loss": 2.5048,
+      "step": 1714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.416684362425846e-06,
+      "loss": 2.5556,
+      "step": 1715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.415742936580156e-06,
+      "loss": 2.5666,
+      "step": 1716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.41480079879046e-06,
+      "loss": 2.5515,
+      "step": 1717
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.413857949208655e-06,
+      "loss": 2.7239,
+      "step": 1718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.412914387986758e-06,
+      "loss": 2.4883,
+      "step": 1719
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.411970115276897e-06,
+      "loss": 2.6383,
+      "step": 1720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.411025131231313e-06,
+      "loss": 2.5343,
+      "step": 1721
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.410079436002367e-06,
+      "loss": 2.6647,
+      "step": 1722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.409133029742532e-06,
+      "loss": 2.539,
+      "step": 1723
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.408185912604395e-06,
+      "loss": 2.6446,
+      "step": 1724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.407238084740658e-06,
+      "loss": 2.4825,
+      "step": 1725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.40628954630414e-06,
+      "loss": 2.5409,
+      "step": 1726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.405340297447772e-06,
+      "loss": 2.647,
+      "step": 1727
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.404390338324599e-06,
+      "loss": 2.5909,
+      "step": 1728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.40343966908778e-06,
+      "loss": 2.7705,
+      "step": 1729
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.402488289890593e-06,
+      "loss": 2.544,
+      "step": 1730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.401536200886426e-06,
+      "loss": 2.6641,
+      "step": 1731
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.400583402228785e-06,
+      "loss": 2.5942,
+      "step": 1732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.399629894071285e-06,
+      "loss": 2.8705,
+      "step": 1733
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.398675676567662e-06,
+      "loss": 2.5475,
+      "step": 1734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.39772074987176e-06,
+      "loss": 2.6146,
+      "step": 1735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.396765114137544e-06,
+      "loss": 2.5474,
+      "step": 1736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.395808769519086e-06,
+      "loss": 2.6273,
+      "step": 1737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.394851716170582e-06,
+      "loss": 2.3296,
+      "step": 1738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.393893954246328e-06,
+      "loss": 2.5264,
+      "step": 1739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.39293548390075e-06,
+      "loss": 2.5208,
+      "step": 1740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.391976305288376e-06,
+      "loss": 2.5311,
+      "step": 1741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.391016418563856e-06,
+      "loss": 2.5949,
+      "step": 1742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.39005582388195e-06,
+      "loss": 2.664,
+      "step": 1743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.389094521397532e-06,
+      "loss": 2.6469,
+      "step": 1744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.388132511265594e-06,
+      "loss": 2.3803,
+      "step": 1745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.38716979364124e-06,
+      "loss": 2.815,
+      "step": 1746
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.386206368679684e-06,
+      "loss": 2.6434,
+      "step": 1747
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.385242236536259e-06,
+      "loss": 2.6816,
+      "step": 1748
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.384277397366413e-06,
+      "loss": 2.6157,
+      "step": 1749
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.383311851325702e-06,
+      "loss": 2.6216,
+      "step": 1750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.382345598569805e-06,
+      "loss": 2.6737,
+      "step": 1751
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.381378639254504e-06,
+      "loss": 2.5271,
+      "step": 1752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.380410973535702e-06,
+      "loss": 2.7343,
+      "step": 1753
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.379442601569415e-06,
+      "loss": 2.5608,
+      "step": 1754
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.378473523511774e-06,
+      "loss": 2.623,
+      "step": 1755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.377503739519019e-06,
+      "loss": 2.709,
+      "step": 1756
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.37653324974751e-06,
+      "loss": 2.7383,
+      "step": 1757
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.375562054353714e-06,
+      "loss": 2.6502,
+      "step": 1758
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.37459015349422e-06,
+      "loss": 2.6749,
+      "step": 1759
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.37361754732572e-06,
+      "loss": 2.725,
+      "step": 1760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.37264423600503e-06,
+      "loss": 2.5349,
+      "step": 1761
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.371670219689077e-06,
+      "loss": 2.5546,
+      "step": 1762
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.370695498534898e-06,
+      "loss": 2.5613,
+      "step": 1763
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.369720072699648e-06,
+      "loss": 2.5287,
+      "step": 1764
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.36874394234059e-06,
+      "loss": 2.6847,
+      "step": 1765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.367767107615106e-06,
+      "loss": 2.6523,
+      "step": 1766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.36678956868069e-06,
+      "loss": 2.7404,
+      "step": 1767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.365811325694949e-06,
+      "loss": 2.6472,
+      "step": 1768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.364832378815602e-06,
+      "loss": 2.4614,
+      "step": 1769
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.363852728200488e-06,
+      "loss": 2.5167,
+      "step": 1770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.36287237400755e-06,
+      "loss": 2.7275,
+      "step": 1771
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.36189131639485e-06,
+      "loss": 2.6802,
+      "step": 1772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.360909555520565e-06,
+      "loss": 2.7541,
+      "step": 1773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.35992709154298e-06,
+      "loss": 2.5189,
+      "step": 1774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.358943924620498e-06,
+      "loss": 2.5649,
+      "step": 1775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.357960054911633e-06,
+      "loss": 2.5926,
+      "step": 1776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.356975482575013e-06,
+      "loss": 2.5905,
+      "step": 1777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.355990207769378e-06,
+      "loss": 2.7679,
+      "step": 1778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.355004230653586e-06,
+      "loss": 2.4523,
+      "step": 1779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.354017551386599e-06,
+      "loss": 2.6743,
+      "step": 1780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.353030170127505e-06,
+      "loss": 2.5748,
+      "step": 1781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.352042087035491e-06,
+      "loss": 2.5235,
+      "step": 1782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.351053302269871e-06,
+      "loss": 2.5453,
+      "step": 1783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.35006381599006e-06,
+      "loss": 2.8055,
+      "step": 1784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.349073628355593e-06,
+      "loss": 2.5441,
+      "step": 1785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.348082739526116e-06,
+      "loss": 2.569,
+      "step": 1786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.347091149661392e-06,
+      "loss": 2.5793,
+      "step": 1787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.346098858921292e-06,
+      "loss": 2.5131,
+      "step": 1788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.345105867465798e-06,
+      "loss": 2.6083,
+      "step": 1789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.344112175455013e-06,
+      "loss": 2.5978,
+      "step": 1790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.343117783049148e-06,
+      "loss": 2.5766,
+      "step": 1791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.342122690408526e-06,
+      "loss": 2.5388,
+      "step": 1792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.341126897693583e-06,
+      "loss": 2.6094,
+      "step": 1793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.340130405064873e-06,
+      "loss": 2.459,
+      "step": 1794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.339133212683058e-06,
+      "loss": 2.5272,
+      "step": 1795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.338135320708912e-06,
+      "loss": 2.4644,
+      "step": 1796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.337136729303324e-06,
+      "loss": 2.7315,
+      "step": 1797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.336137438627299e-06,
+      "loss": 2.7504,
+      "step": 1798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.335137448841947e-06,
+      "loss": 2.6567,
+      "step": 1799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.334136760108496e-06,
+      "loss": 2.6186,
+      "step": 1800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.333135372588287e-06,
+      "loss": 2.6296,
+      "step": 1801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.332133286442772e-06,
+      "loss": 2.8078,
+      "step": 1802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.331130501833516e-06,
+      "loss": 2.5569,
+      "step": 1803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.330127018922195e-06,
+      "loss": 2.5636,
+      "step": 1804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.329122837870599e-06,
+      "loss": 2.701,
+      "step": 1805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.328117958840632e-06,
+      "loss": 2.6382,
+      "step": 1806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.327112381994312e-06,
+      "loss": 2.7414,
+      "step": 1807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.326106107493762e-06,
+      "loss": 2.7015,
+      "step": 1808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.325099135501223e-06,
+      "loss": 2.4536,
+      "step": 1809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.324091466179051e-06,
+      "loss": 2.4581,
+      "step": 1810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.323083099689705e-06,
+      "loss": 2.5679,
+      "step": 1811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.32207403619577e-06,
+      "loss": 2.9358,
+      "step": 1812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.32106427585993e-06,
+      "loss": 2.6042,
+      "step": 1813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.32005381884499e-06,
+      "loss": 2.6182,
+      "step": 1814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.319042665313865e-06,
+      "loss": 2.6229,
+      "step": 1815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.318030815429579e-06,
+      "loss": 2.5722,
+      "step": 1816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.317018269355276e-06,
+      "loss": 2.7883,
+      "step": 1817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.316005027254202e-06,
+      "loss": 2.607,
+      "step": 1818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.314991089289722e-06,
+      "loss": 2.421,
+      "step": 1819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.313976455625316e-06,
+      "loss": 2.7702,
+      "step": 1820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.312961126424565e-06,
+      "loss": 2.6267,
+      "step": 1821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.311945101851175e-06,
+      "loss": 2.5482,
+      "step": 1822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.310928382068955e-06,
+      "loss": 2.623,
+      "step": 1823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.309910967241831e-06,
+      "loss": 2.6183,
+      "step": 1824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.30889285753384e-06,
+      "loss": 2.5061,
+      "step": 1825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.307874053109127e-06,
+      "loss": 2.4959,
+      "step": 1826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.306854554131954e-06,
+      "loss": 2.6524,
+      "step": 1827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.305834360766695e-06,
+      "loss": 2.6139,
+      "step": 1828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.304813473177834e-06,
+      "loss": 2.6624,
+      "step": 1829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.303791891529966e-06,
+      "loss": 2.5326,
+      "step": 1830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.302769615987798e-06,
+      "loss": 2.6291,
+      "step": 1831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.301746646716152e-06,
+      "loss": 2.4875,
+      "step": 1832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.30072298387996e-06,
+      "loss": 2.5696,
+      "step": 1833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.299698627644266e-06,
+      "loss": 2.4787,
+      "step": 1834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.298673578174225e-06,
+      "loss": 2.5011,
+      "step": 1835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.297647835635102e-06,
+      "loss": 2.5731,
+      "step": 1836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.29662140019228e-06,
+      "loss": 2.6629,
+      "step": 1837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.295594272011246e-06,
+      "loss": 2.5676,
+      "step": 1838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.294566451257607e-06,
+      "loss": 2.3752,
+      "step": 1839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.293537938097073e-06,
+      "loss": 2.6747,
+      "step": 1840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.29250873269547e-06,
+      "loss": 2.6185,
+      "step": 1841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.291478835218737e-06,
+      "loss": 2.6347,
+      "step": 1842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.290448245832922e-06,
+      "loss": 2.6457,
+      "step": 1843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.289416964704186e-06,
+      "loss": 2.5412,
+      "step": 1844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.288384991998801e-06,
+      "loss": 2.5117,
+      "step": 1845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.28735232788315e-06,
+      "loss": 2.5874,
+      "step": 1846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.286318972523728e-06,
+      "loss": 2.5137,
+      "step": 1847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.285284926087144e-06,
+      "loss": 2.5435,
+      "step": 1848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.284250188740111e-06,
+      "loss": 2.7115,
+      "step": 1849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.283214760649462e-06,
+      "loss": 2.5199,
+      "step": 1850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.282178641982136e-06,
+      "loss": 2.4881,
+      "step": 1851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.281141832905185e-06,
+      "loss": 2.6333,
+      "step": 1852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.280104333585771e-06,
+      "loss": 2.6317,
+      "step": 1853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.279066144191173e-06,
+      "loss": 2.6584,
+      "step": 1854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.278027264888774e-06,
+      "loss": 2.6966,
+      "step": 1855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.27698769584607e-06,
+      "loss": 2.523,
+      "step": 1856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.27594743723067e-06,
+      "loss": 2.5527,
+      "step": 1857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.274906489210294e-06,
+      "loss": 2.5175,
+      "step": 1858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.273864851952771e-06,
+      "loss": 2.4365,
+      "step": 1859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.272822525626047e-06,
+      "loss": 2.7369,
+      "step": 1860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.271779510398168e-06,
+      "loss": 2.4506,
+      "step": 1861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.270735806437306e-06,
+      "loss": 2.514,
+      "step": 1862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.269691413911727e-06,
+      "loss": 2.5532,
+      "step": 1863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.268646332989826e-06,
+      "loss": 2.7467,
+      "step": 1864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.267600563840093e-06,
+      "loss": 2.6881,
+      "step": 1865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.266554106631137e-06,
+      "loss": 2.6248,
+      "step": 1866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26550696153168e-06,
+      "loss": 2.5816,
+      "step": 1867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26445912871055e-06,
+      "loss": 2.6071,
+      "step": 1868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.263410608336687e-06,
+      "loss": 2.5906,
+      "step": 1869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.262361400579145e-06,
+      "loss": 2.8241,
+      "step": 1870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26131150560708e-06,
+      "loss": 2.633,
+      "step": 1871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.260260923589773e-06,
+      "loss": 2.5108,
+      "step": 1872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.259209654696602e-06,
+      "loss": 2.5262,
+      "step": 1873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.258157699097065e-06,
+      "loss": 2.8399,
+      "step": 1874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.257105056960767e-06,
+      "loss": 2.447,
+      "step": 1875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.25605172845742e-06,
+      "loss": 2.5732,
+      "step": 1876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.254997713756857e-06,
+      "loss": 2.4732,
+      "step": 1877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.25394301302901e-06,
+      "loss": 2.4912,
+      "step": 1878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.25288762644393e-06,
+      "loss": 2.3482,
+      "step": 1879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.251831554171773e-06,
+      "loss": 2.7347,
+      "step": 1880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.250774796382812e-06,
+      "loss": 2.7068,
+      "step": 1881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.249717353247422e-06,
+      "loss": 2.7201,
+      "step": 1882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.248659224936095e-06,
+      "loss": 2.4898,
+      "step": 1883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.247600411619434e-06,
+      "loss": 2.6139,
+      "step": 1884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.246540913468147e-06,
+      "loss": 2.4889,
+      "step": 1885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.245480730653055e-06,
+      "loss": 2.6162,
+      "step": 1886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.24441986334509e-06,
+      "loss": 2.4791,
+      "step": 1887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.243358311715298e-06,
+      "loss": 2.5717,
+      "step": 1888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.242296075934825e-06,
+      "loss": 2.5317,
+      "step": 1889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.24123315617494e-06,
+      "loss": 2.5668,
+      "step": 1890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.240169552607012e-06,
+      "loss": 2.7414,
+      "step": 1891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.239105265402525e-06,
+      "loss": 2.5536,
+      "step": 1892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.238040294733074e-06,
+      "loss": 2.7504,
+      "step": 1893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.236974640770362e-06,
+      "loss": 2.6142,
+      "step": 1894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.235908303686204e-06,
+      "loss": 2.8001,
+      "step": 1895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.23484128365252e-06,
+      "loss": 2.5903,
+      "step": 1896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.233773580841348e-06,
+      "loss": 2.5884,
+      "step": 1897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.232705195424832e-06,
+      "loss": 2.4989,
+      "step": 1898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.231636127575224e-06,
+      "loss": 2.806,
+      "step": 1899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.23056637746489e-06,
+      "loss": 2.6711,
+      "step": 1900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.229495945266304e-06,
+      "loss": 2.6043,
+      "step": 1901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.22842483115205e-06,
+      "loss": 2.6686,
+      "step": 1902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.227353035294823e-06,
+      "loss": 2.5233,
+      "step": 1903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.226280557867428e-06,
+      "loss": 2.4793,
+      "step": 1904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.225207399042775e-06,
+      "loss": 2.4527,
+      "step": 1905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.224133558993894e-06,
+      "loss": 2.583,
+      "step": 1906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.22305903789391e-06,
+      "loss": 2.4216,
+      "step": 1907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.221983835916074e-06,
+      "loss": 2.3007,
+      "step": 1908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.22090795323374e-06,
+      "loss": 2.493,
+      "step": 1909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.219831390020367e-06,
+      "loss": 2.8029,
+      "step": 1910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.218754146449526e-06,
+      "loss": 2.9164,
+      "step": 1911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.217676222694905e-06,
+      "loss": 2.6063,
+      "step": 1912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.216597618930292e-06,
+      "loss": 2.5349,
+      "step": 1913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.215518335329591e-06,
+      "loss": 2.5932,
+      "step": 1914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.214438372066812e-06,
+      "loss": 2.5948,
+      "step": 1915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.213357729316077e-06,
+      "loss": 2.6243,
+      "step": 1916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.212276407251616e-06,
+      "loss": 2.6098,
+      "step": 1917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.21119440604777e-06,
+      "loss": 2.7271,
+      "step": 1918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.210111725878986e-06,
+      "loss": 2.4493,
+      "step": 1919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.209028366919825e-06,
+      "loss": 2.5795,
+      "step": 1920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.207944329344956e-06,
+      "loss": 2.5114,
+      "step": 1921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.206859613329155e-06,
+      "loss": 2.4511,
+      "step": 1922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.205774219047309e-06,
+      "loss": 2.6197,
+      "step": 1923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.204688146674418e-06,
+      "loss": 2.5737,
+      "step": 1924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.203601396385584e-06,
+      "loss": 2.5748,
+      "step": 1925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.202513968356027e-06,
+      "loss": 2.5756,
+      "step": 1926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.201425862761065e-06,
+      "loss": 2.6434,
+      "step": 1927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.200337079776136e-06,
+      "loss": 2.5419,
+      "step": 1928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.199247619576783e-06,
+      "loss": 2.6994,
+      "step": 1929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.198157482338655e-06,
+      "loss": 2.7211,
+      "step": 1930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.197066668237518e-06,
+      "loss": 2.4605,
+      "step": 1931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.195975177449238e-06,
+      "loss": 2.6158,
+      "step": 1932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.194883010149797e-06,
+      "loss": 2.4616,
+      "step": 1933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.193790166515282e-06,
+      "loss": 2.6718,
+      "step": 1934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.192696646721893e-06,
+      "loss": 2.6205,
+      "step": 1935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.191602450945935e-06,
+      "loss": 2.6955,
+      "step": 1936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.190507579363823e-06,
+      "loss": 2.399,
+      "step": 1937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.189412032152083e-06,
+      "loss": 2.5756,
+      "step": 1938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.188315809487349e-06,
+      "loss": 2.5637,
+      "step": 1939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.187218911546363e-06,
+      "loss": 2.7103,
+      "step": 1940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.186121338505973e-06,
+      "loss": 2.4783,
+      "step": 1941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.185023090543144e-06,
+      "loss": 2.715,
+      "step": 1942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.183924167834943e-06,
+      "loss": 2.6709,
+      "step": 1943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.182824570558546e-06,
+      "loss": 2.6041,
+      "step": 1944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.181724298891243e-06,
+      "loss": 2.5058,
+      "step": 1945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.180623353010428e-06,
+      "loss": 2.592,
+      "step": 1946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.179521733093602e-06,
+      "loss": 2.5656,
+      "step": 1947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.178419439318382e-06,
+      "loss": 2.512,
+      "step": 1948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.177316471862485e-06,
+      "loss": 2.5563,
+      "step": 1949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.176212830903746e-06,
+      "loss": 2.7808,
+      "step": 1950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.175108516620099e-06,
+      "loss": 2.7265,
+      "step": 1951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.174003529189592e-06,
+      "loss": 2.5679,
+      "step": 1952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.172897868790383e-06,
+      "loss": 2.3957,
+      "step": 1953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.171791535600733e-06,
+      "loss": 2.641,
+      "step": 1954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.170684529799017e-06,
+      "loss": 2.4124,
+      "step": 1955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.169576851563715e-06,
+      "loss": 2.7982,
+      "step": 1956
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.168468501073415e-06,
+      "loss": 2.5457,
+      "step": 1957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.167359478506816e-06,
+      "loss": 2.6227,
+      "step": 1958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.166249784042725e-06,
+      "loss": 2.5325,
+      "step": 1959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.165139417860057e-06,
+      "loss": 2.4889,
+      "step": 1960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.16402838013783e-06,
+      "loss": 2.5367,
+      "step": 1961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.162916671055182e-06,
+      "loss": 2.4938,
+      "step": 1962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.161804290791346e-06,
+      "loss": 2.5626,
+      "step": 1963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.160691239525675e-06,
+      "loss": 2.7628,
+      "step": 1964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.15957751743762e-06,
+      "loss": 2.5254,
+      "step": 1965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.158463124706746e-06,
+      "loss": 2.3984,
+      "step": 1966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.157348061512728e-06,
+      "loss": 2.4631,
+      "step": 1967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.156232328035342e-06,
+      "loss": 2.7109,
+      "step": 1968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.155115924454478e-06,
+      "loss": 2.4724,
+      "step": 1969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.153998850950133e-06,
+      "loss": 2.538,
+      "step": 1970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.15288110770241e-06,
+      "loss": 2.6395,
+      "step": 1971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.151762694891522e-06,
+      "loss": 2.5424,
+      "step": 1972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.150643612697789e-06,
+      "loss": 2.4426,
+      "step": 1973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.149523861301636e-06,
+      "loss": 2.5139,
+      "step": 1974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.148403440883604e-06,
+      "loss": 2.5783,
+      "step": 1975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.147282351624334e-06,
+      "loss": 2.6957,
+      "step": 1976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.146160593704578e-06,
+      "loss": 2.5462,
+      "step": 1977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.145038167305196e-06,
+      "loss": 2.7778,
+      "step": 1978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.143915072607154e-06,
+      "loss": 2.6598,
+      "step": 1979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.142791309791528e-06,
+      "loss": 2.6909,
+      "step": 1980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.1416668790395e-06,
+      "loss": 2.5852,
+      "step": 1981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.140541780532362e-06,
+      "loss": 2.4709,
+      "step": 1982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.139416014451508e-06,
+      "loss": 2.5523,
+      "step": 1983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.138289580978449e-06,
+      "loss": 2.6897,
+      "step": 1984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.137162480294795e-06,
+      "loss": 2.515,
+      "step": 1985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.136034712582267e-06,
+      "loss": 2.608,
+      "step": 1986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.134906278022694e-06,
+      "loss": 2.5264,
+      "step": 1987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.133777176798013e-06,
+      "loss": 2.5854,
+      "step": 1988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.132647409090267e-06,
+      "loss": 2.626,
+      "step": 1989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.131516975081604e-06,
+      "loss": 2.6434,
+      "step": 1990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.130385874954286e-06,
+      "loss": 2.4179,
+      "step": 1991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.129254108890678e-06,
+      "loss": 2.5204,
+      "step": 1992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.128121677073251e-06,
+      "loss": 2.5353,
+      "step": 1993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.126988579684587e-06,
+      "loss": 2.7299,
+      "step": 1994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.125854816907371e-06,
+      "loss": 2.6015,
+      "step": 1995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.124720388924403e-06,
+      "loss": 2.4297,
+      "step": 1996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.123585295918581e-06,
+      "loss": 2.644,
+      "step": 1997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.122449538072918e-06,
+      "loss": 2.6259,
+      "step": 1998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.121313115570525e-06,
+      "loss": 2.674,
+      "step": 1999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.120176028594632e-06,
+      "loss": 2.5436,
+      "step": 2000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.119038277328568e-06,
+      "loss": 2.4888,
+      "step": 2001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.117899861955767e-06,
+      "loss": 2.7402,
+      "step": 2002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.116760782659781e-06,
+      "loss": 2.6142,
+      "step": 2003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.115621039624256e-06,
+      "loss": 2.7426,
+      "step": 2004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.114480633032956e-06,
+      "loss": 2.5397,
+      "step": 2005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.113339563069742e-06,
+      "loss": 2.5849,
+      "step": 2006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.112197829918591e-06,
+      "loss": 2.7125,
+      "step": 2007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.111055433763582e-06,
+      "loss": 2.6083,
+      "step": 2008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.109912374788904e-06,
+      "loss": 2.6097,
+      "step": 2009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.108768653178848e-06,
+      "loss": 2.733,
+      "step": 2010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.107624269117813e-06,
+      "loss": 2.3806,
+      "step": 2011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.106479222790312e-06,
+      "loss": 2.4341,
+      "step": 2012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.105333514380953e-06,
+      "loss": 2.4488,
+      "step": 2013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.104187144074461e-06,
+      "loss": 2.6244,
+      "step": 2014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.103040112055663e-06,
+      "loss": 2.6491,
+      "step": 2015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.101892418509492e-06,
+      "loss": 2.5754,
+      "step": 2016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.100744063620991e-06,
+      "loss": 2.562,
+      "step": 2017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.099595047575308e-06,
+      "loss": 2.4391,
+      "step": 2018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.098445370557694e-06,
+      "loss": 2.6503,
+      "step": 2019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.09729503275351e-06,
+      "loss": 2.5,
+      "step": 2020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.096144034348227e-06,
+      "loss": 2.4237,
+      "step": 2021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.094992375527415e-06,
+      "loss": 2.8561,
+      "step": 2022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.093840056476759e-06,
+      "loss": 2.5787,
+      "step": 2023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.092687077382043e-06,
+      "loss": 2.6063,
+      "step": 2024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.091533438429159e-06,
+      "loss": 2.5558,
+      "step": 2025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.090379139804107e-06,
+      "loss": 2.6062,
+      "step": 2026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.089224181692992e-06,
+      "loss": 2.5327,
+      "step": 2027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08806856428203e-06,
+      "loss": 2.7563,
+      "step": 2028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.086912287757538e-06,
+      "loss": 2.555,
+      "step": 2029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08575535230594e-06,
+      "loss": 2.6447,
+      "step": 2030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08459775811377e-06,
+      "loss": 2.5739,
+      "step": 2031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.083439505367658e-06,
+      "loss": 2.7103,
+      "step": 2032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.082280594254353e-06,
+      "loss": 2.643,
+      "step": 2033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.081121024960707e-06,
+      "loss": 2.6475,
+      "step": 2034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.079960797673671e-06,
+      "loss": 2.4778,
+      "step": 2035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.078799912580305e-06,
+      "loss": 2.7608,
+      "step": 2036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.077638369867783e-06,
+      "loss": 2.6815,
+      "step": 2037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.076476169723376e-06,
+      "loss": 2.6208,
+      "step": 2038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.075313312334465e-06,
+      "loss": 2.6385,
+      "step": 2039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.074149797888532e-06,
+      "loss": 2.6229,
+      "step": 2040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.072985626573171e-06,
+      "loss": 2.5069,
+      "step": 2041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.071820798576083e-06,
+      "loss": 2.6007,
+      "step": 2042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.070655314085068e-06,
+      "loss": 2.3999,
+      "step": 2043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.069489173288037e-06,
+      "loss": 2.5186,
+      "step": 2044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.068322376373004e-06,
+      "loss": 2.5328,
+      "step": 2045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.06715492352809e-06,
+      "loss": 2.6631,
+      "step": 2046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.065986814941523e-06,
+      "loss": 2.5228,
+      "step": 2047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.064818050801634e-06,
+      "loss": 2.711,
+      "step": 2048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.063648631296861e-06,
+      "loss": 2.5728,
+      "step": 2049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.062478556615751e-06,
+      "loss": 2.3394,
+      "step": 2050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.06130782694695e-06,
+      "loss": 2.6653,
+      "step": 2051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.060136442479215e-06,
+      "loss": 2.6855,
+      "step": 2052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.058964403401407e-06,
+      "loss": 2.7237,
+      "step": 2053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.057791709902491e-06,
+      "loss": 2.5775,
+      "step": 2054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.056618362171537e-06,
+      "loss": 2.6503,
+      "step": 2055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.055444360397727e-06,
+      "loss": 2.722,
+      "step": 2056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.05426970477034e-06,
+      "loss": 2.5831,
+      "step": 2057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.053094395478763e-06,
+      "loss": 2.6116,
+      "step": 2058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.051918432712492e-06,
+      "loss": 2.8308,
+      "step": 2059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.050741816661128e-06,
+      "loss": 2.5763,
+      "step": 2060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.04956454751437e-06,
+      "loss": 2.5228,
+      "step": 2061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.048386625462028e-06,
+      "loss": 2.6537,
+      "step": 2062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.047208050694021e-06,
+      "loss": 2.7104,
+      "step": 2063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.046028823400368e-06,
+      "loss": 2.7202,
+      "step": 2064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.04484894377119e-06,
+      "loss": 2.7263,
+      "step": 2065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.04366841199672e-06,
+      "loss": 2.6479,
+      "step": 2066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.042487228267297e-06,
+      "loss": 2.8151,
+      "step": 2067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.041305392773355e-06,
+      "loss": 2.6305,
+      "step": 2068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.040122905705443e-06,
+      "loss": 2.5059,
+      "step": 2069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.03893976725421e-06,
+      "loss": 2.4531,
+      "step": 2070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.037755977610413e-06,
+      "loss": 2.7036,
+      "step": 2071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.036571536964915e-06,
+      "loss": 2.5204,
+      "step": 2072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.035386445508681e-06,
+      "loss": 2.6363,
+      "step": 2073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.034200703432776e-06,
+      "loss": 2.5696,
+      "step": 2074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.033014310928383e-06,
+      "loss": 2.7125,
+      "step": 2075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.03182726818678e-06,
+      "loss": 2.6088,
+      "step": 2076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.030639575399347e-06,
+      "loss": 2.621,
+      "step": 2077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.029451232757581e-06,
+      "loss": 2.5088,
+      "step": 2078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.028262240453074e-06,
+      "loss": 2.4172,
+      "step": 2079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.027072598677526e-06,
+      "loss": 2.7664,
+      "step": 2080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.025882307622741e-06,
+      "loss": 2.6369,
+      "step": 2081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.024691367480627e-06,
+      "loss": 2.5216,
+      "step": 2082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.0234997784432e-06,
+      "loss": 2.6354,
+      "step": 2083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.022307540702576e-06,
+      "loss": 2.6607,
+      "step": 2084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.021114654450978e-06,
+      "loss": 2.5529,
+      "step": 2085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.019921119880735e-06,
+      "loss": 2.457,
+      "step": 2086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.018726937184278e-06,
+      "loss": 2.4658,
+      "step": 2087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.017532106554143e-06,
+      "loss": 2.5262,
+      "step": 2088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01633662818297e-06,
+      "loss": 2.6535,
+      "step": 2089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.015140502263506e-06,
+      "loss": 2.7527,
+      "step": 2090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.0139437289886e-06,
+      "loss": 2.7131,
+      "step": 2091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.012746308551208e-06,
+      "loss": 2.6539,
+      "step": 2092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.011548241144386e-06,
+      "loss": 2.6242,
+      "step": 2093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.010349526961295e-06,
+      "loss": 2.7233,
+      "step": 2094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.009150166195206e-06,
+      "loss": 2.5733,
+      "step": 2095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.007950159039488e-06,
+      "loss": 2.6706,
+      "step": 2096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.006749505687618e-06,
+      "loss": 2.7089,
+      "step": 2097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.005548206333172e-06,
+      "loss": 2.6481,
+      "step": 2098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.004346261169838e-06,
+      "loss": 2.5028,
+      "step": 2099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.003143670391403e-06,
+      "loss": 2.646,
+      "step": 2100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.001940434191759e-06,
+      "loss": 2.7566,
+      "step": 2101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.0007365527649e-06,
+      "loss": 2.7202,
+      "step": 2102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.999532026304928e-06,
+      "loss": 2.5645,
+      "step": 2103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.998326855006047e-06,
+      "loss": 2.5185,
+      "step": 2104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.997121039062564e-06,
+      "loss": 2.7027,
+      "step": 2105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.995914578668892e-06,
+      "loss": 2.5019,
+      "step": 2106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.994707474019547e-06,
+      "loss": 2.694,
+      "step": 2107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.993499725309148e-06,
+      "loss": 2.4672,
+      "step": 2108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.99229133273242e-06,
+      "loss": 2.7499,
+      "step": 2109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.991082296484188e-06,
+      "loss": 2.5739,
+      "step": 2110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.989872616759387e-06,
+      "loss": 2.5765,
+      "step": 2111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.988662293753047e-06,
+      "loss": 2.5603,
+      "step": 2112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.98745132766031e-06,
+      "loss": 2.4647,
+      "step": 2113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.98623971867642e-06,
+      "loss": 2.5239,
+      "step": 2114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.985027466996718e-06,
+      "loss": 2.5029,
+      "step": 2115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.983814572816656e-06,
+      "loss": 2.5892,
+      "step": 2116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.982601036331788e-06,
+      "loss": 2.5345,
+      "step": 2117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.981386857737773e-06,
+      "loss": 2.4747,
+      "step": 2118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.980172037230365e-06,
+      "loss": 2.6332,
+      "step": 2119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.978956575005432e-06,
+      "loss": 2.4839,
+      "step": 2120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.977740471258943e-06,
+      "loss": 2.5226,
+      "step": 2121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.976523726186966e-06,
+      "loss": 2.5374,
+      "step": 2122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.975306339985674e-06,
+      "loss": 2.5473,
+      "step": 2123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.974088312851346e-06,
+      "loss": 2.6036,
+      "step": 2124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.972869644980363e-06,
+      "loss": 2.641,
+      "step": 2125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.971650336569208e-06,
+      "loss": 2.4896,
+      "step": 2126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.97043038781447e-06,
+      "loss": 2.7799,
+      "step": 2127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.96920979891284e-06,
+      "loss": 2.481,
+      "step": 2128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.96798857006111e-06,
+      "loss": 2.5405,
+      "step": 2129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.966766701456177e-06,
+      "loss": 2.7012,
+      "step": 2130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.96554419329504e-06,
+      "loss": 2.6387,
+      "step": 2131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.964321045774808e-06,
+      "loss": 2.5018,
+      "step": 2132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.963097259092682e-06,
+      "loss": 2.6597,
+      "step": 2133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.961872833445971e-06,
+      "loss": 2.6073,
+      "step": 2134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.960647769032091e-06,
+      "loss": 2.7935,
+      "step": 2135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.959422066048556e-06,
+      "loss": 2.6213,
+      "step": 2136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.958195724692982e-06,
+      "loss": 2.5845,
+      "step": 2137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.956968745163093e-06,
+      "loss": 2.5792,
+      "step": 2138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.955741127656713e-06,
+      "loss": 2.515,
+      "step": 2139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.954512872371768e-06,
+      "loss": 2.4261,
+      "step": 2140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.95328397950629e-06,
+      "loss": 2.4368,
+      "step": 2141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.952054449258409e-06,
+      "loss": 2.7058,
+      "step": 2142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.95082428182636e-06,
+      "loss": 2.3984,
+      "step": 2143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.949593477408485e-06,
+      "loss": 2.5358,
+      "step": 2144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.948362036203222e-06,
+      "loss": 2.5752,
+      "step": 2145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.947129958409113e-06,
+      "loss": 2.5008,
+      "step": 2146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.945897244224808e-06,
+      "loss": 2.5538,
+      "step": 2147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.944663893849053e-06,
+      "loss": 2.5601,
+      "step": 2148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.9434299074807e-06,
+      "loss": 2.7238,
+      "step": 2149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.942195285318705e-06,
+      "loss": 2.5887,
+      "step": 2150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.940960027562122e-06,
+      "loss": 2.7081,
+      "step": 2151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.939724134410111e-06,
+      "loss": 2.5943,
+      "step": 2152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.93848760606193e-06,
+      "loss": 2.6945,
+      "step": 2153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.937250442716947e-06,
+      "loss": 2.5849,
+      "step": 2154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.936012644574627e-06,
+      "loss": 2.634,
+      "step": 2155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.934774211834538e-06,
+      "loss": 2.6516,
+      "step": 2156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.933535144696352e-06,
+      "loss": 2.5767,
+      "step": 2157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.932295443359837e-06,
+      "loss": 2.6149,
+      "step": 2158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.931055108024875e-06,
+      "loss": 2.4116,
+      "step": 2159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.929814138891442e-06,
+      "loss": 2.553,
+      "step": 2160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.928572536159614e-06,
+      "loss": 2.3409,
+      "step": 2161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.927330300029575e-06,
+      "loss": 2.6586,
+      "step": 2162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.92608743070161e-06,
+      "loss": 2.5511,
+      "step": 2163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.924843928376105e-06,
+      "loss": 2.475,
+      "step": 2164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.923599793253547e-06,
+      "loss": 2.6248,
+      "step": 2165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.922355025534526e-06,
+      "loss": 2.5138,
+      "step": 2166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.921109625419734e-06,
+      "loss": 2.7369,
+      "step": 2167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.919863593109967e-06,
+      "loss": 2.6541,
+      "step": 2168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.91861692880612e-06,
+      "loss": 2.8143,
+      "step": 2169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.917369632709189e-06,
+      "loss": 2.6361,
+      "step": 2170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.916121705020277e-06,
+      "loss": 2.5115,
+      "step": 2171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.914873145940585e-06,
+      "loss": 2.6785,
+      "step": 2172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.913623955671413e-06,
+      "loss": 2.4036,
+      "step": 2173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.912374134414171e-06,
+      "loss": 2.4772,
+      "step": 2174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.911123682370364e-06,
+      "loss": 2.7584,
+      "step": 2175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.909872599741598e-06,
+      "loss": 2.6212,
+      "step": 2176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.908620886729586e-06,
+      "loss": 2.6147,
+      "step": 2177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.907368543536141e-06,
+      "loss": 2.6618,
+      "step": 2178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.906115570363174e-06,
+      "loss": 2.7665,
+      "step": 2179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.904861967412702e-06,
+      "loss": 2.5554,
+      "step": 2180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.903607734886842e-06,
+      "loss": 2.4749,
+      "step": 2181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.902352872987809e-06,
+      "loss": 2.7009,
+      "step": 2182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.901097381917925e-06,
+      "loss": 2.4677,
+      "step": 2183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.899841261879609e-06,
+      "loss": 2.7227,
+      "step": 2184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.898584513075387e-06,
+      "loss": 2.4948,
+      "step": 2185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.897327135707881e-06,
+      "loss": 2.5652,
+      "step": 2186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.896069129979814e-06,
+      "loss": 2.4784,
+      "step": 2187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.894810496094016e-06,
+      "loss": 2.5692,
+      "step": 2188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.893551234253414e-06,
+      "loss": 2.4778,
+      "step": 2189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.892291344661035e-06,
+      "loss": 2.6541,
+      "step": 2190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.89103082752001e-06,
+      "loss": 2.5475,
+      "step": 2191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.889769683033573e-06,
+      "loss": 2.6381,
+      "step": 2192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.888507911405055e-06,
+      "loss": 2.6299,
+      "step": 2193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.887245512837887e-06,
+      "loss": 2.5083,
+      "step": 2194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.885982487535607e-06,
+      "loss": 2.7068,
+      "step": 2195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.88471883570185e-06,
+      "loss": 2.648,
+      "step": 2196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.88345455754035e-06,
+      "loss": 2.6323,
+      "step": 2197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.882189653254951e-06,
+      "loss": 2.6511,
+      "step": 2198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.880924123049586e-06,
+      "loss": 2.578,
+      "step": 2199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.879657967128297e-06,
+      "loss": 2.5624,
+      "step": 2200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.878391185695223e-06,
+      "loss": 2.5314,
+      "step": 2201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.877123778954608e-06,
+      "loss": 2.5884,
+      "step": 2202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.875855747110792e-06,
+      "loss": 2.7062,
+      "step": 2203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.874587090368221e-06,
+      "loss": 2.7647,
+      "step": 2204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.873317808931433e-06,
+      "loss": 2.663,
+      "step": 2205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.872047903005079e-06,
+      "loss": 2.4433,
+      "step": 2206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.8707773727939e-06,
+      "loss": 2.6787,
+      "step": 2207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.869506218502742e-06,
+      "loss": 2.6936,
+      "step": 2208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.868234440336555e-06,
+      "loss": 2.8175,
+      "step": 2209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.86696203850038e-06,
+      "loss": 2.7346,
+      "step": 2210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.865689013199369e-06,
+      "loss": 2.5446,
+      "step": 2211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.86441536463877e-06,
+      "loss": 2.546,
+      "step": 2212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.863141093023928e-06,
+      "loss": 2.4981,
+      "step": 2213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.861866198560297e-06,
+      "loss": 2.8057,
+      "step": 2214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.860590681453424e-06,
+      "loss": 2.7094,
+      "step": 2215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.859314541908957e-06,
+      "loss": 2.3175,
+      "step": 2216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.858037780132649e-06,
+      "loss": 2.4956,
+      "step": 2217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.85676039633035e-06,
+      "loss": 2.6197,
+      "step": 2218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.85548239070801e-06,
+      "loss": 2.6171,
+      "step": 2219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.85420376347168e-06,
+      "loss": 2.7136,
+      "step": 2220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.852924514827512e-06,
+      "loss": 2.509,
+      "step": 2221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.851644644981756e-06,
+      "loss": 2.5588,
+      "step": 2222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.850364154140764e-06,
+      "loss": 2.458,
+      "step": 2223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.849083042510991e-06,
+      "loss": 2.4914,
+      "step": 2224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.847801310298985e-06,
+      "loss": 2.6884,
+      "step": 2225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.846518957711398e-06,
+      "loss": 2.5802,
+      "step": 2226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.845235984954985e-06,
+      "loss": 2.596,
+      "step": 2227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.843952392236595e-06,
+      "loss": 2.6337,
+      "step": 2228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.84266817976318e-06,
+      "loss": 2.6232,
+      "step": 2229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.841383347741794e-06,
+      "loss": 2.5654,
+      "step": 2230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.840097896379585e-06,
+      "loss": 2.6835,
+      "step": 2231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.838811825883808e-06,
+      "loss": 2.602,
+      "step": 2232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.837525136461814e-06,
+      "loss": 2.6585,
+      "step": 2233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.83623782832105e-06,
+      "loss": 2.5441,
+      "step": 2234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.834949901669073e-06,
+      "loss": 2.5665,
+      "step": 2235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.833661356713528e-06,
+      "loss": 2.5135,
+      "step": 2236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.83237219366217e-06,
+      "loss": 2.5457,
+      "step": 2237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.831082412722845e-06,
+      "loss": 2.4427,
+      "step": 2238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.829792014103505e-06,
+      "loss": 2.551,
+      "step": 2239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.8285009980122e-06,
+      "loss": 2.6009,
+      "step": 2240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.827209364657074e-06,
+      "loss": 2.6982,
+      "step": 2241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.825917114246382e-06,
+      "loss": 2.5978,
+      "step": 2242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.824624246988467e-06,
+      "loss": 2.5223,
+      "step": 2243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.823330763091775e-06,
+      "loss": 2.5859,
+      "step": 2244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.822036662764856e-06,
+      "loss": 2.7149,
+      "step": 2245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.820741946216357e-06,
+      "loss": 2.6577,
+      "step": 2246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.819446613655021e-06,
+      "loss": 2.6481,
+      "step": 2247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.81815066528969e-06,
+      "loss": 2.7125,
+      "step": 2248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.816854101329313e-06,
+      "loss": 2.6123,
+      "step": 2249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.81555692198293e-06,
+      "loss": 2.5834,
+      "step": 2250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.814259127459686e-06,
+      "loss": 2.5848,
+      "step": 2251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.81296071796882e-06,
+      "loss": 2.531,
+      "step": 2252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.811661693719673e-06,
+      "loss": 2.5899,
+      "step": 2253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.810362054921685e-06,
+      "loss": 2.5189,
+      "step": 2254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.809061801784395e-06,
+      "loss": 2.4461,
+      "step": 2255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.807760934517444e-06,
+      "loss": 2.7421,
+      "step": 2256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.806459453330565e-06,
+      "loss": 2.7232,
+      "step": 2257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.805157358433595e-06,
+      "loss": 2.5685,
+      "step": 2258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.80385465003647e-06,
+      "loss": 2.5733,
+      "step": 2259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.802551328349222e-06,
+      "loss": 2.6334,
+      "step": 2260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.801247393581987e-06,
+      "loss": 2.6151,
+      "step": 2261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.799942845944993e-06,
+      "loss": 2.6007,
+      "step": 2262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.79863768564857e-06,
+      "loss": 2.6084,
+      "step": 2263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.797331912903152e-06,
+      "loss": 2.5446,
+      "step": 2264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.796025527919263e-06,
+      "loss": 2.718,
+      "step": 2265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.794718530907529e-06,
+      "loss": 2.5722,
+      "step": 2266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.793410922078677e-06,
+      "loss": 2.5749,
+      "step": 2267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.792102701643532e-06,
+      "loss": 2.386,
+      "step": 2268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.790793869813014e-06,
+      "loss": 2.7055,
+      "step": 2269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.789484426798148e-06,
+      "loss": 2.6045,
+      "step": 2270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.788174372810048e-06,
+      "loss": 2.5907,
+      "step": 2271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.786863708059936e-06,
+      "loss": 2.5085,
+      "step": 2272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.785552432759127e-06,
+      "loss": 2.6217,
+      "step": 2273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.784240547119039e-06,
+      "loss": 2.746,
+      "step": 2274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.782928051351183e-06,
+      "loss": 2.4063,
+      "step": 2275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.78161494566717e-06,
+      "loss": 2.662,
+      "step": 2276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.780301230278711e-06,
+      "loss": 2.6546,
+      "step": 2277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.778986905397615e-06,
+      "loss": 2.5244,
+      "step": 2278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.77767197123579e-06,
+      "loss": 2.5505,
+      "step": 2279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.776356428005237e-06,
+      "loss": 2.7441,
+      "step": 2280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.775040275918063e-06,
+      "loss": 2.69,
+      "step": 2281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.77372351518647e-06,
+      "loss": 2.5176,
+      "step": 2282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.772406146022751e-06,
+      "loss": 2.5723,
+      "step": 2283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.771088168639312e-06,
+      "loss": 2.5704,
+      "step": 2284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.769769583248645e-06,
+      "loss": 2.7373,
+      "step": 2285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.768450390063342e-06,
+      "loss": 2.5102,
+      "step": 2286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.767130589296095e-06,
+      "loss": 2.4319,
+      "step": 2287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.765810181159696e-06,
+      "loss": 2.6602,
+      "step": 2288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.764489165867029e-06,
+      "loss": 2.5159,
+      "step": 2289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.763167543631084e-06,
+      "loss": 2.6135,
+      "step": 2290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.761845314664942e-06,
+      "loss": 2.7723,
+      "step": 2291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.760522479181784e-06,
+      "loss": 2.6285,
+      "step": 2292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.759199037394888e-06,
+      "loss": 2.5177,
+      "step": 2293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.757874989517631e-06,
+      "loss": 2.6341,
+      "step": 2294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.75655033576349e-06,
+      "loss": 2.5947,
+      "step": 2295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.755225076346035e-06,
+      "loss": 2.5884,
+      "step": 2296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.753899211478933e-06,
+      "loss": 2.5412,
+      "step": 2297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.752572741375956e-06,
+      "loss": 2.6923,
+      "step": 2298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.751245666250968e-06,
+      "loss": 2.6304,
+      "step": 2299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.74991798631793e-06,
+      "loss": 2.4996,
+      "step": 2300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.7485897017909e-06,
+      "loss": 2.6367,
+      "step": 2301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.74726081288404e-06,
+      "loss": 2.703,
+      "step": 2302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.745931319811603e-06,
+      "loss": 2.4673,
+      "step": 2303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.744601222787938e-06,
+      "loss": 2.5757,
+      "step": 2304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.7432705220275e-06,
+      "loss": 2.6648,
+      "step": 2305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.741939217744835e-06,
+      "loss": 2.6826,
+      "step": 2306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.740607310154583e-06,
+      "loss": 2.7921,
+      "step": 2307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.739274799471492e-06,
+      "loss": 2.6572,
+      "step": 2308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.737941685910395e-06,
+      "loss": 2.2318,
+      "step": 2309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.736607969686231e-06,
+      "loss": 2.6085,
+      "step": 2310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.735273651014031e-06,
+      "loss": 2.6578,
+      "step": 2311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.733938730108928e-06,
+      "loss": 2.6034,
+      "step": 2312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.73260320718615e-06,
+      "loss": 2.6364,
+      "step": 2313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.731267082461018e-06,
+      "loss": 2.6456,
+      "step": 2314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.729930356148956e-06,
+      "loss": 2.5852,
+      "step": 2315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.728593028465481e-06,
+      "loss": 2.6869,
+      "step": 2316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.72725509962621e-06,
+      "loss": 2.5301,
+      "step": 2317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.725916569846854e-06,
+      "loss": 2.5863,
+      "step": 2318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.724577439343223e-06,
+      "loss": 2.5897,
+      "step": 2319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.723237708331225e-06,
+      "loss": 2.6298,
+      "step": 2320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.721897377026859e-06,
+      "loss": 2.7035,
+      "step": 2321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.720556445646226e-06,
+      "loss": 2.4774,
+      "step": 2322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.719214914405525e-06,
+      "loss": 2.4448,
+      "step": 2323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.717872783521048e-06,
+      "loss": 2.4175,
+      "step": 2324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.716530053209182e-06,
+      "loss": 2.4623,
+      "step": 2325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.715186723686418e-06,
+      "loss": 2.6693,
+      "step": 2326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.713842795169334e-06,
+      "loss": 2.7202,
+      "step": 2327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.712498267874615e-06,
+      "loss": 2.7234,
+      "step": 2328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.711153142019035e-06,
+      "loss": 2.4721,
+      "step": 2329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.709807417819464e-06,
+      "loss": 2.6127,
+      "step": 2330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.708461095492876e-06,
+      "loss": 2.4876,
+      "step": 2331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.707114175256335e-06,
+      "loss": 2.6058,
+      "step": 2332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.705766657326999e-06,
+      "loss": 2.3585,
+      "step": 2333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.704418541922132e-06,
+      "loss": 2.5525,
+      "step": 2334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.703069829259087e-06,
+      "loss": 2.4915,
+      "step": 2335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.701720519555315e-06,
+      "loss": 2.5493,
+      "step": 2336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.700370613028364e-06,
+      "loss": 2.6009,
+      "step": 2337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.699020109895875e-06,
+      "loss": 2.5307,
+      "step": 2338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.69766901037559e-06,
+      "loss": 2.3723,
+      "step": 2339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.696317314685342e-06,
+      "loss": 2.5767,
+      "step": 2340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.694965023043067e-06,
+      "loss": 2.5408,
+      "step": 2341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.69361213566679e-06,
+      "loss": 2.5031,
+      "step": 2342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.692258652774636e-06,
+      "loss": 2.4459,
+      "step": 2343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.690904574584828e-06,
+      "loss": 2.5701,
+      "step": 2344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.689549901315677e-06,
+      "loss": 2.3174,
+      "step": 2345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.6881946331856e-06,
+      "loss": 2.6292,
+      "step": 2346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.6868387704131e-06,
+      "loss": 2.5507,
+      "step": 2347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.685482313216784e-06,
+      "loss": 2.6359,
+      "step": 2348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.684125261815352e-06,
+      "loss": 2.5649,
+      "step": 2349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.682767616427598e-06,
+      "loss": 2.5765,
+      "step": 2350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.681409377272414e-06,
+      "loss": 2.7976,
+      "step": 2351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.680050544568788e-06,
+      "loss": 2.8242,
+      "step": 2352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.6786911185358e-06,
+      "loss": 2.5279,
+      "step": 2353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.67733109939263e-06,
+      "loss": 2.7454,
+      "step": 2354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.675970487358552e-06,
+      "loss": 2.7214,
+      "step": 2355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.674609282652936e-06,
+      "loss": 2.4745,
+      "step": 2356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.673247485495247e-06,
+      "loss": 2.7342,
+      "step": 2357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.671885096105042e-06,
+      "loss": 2.4902,
+      "step": 2358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.670522114701983e-06,
+      "loss": 2.7203,
+      "step": 2359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.669158541505817e-06,
+      "loss": 2.5739,
+      "step": 2360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.667794376736395e-06,
+      "loss": 2.5216,
+      "step": 2361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.666429620613656e-06,
+      "loss": 2.7359,
+      "step": 2362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.665064273357638e-06,
+      "loss": 2.547,
+      "step": 2363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.663698335188477e-06,
+      "loss": 2.5363,
+      "step": 2364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.662331806326399e-06,
+      "loss": 2.5336,
+      "step": 2365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.660964686991725e-06,
+      "loss": 2.6045,
+      "step": 2366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.659596977404879e-06,
+      "loss": 2.6891,
+      "step": 2367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.65822867778637e-06,
+      "loss": 2.523,
+      "step": 2368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.656859788356811e-06,
+      "loss": 2.6495,
+      "step": 2369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.655490309336905e-06,
+      "loss": 2.5656,
+      "step": 2370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.65412024094745e-06,
+      "loss": 2.5992,
+      "step": 2371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.65274958340934e-06,
+      "loss": 2.699,
+      "step": 2372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.651378336943565e-06,
+      "loss": 2.5365,
+      "step": 2373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.65000650177121e-06,
+      "loss": 2.5822,
+      "step": 2374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.648634078113453e-06,
+      "loss": 2.5401,
+      "step": 2375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.647261066191569e-06,
+      "loss": 2.5774,
+      "step": 2376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.645887466226923e-06,
+      "loss": 2.6625,
+      "step": 2377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.644513278440984e-06,
+      "loss": 2.5682,
+      "step": 2378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.643138503055306e-06,
+      "loss": 2.6275,
+      "step": 2379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.641763140291546e-06,
+      "loss": 2.5354,
+      "step": 2380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.640387190371448e-06,
+      "loss": 2.3807,
+      "step": 2381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.639010653516858e-06,
+      "loss": 2.6182,
+      "step": 2382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.63763352994971e-06,
+      "loss": 2.5359,
+      "step": 2383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.636255819892037e-06,
+      "loss": 2.6359,
+      "step": 2384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.634877523565964e-06,
+      "loss": 2.6336,
+      "step": 2385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.633498641193713e-06,
+      "loss": 2.6452,
+      "step": 2386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.632119172997601e-06,
+      "loss": 2.4194,
+      "step": 2387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.630739119200035e-06,
+      "loss": 2.5368,
+      "step": 2388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.629358480023521e-06,
+      "loss": 2.4869,
+      "step": 2389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.627977255690658e-06,
+      "loss": 2.4915,
+      "step": 2390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.626595446424133e-06,
+      "loss": 2.782,
+      "step": 2391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.62521305244674e-06,
+      "loss": 2.5767,
+      "step": 2392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.623830073981361e-06,
+      "loss": 2.446,
+      "step": 2393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.622446511250964e-06,
+      "loss": 2.4755,
+      "step": 2394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.621062364478627e-06,
+      "loss": 2.5466,
+      "step": 2395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.61967763388751e-06,
+      "loss": 2.4267,
+      "step": 2396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.61829231970087e-06,
+      "loss": 2.5937,
+      "step": 2397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.616906422142061e-06,
+      "loss": 2.5518,
+      "step": 2398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.61551994143453e-06,
+      "loss": 2.5754,
+      "step": 2399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.614132877801817e-06,
+      "loss": 2.7716,
+      "step": 2400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.612745231467556e-06,
+      "loss": 2.5672,
+      "step": 2401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.611357002655475e-06,
+      "loss": 2.6149,
+      "step": 2402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.609968191589396e-06,
+      "loss": 2.6132,
+      "step": 2403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.608578798493237e-06,
+      "loss": 2.6278,
+      "step": 2404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.607188823591004e-06,
+      "loss": 2.6757,
+      "step": 2405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.605798267106805e-06,
+      "loss": 2.7148,
+      "step": 2406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.604407129264835e-06,
+      "loss": 2.7476,
+      "step": 2407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.603015410289387e-06,
+      "loss": 2.5226,
+      "step": 2408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.601623110404841e-06,
+      "loss": 2.5611,
+      "step": 2409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.600230229835682e-06,
+      "loss": 2.6794,
+      "step": 2410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.598836768806478e-06,
+      "loss": 2.6658,
+      "step": 2411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.597442727541898e-06,
+      "loss": 2.6075,
+      "step": 2412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.596048106266697e-06,
+      "loss": 2.567,
+      "step": 2413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.594652905205731e-06,
+      "loss": 2.5864,
+      "step": 2414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.593257124583946e-06,
+      "loss": 2.5504,
+      "step": 2415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.591860764626381e-06,
+      "loss": 2.5646,
+      "step": 2416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.590463825558169e-06,
+      "loss": 2.6392,
+      "step": 2417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.589066307604537e-06,
+      "loss": 2.758,
+      "step": 2418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.587668210990805e-06,
+      "loss": 2.4847,
+      "step": 2419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.586269535942386e-06,
+      "loss": 2.5166,
+      "step": 2420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.584870282684786e-06,
+      "loss": 2.6342,
+      "step": 2421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.583470451443605e-06,
+      "loss": 2.6081,
+      "step": 2422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.582070042444535e-06,
+      "loss": 2.5615,
+      "step": 2423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.580669055913364e-06,
+      "loss": 2.5954,
+      "step": 2424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.579267492075968e-06,
+      "loss": 2.4868,
+      "step": 2425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.577865351158324e-06,
+      "loss": 2.4834,
+      "step": 2426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.57646263338649e-06,
+      "loss": 2.5618,
+      "step": 2427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.575059338986632e-06,
+      "loss": 2.4996,
+      "step": 2428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.573655468184997e-06,
+      "loss": 2.5122,
+      "step": 2429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.57225102120793e-06,
+      "loss": 2.4806,
+      "step": 2430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.570845998281865e-06,
+      "loss": 2.6045,
+      "step": 2431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.569440399633337e-06,
+      "loss": 2.4304,
+      "step": 2432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.568034225488964e-06,
+      "loss": 2.4538,
+      "step": 2433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.566627476075467e-06,
+      "loss": 2.64,
+      "step": 2434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.565220151619648e-06,
+      "loss": 2.6859,
+      "step": 2435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.563812252348412e-06,
+      "loss": 2.7395,
+      "step": 2436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.56240377848875e-06,
+      "loss": 2.5786,
+      "step": 2437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.560994730267751e-06,
+      "loss": 2.4372,
+      "step": 2438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.559585107912593e-06,
+      "loss": 2.7312,
+      "step": 2439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.558174911650546e-06,
+      "loss": 2.54,
+      "step": 2440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.556764141708975e-06,
+      "loss": 2.3839,
+      "step": 2441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.555352798315339e-06,
+      "loss": 2.3791,
+      "step": 2442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.553940881697182e-06,
+      "loss": 2.6749,
+      "step": 2443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.552528392082147e-06,
+      "loss": 2.7619,
+      "step": 2444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.55111532969797e-06,
+      "loss": 2.4611,
+      "step": 2445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.549701694772475e-06,
+      "loss": 2.5515,
+      "step": 2446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.548287487533579e-06,
+      "loss": 2.4892,
+      "step": 2447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.546872708209297e-06,
+      "loss": 2.4429,
+      "step": 2448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.545457357027727e-06,
+      "loss": 2.6093,
+      "step": 2449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.544041434217067e-06,
+      "loss": 2.7221,
+      "step": 2450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.542624940005604e-06,
+      "loss": 2.6011,
+      "step": 2451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.541207874621718e-06,
+      "loss": 2.5159,
+      "step": 2452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.539790238293878e-06,
+      "loss": 2.514,
+      "step": 2453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.53837203125065e-06,
+      "loss": 2.4945,
+      "step": 2454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.536953253720687e-06,
+      "loss": 2.4896,
+      "step": 2455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 2.4572,
+      "step": 2456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.534113988115645e-06,
+      "loss": 2.7591,
+      "step": 2457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.532693500498334e-06,
+      "loss": 2.7264,
+      "step": 2458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.53127244330983e-06,
+      "loss": 2.6027,
+      "step": 2459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.529850816779252e-06,
+      "loss": 2.4524,
+      "step": 2460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.528428621135802e-06,
+      "loss": 2.6452,
+      "step": 2461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.52700585660878e-06,
+      "loss": 2.6584,
+      "step": 2462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.525582523427578e-06,
+      "loss": 2.6005,
+      "step": 2463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.524158621821678e-06,
+      "loss": 2.5459,
+      "step": 2464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.52273415202065e-06,
+      "loss": 2.6272,
+      "step": 2465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.521309114254164e-06,
+      "loss": 2.4999,
+      "step": 2466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.519883508751972e-06,
+      "loss": 2.6229,
+      "step": 2467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.518457335743927e-06,
+      "loss": 2.5676,
+      "step": 2468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.517030595459966e-06,
+      "loss": 2.4274,
+      "step": 2469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.515603288130119e-06,
+      "loss": 2.4998,
+      "step": 2470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.514175413984513e-06,
+      "loss": 2.6171,
+      "step": 2471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.512746973253358e-06,
+      "loss": 2.6117,
+      "step": 2472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.511317966166963e-06,
+      "loss": 2.6337,
+      "step": 2473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.509888392955725e-06,
+      "loss": 2.5162,
+      "step": 2474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.508458253850126e-06,
+      "loss": 2.6302,
+      "step": 2475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.507027549080753e-06,
+      "loss": 2.5069,
+      "step": 2476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.505596278878272e-06,
+      "loss": 2.5346,
+      "step": 2477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.504164443473446e-06,
+      "loss": 2.6742,
+      "step": 2478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.502732043097128e-06,
+      "loss": 2.468,
+      "step": 2479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.50129907798026e-06,
+      "loss": 2.4726,
+      "step": 2480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.499865548353879e-06,
+      "loss": 2.6581,
+      "step": 2481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.49843145444911e-06,
+      "loss": 2.5507,
+      "step": 2482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.496996796497172e-06,
+      "loss": 2.5632,
+      "step": 2483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.49556157472937e-06,
+      "loss": 2.6134,
+      "step": 2484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.494125789377104e-06,
+      "loss": 2.7207,
+      "step": 2485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.492689440671865e-06,
+      "loss": 2.6244,
+      "step": 2486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.491252528845229e-06,
+      "loss": 2.5673,
+      "step": 2487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.489815054128874e-06,
+      "loss": 2.6189,
+      "step": 2488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.488377016754556e-06,
+      "loss": 2.5662,
+      "step": 2489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.48693841695413e-06,
+      "loss": 2.5665,
+      "step": 2490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.485499254959542e-06,
+      "loss": 2.4691,
+      "step": 2491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.484059531002822e-06,
+      "loss": 2.4533,
+      "step": 2492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.482619245316094e-06,
+      "loss": 2.6268,
+      "step": 2493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.481178398131578e-06,
+      "loss": 2.6128,
+      "step": 2494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.47973698968158e-06,
+      "loss": 2.6875,
+      "step": 2495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.478295020198488e-06,
+      "loss": 2.5892,
+      "step": 2496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.476852489914797e-06,
+      "loss": 2.6404,
+      "step": 2497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.475409399063081e-06,
+      "loss": 2.629,
+      "step": 2498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.473965747876007e-06,
+      "loss": 2.6444,
+      "step": 2499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.472521536586336e-06,
+      "loss": 2.362,
+      "step": 2500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.471076765426913e-06,
+      "loss": 2.6292,
+      "step": 2501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.469631434630677e-06,
+      "loss": 2.6224,
+      "step": 2502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.468185544430656e-06,
+      "loss": 2.6199,
+      "step": 2503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.46673909505997e-06,
+      "loss": 2.6567,
+      "step": 2504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.465292086751829e-06,
+      "loss": 2.5208,
+      "step": 2505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.46384451973953e-06,
+      "loss": 2.571,
+      "step": 2506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.462396394256462e-06,
+      "loss": 2.7089,
+      "step": 2507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.460947710536108e-06,
+      "loss": 2.5557,
+      "step": 2508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.459498468812032e-06,
+      "loss": 2.6597,
+      "step": 2509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.458048669317897e-06,
+      "loss": 2.6636,
+      "step": 2510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.45659831228745e-06,
+      "loss": 2.5195,
+      "step": 2511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.45514739795453e-06,
+      "loss": 2.7085,
+      "step": 2512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.45369592655307e-06,
+      "loss": 2.3491,
+      "step": 2513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.452243898317083e-06,
+      "loss": 2.5634,
+      "step": 2514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.450791313480679e-06,
+      "loss": 2.6207,
+      "step": 2515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.44933817227806e-06,
+      "loss": 2.6451,
+      "step": 2516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.447884474943509e-06,
+      "loss": 2.7452,
+      "step": 2517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.446430221711406e-06,
+      "loss": 2.5703,
+      "step": 2518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.444975412816216e-06,
+      "loss": 2.5749,
+      "step": 2519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.4435200484925e-06,
+      "loss": 2.3245,
+      "step": 2520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.442064128974901e-06,
+      "loss": 2.7889,
+      "step": 2521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.440607654498156e-06,
+      "loss": 2.5646,
+      "step": 2522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.43915062529709e-06,
+      "loss": 2.4603,
+      "step": 2523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.437693041606619e-06,
+      "loss": 2.5879,
+      "step": 2524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.436234903661744e-06,
+      "loss": 2.655,
+      "step": 2525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.434776211697559e-06,
+      "loss": 2.4387,
+      "step": 2526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.433316965949253e-06,
+      "loss": 2.5422,
+      "step": 2527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.43185716665209e-06,
+      "loss": 2.655,
+      "step": 2528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.430396814041436e-06,
+      "loss": 2.5647,
+      "step": 2529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.42893590835274e-06,
+      "loss": 2.4441,
+      "step": 2530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.427474449821543e-06,
+      "loss": 2.7744,
+      "step": 2531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.426012438683472e-06,
+      "loss": 2.5379,
+      "step": 2532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.424549875174248e-06,
+      "loss": 2.4646,
+      "step": 2533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.423086759529676e-06,
+      "loss": 2.772,
+      "step": 2534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.42162309198565e-06,
+      "loss": 2.6128,
+      "step": 2535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.42015887277816e-06,
+      "loss": 2.62,
+      "step": 2536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.418694102143274e-06,
+      "loss": 2.6508,
+      "step": 2537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.417228780317159e-06,
+      "loss": 2.7049,
+      "step": 2538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.41576290753607e-06,
+      "loss": 2.634,
+      "step": 2539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.41429648403634e-06,
+      "loss": 2.5676,
+      "step": 2540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.4128295100544e-06,
+      "loss": 2.625,
+      "step": 2541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.411361985826775e-06,
+      "loss": 2.4087,
+      "step": 2542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.409893911590063e-06,
+      "loss": 2.5044,
+      "step": 2543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.408425287580966e-06,
+      "loss": 2.5427,
+      "step": 2544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.406956114036265e-06,
+      "loss": 2.6852,
+      "step": 2545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.405486391192834e-06,
+      "loss": 2.4742,
+      "step": 2546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.404016119287633e-06,
+      "loss": 2.6787,
+      "step": 2547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.402545298557712e-06,
+      "loss": 2.6733,
+      "step": 2548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.401073929240211e-06,
+      "loss": 2.5572,
+      "step": 2549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.399602011572356e-06,
+      "loss": 2.4537,
+      "step": 2550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.398129545791463e-06,
+      "loss": 2.6312,
+      "step": 2551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.396656532134934e-06,
+      "loss": 2.5922,
+      "step": 2552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.395182970840261e-06,
+      "loss": 2.5495,
+      "step": 2553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.393708862145024e-06,
+      "loss": 2.4656,
+      "step": 2554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.392234206286894e-06,
+      "loss": 2.5872,
+      "step": 2555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.390759003503624e-06,
+      "loss": 2.5661,
+      "step": 2556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.389283254033061e-06,
+      "loss": 2.5764,
+      "step": 2557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.387806958113139e-06,
+      "loss": 2.648,
+      "step": 2558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.386330115981877e-06,
+      "loss": 2.5034,
+      "step": 2559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.384852727877386e-06,
+      "loss": 2.5501,
+      "step": 2560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.383374794037862e-06,
+      "loss": 2.4438,
+      "step": 2561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.381896314701589e-06,
+      "loss": 2.6433,
+      "step": 2562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.380417290106943e-06,
+      "loss": 2.5714,
+      "step": 2563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.378937720492384e-06,
+      "loss": 2.4141,
+      "step": 2564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.37745760609646e-06,
+      "loss": 2.5779,
+      "step": 2565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.375976947157808e-06,
+      "loss": 2.604,
+      "step": 2566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.374495743915153e-06,
+      "loss": 2.5995,
+      "step": 2567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.373013996607309e-06,
+      "loss": 2.7605,
+      "step": 2568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.371531705473172e-06,
+      "loss": 2.6018,
+      "step": 2569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.370048870751732e-06,
+      "loss": 2.4328,
+      "step": 2570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.368565492682064e-06,
+      "loss": 2.5006,
+      "step": 2571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.367081571503332e-06,
+      "loss": 2.4701,
+      "step": 2572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.365597107454786e-06,
+      "loss": 2.5793,
+      "step": 2573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.364112100775765e-06,
+      "loss": 2.6341,
+      "step": 2574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.36262655170569e-06,
+      "loss": 2.5406,
+      "step": 2575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.361140460484081e-06,
+      "loss": 2.5491,
+      "step": 2576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.359653827350533e-06,
+      "loss": 2.3944,
+      "step": 2577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.358166652544734e-06,
+      "loss": 2.4618,
+      "step": 2578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.356678936306463e-06,
+      "loss": 2.773,
+      "step": 2579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.355190678875577e-06,
+      "loss": 2.5812,
+      "step": 2580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.353701880492031e-06,
+      "loss": 2.6375,
+      "step": 2581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.352212541395859e-06,
+      "loss": 2.5355,
+      "step": 2582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.350722661827186e-06,
+      "loss": 2.5235,
+      "step": 2583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.349232242026221e-06,
+      "loss": 2.6211,
+      "step": 2584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.347741282233265e-06,
+      "loss": 2.4973,
+      "step": 2585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.346249782688705e-06,
+      "loss": 2.5139,
+      "step": 2586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.344757743633007e-06,
+      "loss": 2.6732,
+      "step": 2587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.343265165306736e-06,
+      "loss": 2.5139,
+      "step": 2588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.341772047950534e-06,
+      "loss": 2.5775,
+      "step": 2589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.34027839180514e-06,
+      "loss": 2.4471,
+      "step": 2590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.338784197111367e-06,
+      "loss": 2.5801,
+      "step": 2591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.337289464110128e-06,
+      "loss": 2.5727,
+      "step": 2592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.335794193042413e-06,
+      "loss": 2.5963,
+      "step": 2593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.334298384149305e-06,
+      "loss": 2.6013,
+      "step": 2594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.332802037671968e-06,
+      "loss": 2.529,
+      "step": 2595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.331305153851659e-06,
+      "loss": 2.4846,
+      "step": 2596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.329807732929715e-06,
+      "loss": 2.57,
+      "step": 2597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.328309775147567e-06,
+      "loss": 2.6741,
+      "step": 2598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.326811280746727e-06,
+      "loss": 2.7164,
+      "step": 2599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.325312249968795e-06,
+      "loss": 2.4808,
+      "step": 2600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.323812683055457e-06,
+      "loss": 2.564,
+      "step": 2601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.322312580248488e-06,
+      "loss": 2.6342,
+      "step": 2602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.320811941789746e-06,
+      "loss": 2.668,
+      "step": 2603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.319310767921174e-06,
+      "loss": 2.5528,
+      "step": 2604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.317809058884809e-06,
+      "loss": 2.4981,
+      "step": 2605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.316306814922769e-06,
+      "loss": 2.3741,
+      "step": 2606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.314804036277254e-06,
+      "loss": 2.436,
+      "step": 2607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.313300723190561e-06,
+      "loss": 2.5505,
+      "step": 2608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.311796875905061e-06,
+      "loss": 2.635,
+      "step": 2609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.310292494663222e-06,
+      "loss": 2.671,
+      "step": 2610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.308787579707591e-06,
+      "loss": 2.5532,
+      "step": 2611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.307282131280805e-06,
+      "loss": 2.4858,
+      "step": 2612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.305776149625584e-06,
+      "loss": 2.5576,
+      "step": 2613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.304269634984735e-06,
+      "loss": 2.4927,
+      "step": 2614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.30276258760115e-06,
+      "loss": 2.5737,
+      "step": 2615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.301255007717812e-06,
+      "loss": 2.5424,
+      "step": 2616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.299746895577785e-06,
+      "loss": 2.6324,
+      "step": 2617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.298238251424216e-06,
+      "loss": 2.551,
+      "step": 2618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.296729075500345e-06,
+      "loss": 2.5114,
+      "step": 2619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.295219368049494e-06,
+      "loss": 2.5515,
+      "step": 2620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.29370912931507e-06,
+      "loss": 2.7195,
+      "step": 2621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.292198359540566e-06,
+      "loss": 2.5085,
+      "step": 2622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.290687058969564e-06,
+      "loss": 2.5845,
+      "step": 2623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.289175227845727e-06,
+      "loss": 2.6207,
+      "step": 2624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.287662866412807e-06,
+      "loss": 2.5274,
+      "step": 2625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.286149974914637e-06,
+      "loss": 2.3412,
+      "step": 2626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.28463655359514e-06,
+      "loss": 2.6722,
+      "step": 2627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.283122602698324e-06,
+      "loss": 2.5865,
+      "step": 2628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.281608122468278e-06,
+      "loss": 2.5272,
+      "step": 2629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.280093113149182e-06,
+      "loss": 2.4534,
+      "step": 2630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.2785775749853e-06,
+      "loss": 2.6454,
+      "step": 2631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.277061508220975e-06,
+      "loss": 2.6761,
+      "step": 2632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.275544913100646e-06,
+      "loss": 2.7751,
+      "step": 2633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.274027789868828e-06,
+      "loss": 2.6434,
+      "step": 2634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.272510138770128e-06,
+      "loss": 2.5789,
+      "step": 2635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.270991960049231e-06,
+      "loss": 2.5515,
+      "step": 2636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.26947325395091e-06,
+      "loss": 2.723,
+      "step": 2637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.267954020720029e-06,
+      "loss": 2.554,
+      "step": 2638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.266434260601528e-06,
+      "loss": 2.5653,
+      "step": 2639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.264913973840439e-06,
+      "loss": 2.7289,
+      "step": 2640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.26339316068187e-06,
+      "loss": 2.6784,
+      "step": 2641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.261871821371026e-06,
+      "loss": 2.6692,
+      "step": 2642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.260349956153185e-06,
+      "loss": 2.5945,
+      "step": 2643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.258827565273717e-06,
+      "loss": 2.7162,
+      "step": 2644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.25730464897808e-06,
+      "loss": 2.5601,
+      "step": 2645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.255781207511804e-06,
+      "loss": 2.6387,
+      "step": 2646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.254257241120514e-06,
+      "loss": 2.5941,
+      "step": 2647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.252732750049918e-06,
+      "loss": 2.6675,
+      "step": 2648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.251207734545807e-06,
+      "loss": 2.6384,
+      "step": 2649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.249682194854059e-06,
+      "loss": 2.5124,
+      "step": 2650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.248156131220631e-06,
+      "loss": 2.6139,
+      "step": 2651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.24662954389157e-06,
+      "loss": 2.627,
+      "step": 2652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.245102433113005e-06,
+      "loss": 2.6104,
+      "step": 2653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.24357479913115e-06,
+      "loss": 2.7312,
+      "step": 2654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.242046642192304e-06,
+      "loss": 2.553,
+      "step": 2655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.240517962542849e-06,
+      "loss": 2.5021,
+      "step": 2656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.238988760429252e-06,
+      "loss": 2.5739,
+      "step": 2657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.237459036098064e-06,
+      "loss": 2.6064,
+      "step": 2658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.23592878979592e-06,
+      "loss": 2.7111,
+      "step": 2659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.234398021769541e-06,
+      "loss": 2.6277,
+      "step": 2660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.23286673226573e-06,
+      "loss": 2.5882,
+      "step": 2661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.23133492153137e-06,
+      "loss": 2.7023,
+      "step": 2662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.229802589813442e-06,
+      "loss": 2.6135,
+      "step": 2663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.228269737358993e-06,
+      "loss": 2.3393,
+      "step": 2664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.226736364415167e-06,
+      "loss": 2.5233,
+      "step": 2665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.225202471229187e-06,
+      "loss": 2.5155,
+      "step": 2666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.22366805804836e-06,
+      "loss": 2.555,
+      "step": 2667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.222133125120076e-06,
+      "loss": 2.5612,
+      "step": 2668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.220597672691812e-06,
+      "loss": 2.6521,
+      "step": 2669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.219061701011125e-06,
+      "loss": 2.4864,
+      "step": 2670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.217525210325662e-06,
+      "loss": 2.5519,
+      "step": 2671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.215988200883145e-06,
+      "loss": 2.4437,
+      "step": 2672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.214450672931385e-06,
+      "loss": 2.41,
+      "step": 2673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.212912626718273e-06,
+      "loss": 2.6167,
+      "step": 2674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.21137406249179e-06,
+      "loss": 2.642,
+      "step": 2675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.209834980499995e-06,
+      "loss": 2.8119,
+      "step": 2676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.20829538099103e-06,
+      "loss": 2.6948,
+      "step": 2677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.206755264213128e-06,
+      "loss": 2.5394,
+      "step": 2678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.205214630414594e-06,
+      "loss": 2.6643,
+      "step": 2679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.203673479843824e-06,
+      "loss": 2.3948,
+      "step": 2680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.202131812749296e-06,
+      "loss": 2.5228,
+      "step": 2681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.200589629379572e-06,
+      "loss": 2.5046,
+      "step": 2682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.199046929983294e-06,
+      "loss": 2.6575,
+      "step": 2683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.19750371480919e-06,
+      "loss": 2.5759,
+      "step": 2684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.195959984106072e-06,
+      "loss": 2.6241,
+      "step": 2685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.19441573812283e-06,
+      "loss": 2.5886,
+      "step": 2686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.192870977108443e-06,
+      "loss": 2.594,
+      "step": 2687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.191325701311971e-06,
+      "loss": 2.581,
+      "step": 2688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.189779910982556e-06,
+      "loss": 2.6978,
+      "step": 2689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.188233606369422e-06,
+      "loss": 2.5803,
+      "step": 2690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.18668678772188e-06,
+      "loss": 2.642,
+      "step": 2691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.185139455289322e-06,
+      "loss": 2.6413,
+      "step": 2692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.183591609321217e-06,
+      "loss": 2.6384,
+      "step": 2693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.182043250067128e-06,
+      "loss": 2.4781,
+      "step": 2694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.180494377776694e-06,
+      "loss": 2.6023,
+      "step": 2695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.178944992699632e-06,
+      "loss": 2.5401,
+      "step": 2696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.177395095085754e-06,
+      "loss": 2.6168,
+      "step": 2697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.175844685184942e-06,
+      "loss": 2.4881,
+      "step": 2698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.174293763247172e-06,
+      "loss": 2.5601,
+      "step": 2699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.172742329522493e-06,
+      "loss": 2.4295,
+      "step": 2700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.171190384261043e-06,
+      "loss": 2.5413,
+      "step": 2701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.169637927713037e-06,
+      "loss": 2.589,
+      "step": 2702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.16808496012878e-06,
+      "loss": 2.5369,
+      "step": 2703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.166531481758649e-06,
+      "loss": 2.5705,
+      "step": 2704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.164977492853113e-06,
+      "loss": 2.527,
+      "step": 2705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.163422993662718e-06,
+      "loss": 2.6226,
+      "step": 2706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.161867984438094e-06,
+      "loss": 2.7103,
+      "step": 2707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.160312465429952e-06,
+      "loss": 2.6387,
+      "step": 2708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.158756436889088e-06,
+      "loss": 2.4585,
+      "step": 2709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.157199899066379e-06,
+      "loss": 2.6166,
+      "step": 2710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.15564285221278e-06,
+      "loss": 2.6646,
+      "step": 2711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.154085296579335e-06,
+      "loss": 2.5158,
+      "step": 2712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.152527232417163e-06,
+      "loss": 2.6282,
+      "step": 2713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.150968659977472e-06,
+      "loss": 2.5294,
+      "step": 2714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.149409579511545e-06,
+      "loss": 2.5221,
+      "step": 2715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.147849991270753e-06,
+      "loss": 2.4804,
+      "step": 2716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.146289895506546e-06,
+      "loss": 2.5638,
+      "step": 2717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.144729292470454e-06,
+      "loss": 2.5047,
+      "step": 2718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.143168182414094e-06,
+      "loss": 2.5307,
+      "step": 2719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.141606565589157e-06,
+      "loss": 2.4673,
+      "step": 2720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.140044442247425e-06,
+      "loss": 2.7557,
+      "step": 2721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.138481812640755e-06,
+      "loss": 2.3536,
+      "step": 2722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.136918677021088e-06,
+      "loss": 2.5491,
+      "step": 2723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.135355035640445e-06,
+      "loss": 2.4149,
+      "step": 2724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.13379088875093e-06,
+      "loss": 2.4424,
+      "step": 2725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.132226236604728e-06,
+      "loss": 2.5938,
+      "step": 2726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.130661079454108e-06,
+      "loss": 2.5032,
+      "step": 2727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.129095417551416e-06,
+      "loss": 2.5452,
+      "step": 2728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.127529251149081e-06,
+      "loss": 2.6004,
+      "step": 2729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.125962580499616e-06,
+      "loss": 2.3647,
+      "step": 2730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.124395405855609e-06,
+      "loss": 2.6357,
+      "step": 2731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.122827727469737e-06,
+      "loss": 2.6049,
+      "step": 2732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.121259545594756e-06,
+      "loss": 2.545,
+      "step": 2733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.119690860483497e-06,
+      "loss": 2.4327,
+      "step": 2734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.11812167238888e-06,
+      "loss": 2.5694,
+      "step": 2735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.116551981563903e-06,
+      "loss": 2.536,
+      "step": 2736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.114981788261642e-06,
+      "loss": 2.6239,
+      "step": 2737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.113411092735262e-06,
+      "loss": 2.4997,
+      "step": 2738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.111839895238e-06,
+      "loss": 2.4794,
+      "step": 2739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.110268196023179e-06,
+      "loss": 2.5425,
+      "step": 2740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.108695995344202e-06,
+      "loss": 2.4319,
+      "step": 2741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.107123293454555e-06,
+      "loss": 2.4695,
+      "step": 2742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.1055500906078e-06,
+      "loss": 2.3558,
+      "step": 2743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.103976387057579e-06,
+      "loss": 2.6216,
+      "step": 2744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.102402183057624e-06,
+      "loss": 2.7285,
+      "step": 2745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.100827478861741e-06,
+      "loss": 2.6063,
+      "step": 2746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.099252274723814e-06,
+      "loss": 2.4137,
+      "step": 2747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.097676570897814e-06,
+      "loss": 2.496,
+      "step": 2748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.096100367637787e-06,
+      "loss": 2.5072,
+      "step": 2749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.094523665197862e-06,
+      "loss": 2.7217,
+      "step": 2750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.092946463832251e-06,
+      "loss": 2.4987,
+      "step": 2751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.091368763795244e-06,
+      "loss": 2.4136,
+      "step": 2752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.089790565341208e-06,
+      "loss": 2.4752,
+      "step": 2753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.088211868724595e-06,
+      "loss": 2.5046,
+      "step": 2754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.086632674199939e-06,
+      "loss": 2.6758,
+      "step": 2755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.085052982021849e-06,
+      "loss": 2.6192,
+      "step": 2756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.083472792445014e-06,
+      "loss": 2.6245,
+      "step": 2757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.081892105724212e-06,
+      "loss": 2.6182,
+      "step": 2758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.080310922114287e-06,
+      "loss": 2.5678,
+      "step": 2759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.078729241870177e-06,
+      "loss": 2.6756,
+      "step": 2760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.077147065246895e-06,
+      "loss": 2.4666,
+      "step": 2761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.075564392499527e-06,
+      "loss": 2.6777,
+      "step": 2762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.07398122388325e-06,
+      "loss": 2.6701,
+      "step": 2763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.072397559653314e-06,
+      "loss": 2.6755,
+      "step": 2764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.070813400065051e-06,
+      "loss": 2.6468,
+      "step": 2765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.069228745373875e-06,
+      "loss": 2.6097,
+      "step": 2766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.067643595835274e-06,
+      "loss": 2.5563,
+      "step": 2767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.066057951704821e-06,
+      "loss": 2.6363,
+      "step": 2768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.06447181323817e-06,
+      "loss": 2.5664,
+      "step": 2769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.062885180691046e-06,
+      "loss": 2.5854,
+      "step": 2770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.061298054319267e-06,
+      "loss": 2.7208,
+      "step": 2771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.059710434378717e-06,
+      "loss": 2.6048,
+      "step": 2772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.058122321125368e-06,
+      "loss": 2.6209,
+      "step": 2773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.056533714815268e-06,
+      "loss": 2.6068,
+      "step": 2774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.054944615704548e-06,
+      "loss": 2.563,
+      "step": 2775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.053355024049415e-06,
+      "loss": 2.6052,
+      "step": 2776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.051764940106157e-06,
+      "loss": 2.5926,
+      "step": 2777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.05017436413114e-06,
+      "loss": 2.4734,
+      "step": 2778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.04858329638081e-06,
+      "loss": 2.4617,
+      "step": 2779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.046991737111696e-06,
+      "loss": 2.4535,
+      "step": 2780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.045399686580399e-06,
+      "loss": 2.7043,
+      "step": 2781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.043807145043604e-06,
+      "loss": 2.691,
+      "step": 2782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.042214112758075e-06,
+      "loss": 2.5659,
+      "step": 2783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.040620589980654e-06,
+      "loss": 2.6798,
+      "step": 2784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.039026576968262e-06,
+      "loss": 2.5984,
+      "step": 2785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.037432073977902e-06,
+      "loss": 2.3856,
+      "step": 2786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.03583708126665e-06,
+      "loss": 2.6073,
+      "step": 2787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.034241599091666e-06,
+      "loss": 2.572,
+      "step": 2788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.032645627710188e-06,
+      "loss": 2.6734,
+      "step": 2789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.03104916737953e-06,
+      "loss": 2.5634,
+      "step": 2790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.02945221835709e-06,
+      "loss": 2.5344,
+      "step": 2791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.027854780900339e-06,
+      "loss": 2.5693,
+      "step": 2792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.02625685526683e-06,
+      "loss": 2.5066,
+      "step": 2793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.024658441714197e-06,
+      "loss": 2.4872,
+      "step": 2794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.023059540500149e-06,
+      "loss": 2.7201,
+      "step": 2795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.021460151882472e-06,
+      "loss": 2.4534,
+      "step": 2796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.019860276119036e-06,
+      "loss": 2.5352,
+      "step": 2797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.018259913467785e-06,
+      "loss": 2.4837,
+      "step": 2798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.016659064186742e-06,
+      "loss": 2.715,
+      "step": 2799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.015057728534015e-06,
+      "loss": 2.5463,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.01345590676778e-06,
+      "loss": 2.5789,
+      "step": 2801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.011853599146297e-06,
+      "loss": 2.6799,
+      "step": 2802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.010250805927907e-06,
+      "loss": 2.651,
+      "step": 2803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.008647527371022e-06,
+      "loss": 2.4357,
+      "step": 2804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.007043763734139e-06,
+      "loss": 2.6758,
+      "step": 2805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.00543951527583e-06,
+      "loss": 2.6267,
+      "step": 2806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.003834782254747e-06,
+      "loss": 2.5666,
+      "step": 2807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.002229564929616e-06,
+      "loss": 2.2994,
+      "step": 2808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.000623863559245e-06,
+      "loss": 2.4898,
+      "step": 2809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.999017678402518e-06,
+      "loss": 2.8418,
+      "step": 2810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.997411009718401e-06,
+      "loss": 2.5341,
+      "step": 2811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.995803857765934e-06,
+      "loss": 2.4653,
+      "step": 2812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.994196222804234e-06,
+      "loss": 2.8119,
+      "step": 2813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.992588105092498e-06,
+      "loss": 2.5161,
+      "step": 2814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.990979504890003e-06,
+      "loss": 2.5044,
+      "step": 2815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.989370422456098e-06,
+      "loss": 2.7166,
+      "step": 2816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.987760858050216e-06,
+      "loss": 2.5338,
+      "step": 2817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.986150811931864e-06,
+      "loss": 2.5608,
+      "step": 2818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.984540284360626e-06,
+      "loss": 2.5387,
+      "step": 2819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.982929275596164e-06,
+      "loss": 2.5447,
+      "step": 2820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.981317785898225e-06,
+      "loss": 2.5504,
+      "step": 2821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.979705815526621e-06,
+      "loss": 2.6578,
+      "step": 2822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.97809336474125e-06,
+      "loss": 2.5047,
+      "step": 2823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.976480433802085e-06,
+      "loss": 2.5139,
+      "step": 2824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.974867022969176e-06,
+      "loss": 2.6403,
+      "step": 2825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.973253132502652e-06,
+      "loss": 2.676,
+      "step": 2826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.971638762662717e-06,
+      "loss": 2.5525,
+      "step": 2827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.970023913709652e-06,
+      "loss": 2.6945,
+      "step": 2828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.968408585903823e-06,
+      "loss": 2.5899,
+      "step": 2829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.96679277950566e-06,
+      "loss": 2.5245,
+      "step": 2830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.965176494775684e-06,
+      "loss": 2.556,
+      "step": 2831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.96355973197448e-06,
+      "loss": 2.5199,
+      "step": 2832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.961942491362719e-06,
+      "loss": 2.543,
+      "step": 2833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.960324773201147e-06,
+      "loss": 2.5165,
+      "step": 2834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.958706577750586e-06,
+      "loss": 2.6465,
+      "step": 2835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.957087905271934e-06,
+      "loss": 2.5395,
+      "step": 2836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.955468756026168e-06,
+      "loss": 2.5309,
+      "step": 2837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.953849130274344e-06,
+      "loss": 2.5853,
+      "step": 2838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.952229028277588e-06,
+      "loss": 2.5978,
+      "step": 2839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.95060845029711e-06,
+      "loss": 2.4975,
+      "step": 2840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.94898739659419e-06,
+      "loss": 2.5247,
+      "step": 2841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.94736586743019e-06,
+      "loss": 2.527,
+      "step": 2842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.94574386306655e-06,
+      "loss": 2.5681,
+      "step": 2843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.944121383764775e-06,
+      "loss": 2.4952,
+      "step": 2844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.942498429786466e-06,
+      "loss": 2.456,
+      "step": 2845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.94087500139328e-06,
+      "loss": 2.7886,
+      "step": 2846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.939251098846968e-06,
+      "loss": 2.5056,
+      "step": 2847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.937626722409342e-06,
+      "loss": 2.3708,
+      "step": 2848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.936001872342303e-06,
+      "loss": 2.6743,
+      "step": 2849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.93437654890782e-06,
+      "loss": 2.5541,
+      "step": 2850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.932750752367947e-06,
+      "loss": 2.4163,
+      "step": 2851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.931124482984802e-06,
+      "loss": 2.6016,
+      "step": 2852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.929497741020592e-06,
+      "loss": 2.579,
+      "step": 2853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.92787052673759e-06,
+      "loss": 2.7825,
+      "step": 2854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.926242840398152e-06,
+      "loss": 2.5535,
+      "step": 2855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.924614682264707e-06,
+      "loss": 2.5667,
+      "step": 2856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.922986052599759e-06,
+      "loss": 2.6106,
+      "step": 2857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.921356951665894e-06,
+      "loss": 2.6688,
+      "step": 2858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.919727379725763e-06,
+      "loss": 2.4968,
+      "step": 2859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.918097337042106e-06,
+      "loss": 2.5996,
+      "step": 2860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.91646682387773e-06,
+      "loss": 2.5838,
+      "step": 2861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.914835840495518e-06,
+      "loss": 2.542,
+      "step": 2862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.913204387158435e-06,
+      "loss": 2.5822,
+      "step": 2863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.911572464129515e-06,
+      "loss": 2.407,
+      "step": 2864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.909940071671874e-06,
+      "loss": 2.6295,
+      "step": 2865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.908307210048695e-06,
+      "loss": 2.461,
+      "step": 2866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.906673879523249e-06,
+      "loss": 2.4688,
+      "step": 2867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.905040080358869e-06,
+      "loss": 2.5864,
+      "step": 2868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.903405812818974e-06,
+      "loss": 2.5513,
+      "step": 2869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.901771077167051e-06,
+      "loss": 2.5014,
+      "step": 2870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.900135873666672e-06,
+      "loss": 2.606,
+      "step": 2871
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.898500202581474e-06,
+      "loss": 2.7239,
+      "step": 2872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.896864064175176e-06,
+      "loss": 2.5848,
+      "step": 2873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.895227458711567e-06,
+      "loss": 2.6267,
+      "step": 2874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.89359038645452e-06,
+      "loss": 2.5691,
+      "step": 2875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.891952847667973e-06,
+      "loss": 2.6508,
+      "step": 2876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.890314842615948e-06,
+      "loss": 2.4033,
+      "step": 2877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.888676371562534e-06,
+      "loss": 2.5467,
+      "step": 2878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.887037434771902e-06,
+      "loss": 2.6814,
+      "step": 2879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.885398032508296e-06,
+      "loss": 2.6065,
+      "step": 2880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.883758165036032e-06,
+      "loss": 2.5961,
+      "step": 2881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.882117832619506e-06,
+      "loss": 2.7195,
+      "step": 2882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.880477035523187e-06,
+      "loss": 2.6041,
+      "step": 2883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.878835774011615e-06,
+      "loss": 2.4418,
+      "step": 2884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.877194048349413e-06,
+      "loss": 2.515,
+      "step": 2885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.875551858801268e-06,
+      "loss": 2.4094,
+      "step": 2886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.873909205631951e-06,
+      "loss": 2.5801,
+      "step": 2887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.872266089106309e-06,
+      "loss": 2.7062,
+      "step": 2888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.870622509489251e-06,
+      "loss": 2.5632,
+      "step": 2889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.868978467045777e-06,
+      "loss": 2.5447,
+      "step": 2890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.867333962040948e-06,
+      "loss": 2.6751,
+      "step": 2891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.865688994739907e-06,
+      "loss": 2.51,
+      "step": 2892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.864043565407871e-06,
+      "loss": 2.7387,
+      "step": 2893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.862397674310127e-06,
+      "loss": 2.7214,
+      "step": 2894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.860751321712043e-06,
+      "loss": 2.4684,
+      "step": 2895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.859104507879058e-06,
+      "loss": 2.5351,
+      "step": 2896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.857457233076684e-06,
+      "loss": 2.6763,
+      "step": 2897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.85580949757051e-06,
+      "loss": 2.6832,
+      "step": 2898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.854161301626196e-06,
+      "loss": 2.5517,
+      "step": 2899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.85251264550948e-06,
+      "loss": 2.4681,
+      "step": 2900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.850863529486172e-06,
+      "loss": 2.5925,
+      "step": 2901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.849213953822158e-06,
+      "loss": 2.6068,
+      "step": 2902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.847563918783393e-06,
+      "loss": 2.5288,
+      "step": 2903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.845913424635915e-06,
+      "loss": 2.521,
+      "step": 2904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.844262471645827e-06,
+      "loss": 2.4423,
+      "step": 2905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.842611060079312e-06,
+      "loss": 2.3436,
+      "step": 2906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.840959190202623e-06,
+      "loss": 2.6797,
+      "step": 2907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.83930686228209e-06,
+      "loss": 2.7181,
+      "step": 2908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.837654076584115e-06,
+      "loss": 2.5138,
+      "step": 2909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.836000833375174e-06,
+      "loss": 2.6641,
+      "step": 2910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.834347132921817e-06,
+      "loss": 2.5491,
+      "step": 2911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.832692975490668e-06,
+      "loss": 2.4358,
+      "step": 2912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.831038361348425e-06,
+      "loss": 2.6112,
+      "step": 2913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.829383290761859e-06,
+      "loss": 2.7002,
+      "step": 2914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.827727763997815e-06,
+      "loss": 2.4583,
+      "step": 2915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.826071781323208e-06,
+      "loss": 2.5323,
+      "step": 2916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.824415343005035e-06,
+      "loss": 2.4228,
+      "step": 2917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.822758449310357e-06,
+      "loss": 2.4258,
+      "step": 2918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.821101100506317e-06,
+      "loss": 2.6686,
+      "step": 2919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.819443296860121e-06,
+      "loss": 2.6419,
+      "step": 2920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.817785038639058e-06,
+      "loss": 2.7657,
+      "step": 2921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.816126326110487e-06,
+      "loss": 2.5512,
+      "step": 2922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.814467159541841e-06,
+      "loss": 2.651,
+      "step": 2923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.812807539200622e-06,
+      "loss": 2.4833,
+      "step": 2924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.811147465354408e-06,
+      "loss": 2.5305,
+      "step": 2925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.809486938270853e-06,
+      "loss": 2.7539,
+      "step": 2926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.80782595821768e-06,
+      "loss": 2.445,
+      "step": 2927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.806164525462687e-06,
+      "loss": 2.501,
+      "step": 2928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.804502640273743e-06,
+      "loss": 2.7106,
+      "step": 2929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.802840302918794e-06,
+      "loss": 2.7369,
+      "step": 2930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.801177513665856e-06,
+      "loss": 2.5599,
+      "step": 2931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.799514272783014e-06,
+      "loss": 2.3928,
+      "step": 2932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.797850580538435e-06,
+      "loss": 2.4631,
+      "step": 2933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.796186437200352e-06,
+      "loss": 2.6008,
+      "step": 2934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.79452184303707e-06,
+      "loss": 2.3959,
+      "step": 2935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.792856798316975e-06,
+      "loss": 2.7049,
+      "step": 2936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.791191303308513e-06,
+      "loss": 2.6206,
+      "step": 2937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.789525358280216e-06,
+      "loss": 2.6765,
+      "step": 2938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.787858963500676e-06,
+      "loss": 2.4868,
+      "step": 2939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.786192119238568e-06,
+      "loss": 2.5133,
+      "step": 2940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.784524825762632e-06,
+      "loss": 2.6607,
+      "step": 2941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.782857083341685e-06,
+      "loss": 2.6656,
+      "step": 2942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.781188892244613e-06,
+      "loss": 2.447,
+      "step": 2943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.779520252740378e-06,
+      "loss": 2.3656,
+      "step": 2944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.777851165098012e-06,
+      "loss": 2.3781,
+      "step": 2945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.776181629586618e-06,
+      "loss": 2.5658,
+      "step": 2946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.774511646475376e-06,
+      "loss": 2.6343,
+      "step": 2947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.772841216033534e-06,
+      "loss": 2.5894,
+      "step": 2948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.771170338530411e-06,
+      "loss": 2.6371,
+      "step": 2949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.769499014235401e-06,
+      "loss": 2.5845,
+      "step": 2950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.767827243417971e-06,
+      "loss": 2.6005,
+      "step": 2951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.766155026347657e-06,
+      "loss": 2.5869,
+      "step": 2952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.764482363294068e-06,
+      "loss": 2.5016,
+      "step": 2953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.762809254526884e-06,
+      "loss": 2.5446,
+      "step": 2954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.761135700315863e-06,
+      "loss": 2.6095,
+      "step": 2955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.759461700930824e-06,
+      "loss": 2.6276,
+      "step": 2956
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.757787256641666e-06,
+      "loss": 2.5471,
+      "step": 2957
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.756112367718356e-06,
+      "loss": 2.482,
+      "step": 2958
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.754437034430935e-06,
+      "loss": 2.4878,
+      "step": 2959
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.752761257049517e-06,
+      "loss": 2.3928,
+      "step": 2960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.751085035844281e-06,
+      "loss": 2.5217,
+      "step": 2961
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.749408371085484e-06,
+      "loss": 2.61,
+      "step": 2962
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.747731263043453e-06,
+      "loss": 2.5169,
+      "step": 2963
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.746053711988584e-06,
+      "loss": 2.6868,
+      "step": 2964
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.744375718191347e-06,
+      "loss": 2.4561,
+      "step": 2965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.74269728192228e-06,
+      "loss": 2.5379,
+      "step": 2966
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.741018403452e-06,
+      "loss": 2.6429,
+      "step": 2967
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.739339083051186e-06,
+      "loss": 2.5006,
+      "step": 2968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.737659320990598e-06,
+      "loss": 2.5228,
+      "step": 2969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.735979117541054e-06,
+      "loss": 2.4733,
+      "step": 2970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.734298472973454e-06,
+      "loss": 2.727,
+      "step": 2971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.732617387558769e-06,
+      "loss": 2.5065,
+      "step": 2972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.730935861568034e-06,
+      "loss": 2.3727,
+      "step": 2973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.729253895272362e-06,
+      "loss": 2.5114,
+      "step": 2974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.727571488942931e-06,
+      "loss": 2.5958,
+      "step": 2975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.725888642850995e-06,
+      "loss": 2.6005,
+      "step": 2976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.724205357267877e-06,
+      "loss": 2.4978,
+      "step": 2977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.722521632464972e-06,
+      "loss": 2.5824,
+      "step": 2978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.72083746871374e-06,
+      "loss": 2.4628,
+      "step": 2979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.719152866285722e-06,
+      "loss": 2.5842,
+      "step": 2980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.717467825452518e-06,
+      "loss": 2.5439,
+      "step": 2981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.715782346485809e-06,
+      "loss": 2.5301,
+      "step": 2982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.714096429657344e-06,
+      "loss": 2.5519,
+      "step": 2983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.712410075238936e-06,
+      "loss": 2.6597,
+      "step": 2984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.710723283502476e-06,
+      "loss": 2.5035,
+      "step": 2985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.709036054719926e-06,
+      "loss": 2.4965,
+      "step": 2986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.707348389163309e-06,
+      "loss": 2.5823,
+      "step": 2987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.70566028710473e-06,
+      "loss": 2.4176,
+      "step": 2988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.703971748816358e-06,
+      "loss": 2.558,
+      "step": 2989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.702282774570434e-06,
+      "loss": 2.5089,
+      "step": 2990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.700593364639269e-06,
+      "loss": 2.5334,
+      "step": 2991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.698903519295242e-06,
+      "loss": 2.3934,
+      "step": 2992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.697213238810806e-06,
+      "loss": 2.7482,
+      "step": 2993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.695522523458484e-06,
+      "loss": 2.7303,
+      "step": 2994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.693831373510864e-06,
+      "loss": 2.5855,
+      "step": 2995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.692139789240611e-06,
+      "loss": 2.4197,
+      "step": 2996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.690447770920456e-06,
+      "loss": 2.627,
+      "step": 2997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6887553188232e-06,
+      "loss": 2.6645,
+      "step": 2998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.687062433221715e-06,
+      "loss": 2.6122,
+      "step": 2999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.685369114388943e-06,
+      "loss": 2.5169,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.683675362597893e-06,
+      "loss": 2.5068,
+      "step": 3001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.681981178121652e-06,
+      "loss": 2.5219,
+      "step": 3002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.680286561233365e-06,
+      "loss": 2.6009,
+      "step": 3003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.678591512206254e-06,
+      "loss": 2.5132,
+      "step": 3004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.676896031313612e-06,
+      "loss": 2.5926,
+      "step": 3005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.675200118828796e-06,
+      "loss": 2.5614,
+      "step": 3006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.673503775025237e-06,
+      "loss": 2.6382,
+      "step": 3007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.671807000176434e-06,
+      "loss": 2.5241,
+      "step": 3008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.670109794555955e-06,
+      "loss": 2.4602,
+      "step": 3009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.668412158437442e-06,
+      "loss": 2.5793,
+      "step": 3010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.666714092094596e-06,
+      "loss": 2.5783,
+      "step": 3011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.665015595801198e-06,
+      "loss": 2.4706,
+      "step": 3012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.663316669831092e-06,
+      "loss": 2.601,
+      "step": 3013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.661617314458194e-06,
+      "loss": 2.6002,
+      "step": 3014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.659917529956491e-06,
+      "loss": 2.3782,
+      "step": 3015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.658217316600033e-06,
+      "loss": 2.7325,
+      "step": 3016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.656516674662948e-06,
+      "loss": 2.7176,
+      "step": 3017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.654815604419424e-06,
+      "loss": 2.598,
+      "step": 3018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.653114106143723e-06,
+      "loss": 2.5905,
+      "step": 3019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.651412180110176e-06,
+      "loss": 2.5564,
+      "step": 3020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.649709826593182e-06,
+      "loss": 2.5066,
+      "step": 3021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.648007045867208e-06,
+      "loss": 2.6303,
+      "step": 3022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.646303838206791e-06,
+      "loss": 2.512,
+      "step": 3023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.644600203886537e-06,
+      "loss": 2.6508,
+      "step": 3024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.642896143181123e-06,
+      "loss": 2.5663,
+      "step": 3025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.641191656365288e-06,
+      "loss": 2.4999,
+      "step": 3026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.639486743713849e-06,
+      "loss": 2.73,
+      "step": 3027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.637781405501682e-06,
+      "loss": 2.4688,
+      "step": 3028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.636075642003738e-06,
+      "loss": 2.4947,
+      "step": 3029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.634369453495036e-06,
+      "loss": 2.6289,
+      "step": 3030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.632662840250661e-06,
+      "loss": 2.4107,
+      "step": 3031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.630955802545768e-06,
+      "loss": 2.6414,
+      "step": 3032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.629248340655582e-06,
+      "loss": 2.6906,
+      "step": 3033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.627540454855391e-06,
+      "loss": 2.575,
+      "step": 3034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6258321454205586e-06,
+      "loss": 2.5484,
+      "step": 3035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6241234126265115e-06,
+      "loss": 2.6002,
+      "step": 3036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.622414256748747e-06,
+      "loss": 2.5317,
+      "step": 3037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6207046780628295e-06,
+      "loss": 2.6132,
+      "step": 3038
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.618994676844391e-06,
+      "loss": 2.6884,
+      "step": 3039
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.617284253369135e-06,
+      "loss": 2.5946,
+      "step": 3040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.615573407912827e-06,
+      "loss": 2.5349,
+      "step": 3041
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.6138621407513084e-06,
+      "loss": 2.5247,
+      "step": 3042
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.6121504521604805e-06,
+      "loss": 2.5267,
+      "step": 3043
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.61043834241632e-06,
+      "loss": 2.4852,
+      "step": 3044
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.608725811794866e-06,
+      "loss": 2.4774,
+      "step": 3045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.607012860572227e-06,
+      "loss": 2.656,
+      "step": 3046
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.60529948902458e-06,
+      "loss": 2.5886,
+      "step": 3047
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.603585697428169e-06,
+      "loss": 2.6306,
+      "step": 3048
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.601871486059309e-06,
+      "loss": 2.6293,
+      "step": 3049
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.600156855194376e-06,
+      "loss": 2.5962,
+      "step": 3050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.598441805109822e-06,
+      "loss": 2.478,
+      "step": 3051
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.596726336082158e-06,
+      "loss": 2.5075,
+      "step": 3052
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.595010448387967e-06,
+      "loss": 2.7038,
+      "step": 3053
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.593294142303902e-06,
+      "loss": 2.3905,
+      "step": 3054
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.591577418106678e-06,
+      "loss": 2.4451,
+      "step": 3055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.589860276073082e-06,
+      "loss": 2.6799,
+      "step": 3056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.588142716479965e-06,
+      "loss": 2.3405,
+      "step": 3057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.586424739604248e-06,
+      "loss": 2.7073,
+      "step": 3058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.584706345722914e-06,
+      "loss": 2.5276,
+      "step": 3059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5829875351130224e-06,
+      "loss": 2.3663,
+      "step": 3060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.581268308051692e-06,
+      "loss": 2.5418,
+      "step": 3061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.579548664816112e-06,
+      "loss": 2.3754,
+      "step": 3062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5778286056835395e-06,
+      "loss": 2.586,
+      "step": 3063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.576108130931292e-06,
+      "loss": 2.5898,
+      "step": 3064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5743872408367645e-06,
+      "loss": 2.5478,
+      "step": 3065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.572665935677412e-06,
+      "loss": 2.472,
+      "step": 3066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.57094421573076e-06,
+      "loss": 2.6841,
+      "step": 3067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.569222081274396e-06,
+      "loss": 2.416,
+      "step": 3068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.567499532585978e-06,
+      "loss": 2.7056,
+      "step": 3069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.565776569943231e-06,
+      "loss": 2.6592,
+      "step": 3070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.564053193623944e-06,
+      "loss": 2.2394,
+      "step": 3071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.562329403905976e-06,
+      "loss": 2.6694,
+      "step": 3072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5606052010672525e-06,
+      "loss": 2.5527,
+      "step": 3073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.558880585385762e-06,
+      "loss": 2.6286,
+      "step": 3074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.557155557139563e-06,
+      "loss": 2.6077,
+      "step": 3075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.555430116606778e-06,
+      "loss": 2.6389,
+      "step": 3076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.553704264065598e-06,
+      "loss": 2.5701,
+      "step": 3077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5519779997942825e-06,
+      "loss": 2.5976,
+      "step": 3078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.550251324071152e-06,
+      "loss": 2.5225,
+      "step": 3079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.548524237174595e-06,
+      "loss": 2.7193,
+      "step": 3080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5467967393830686e-06,
+      "loss": 2.5269,
+      "step": 3081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.545068830975096e-06,
+      "loss": 2.7224,
+      "step": 3082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.543340512229263e-06,
+      "loss": 2.4984,
+      "step": 3083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5416117834242254e-06,
+      "loss": 2.575,
+      "step": 3084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5398826448387055e-06,
+      "loss": 2.6468,
+      "step": 3085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.538153096751486e-06,
+      "loss": 2.5131,
+      "step": 3086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.536423139441421e-06,
+      "loss": 2.5475,
+      "step": 3087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.534692773187431e-06,
+      "loss": 2.4431,
+      "step": 3088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.532961998268498e-06,
+      "loss": 2.7499,
+      "step": 3089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.531230814963675e-06,
+      "loss": 2.5988,
+      "step": 3090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.529499223552075e-06,
+      "loss": 2.4,
+      "step": 3091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.527767224312883e-06,
+      "loss": 2.6532,
+      "step": 3092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.526034817525345e-06,
+      "loss": 2.6422,
+      "step": 3093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.524302003468774e-06,
+      "loss": 2.6479,
+      "step": 3094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.522568782422553e-06,
+      "loss": 2.4361,
+      "step": 3095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.520835154666121e-06,
+      "loss": 2.6634,
+      "step": 3096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.519101120478993e-06,
+      "loss": 2.5177,
+      "step": 3097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.517366680140743e-06,
+      "loss": 2.6319,
+      "step": 3098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.515631833931012e-06,
+      "loss": 2.6566,
+      "step": 3099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.513896582129507e-06,
+      "loss": 2.7131,
+      "step": 3100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5121609250160024e-06,
+      "loss": 2.5292,
+      "step": 3101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.510424862870334e-06,
+      "loss": 2.6434,
+      "step": 3102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.508688395972403e-06,
+      "loss": 2.5826,
+      "step": 3103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.50695152460218e-06,
+      "loss": 2.6154,
+      "step": 3104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.505214249039696e-06,
+      "loss": 2.4873,
+      "step": 3105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.503476569565052e-06,
+      "loss": 2.5442,
+      "step": 3106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.501738486458411e-06,
+      "loss": 2.5804,
+      "step": 3107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.6347,
+      "step": 3108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.498261110470116e-06,
+      "loss": 2.6056,
+      "step": 3109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.496521818149114e-06,
+      "loss": 2.4335,
+      "step": 3110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.49478212331742e-06,
+      "loss": 2.4643,
+      "step": 3111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.493042026255522e-06,
+      "loss": 2.4925,
+      "step": 3112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4913015272439725e-06,
+      "loss": 2.6064,
+      "step": 3113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.489560626563392e-06,
+      "loss": 2.641,
+      "step": 3114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.48781932449446e-06,
+      "loss": 2.5651,
+      "step": 3115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4860776213179264e-06,
+      "loss": 2.6674,
+      "step": 3116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.484335517314603e-06,
+      "loss": 2.6051,
+      "step": 3117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.482593012765368e-06,
+      "loss": 2.6486,
+      "step": 3118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.480850107951162e-06,
+      "loss": 2.7443,
+      "step": 3119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.479106803152991e-06,
+      "loss": 2.6005,
+      "step": 3120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.477363098651925e-06,
+      "loss": 2.5229,
+      "step": 3121
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4756189947291e-06,
+      "loss": 2.5671,
+      "step": 3122
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.473874491665715e-06,
+      "loss": 2.4166,
+      "step": 3123
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.472129589743034e-06,
+      "loss": 2.5951,
+      "step": 3124
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.470384289242385e-06,
+      "loss": 2.402,
+      "step": 3125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.468638590445159e-06,
+      "loss": 2.5425,
+      "step": 3126
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.466892493632816e-06,
+      "loss": 2.6248,
+      "step": 3127
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.465145999086874e-06,
+      "loss": 2.6268,
+      "step": 3128
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.463399107088917e-06,
+      "loss": 2.6184,
+      "step": 3129
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.461651817920595e-06,
+      "loss": 2.417,
+      "step": 3130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.459904131863623e-06,
+      "loss": 2.6385,
+      "step": 3131
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.458156049199775e-06,
+      "loss": 2.5573,
+      "step": 3132
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.456407570210894e-06,
+      "loss": 2.5578,
+      "step": 3133
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.454658695178885e-06,
+      "loss": 2.3566,
+      "step": 3134
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.452909424385712e-06,
+      "loss": 2.6145,
+      "step": 3135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.451159758113414e-06,
+      "loss": 2.4296,
+      "step": 3136
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.449409696644083e-06,
+      "loss": 2.4347,
+      "step": 3137
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.447659240259881e-06,
+      "loss": 2.4913,
+      "step": 3138
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4459083892430315e-06,
+      "loss": 2.5592,
+      "step": 3139
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.44415714387582e-06,
+      "loss": 2.5811,
+      "step": 3140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.442405504440598e-06,
+      "loss": 2.6652,
+      "step": 3141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.440653471219782e-06,
+      "loss": 2.5482,
+      "step": 3142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.438901044495846e-06,
+      "loss": 2.6439,
+      "step": 3143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4371482245513345e-06,
+      "loss": 2.5798,
+      "step": 3144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.435395011668853e-06,
+      "loss": 2.4018,
+      "step": 3145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.433641406131065e-06,
+      "loss": 2.6193,
+      "step": 3146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.431887408220705e-06,
+      "loss": 2.6327,
+      "step": 3147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.430133018220567e-06,
+      "loss": 2.5548,
+      "step": 3148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.428378236413509e-06,
+      "loss": 2.6812,
+      "step": 3149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.426623063082452e-06,
+      "loss": 2.6538,
+      "step": 3150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.424867498510381e-06,
+      "loss": 2.6022,
+      "step": 3151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4231115429803405e-06,
+      "loss": 2.5677,
+      "step": 3152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.421355196775444e-06,
+      "loss": 2.663,
+      "step": 3153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.41959846017886e-06,
+      "loss": 2.6133,
+      "step": 3154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.417841333473832e-06,
+      "loss": 2.6399,
+      "step": 3155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.416083816943653e-06,
+      "loss": 2.732,
+      "step": 3156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.414325910871688e-06,
+      "loss": 2.6252,
+      "step": 3157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.41256761554136e-06,
+      "loss": 2.5641,
+      "step": 3158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.410808931236155e-06,
+      "loss": 2.4764,
+      "step": 3159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.409049858239627e-06,
+      "loss": 2.4578,
+      "step": 3160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.407290396835387e-06,
+      "loss": 2.4636,
+      "step": 3161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.405530547307112e-06,
+      "loss": 2.5102,
+      "step": 3162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.403770309938537e-06,
+      "loss": 2.5542,
+      "step": 3163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4020096850134635e-06,
+      "loss": 2.4984,
+      "step": 3164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.400248672815755e-06,
+      "loss": 2.7014,
+      "step": 3165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.398487273629339e-06,
+      "loss": 2.5029,
+      "step": 3166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.396725487738201e-06,
+      "loss": 2.5174,
+      "step": 3167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.394963315426393e-06,
+      "loss": 2.5466,
+      "step": 3168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.393200756978025e-06,
+      "loss": 2.5507,
+      "step": 3169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.391437812677274e-06,
+      "loss": 2.7138,
+      "step": 3170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.389674482808375e-06,
+      "loss": 2.6353,
+      "step": 3171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.38791076765563e-06,
+      "loss": 2.4847,
+      "step": 3172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3861466675034e-06,
+      "loss": 2.3597,
+      "step": 3173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.384382182636105e-06,
+      "loss": 2.6277,
+      "step": 3174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.382617313338234e-06,
+      "loss": 2.6321,
+      "step": 3175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.380852059894331e-06,
+      "loss": 2.6686,
+      "step": 3176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.379086422589009e-06,
+      "loss": 2.6114,
+      "step": 3177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.377320401706937e-06,
+      "loss": 2.6719,
+      "step": 3178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.375553997532851e-06,
+      "loss": 2.517,
+      "step": 3179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.37378721035154e-06,
+      "loss": 2.4126,
+      "step": 3180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.372020040447866e-06,
+      "loss": 2.5419,
+      "step": 3181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.370252488106744e-06,
+      "loss": 2.5483,
+      "step": 3182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.368484553613157e-06,
+      "loss": 2.5974,
+      "step": 3183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3667162372521436e-06,
+      "loss": 2.6297,
+      "step": 3184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.36494753930881e-06,
+      "loss": 2.6335,
+      "step": 3185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.363178460068319e-06,
+      "loss": 2.429,
+      "step": 3186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.361408999815895e-06,
+      "loss": 2.4682,
+      "step": 3187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.359639158836828e-06,
+      "loss": 2.3934,
+      "step": 3188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.357868937416467e-06,
+      "loss": 2.5834,
+      "step": 3189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.356098335840221e-06,
+      "loss": 2.6682,
+      "step": 3190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3543273543935645e-06,
+      "loss": 2.5951,
+      "step": 3191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.352555993362025e-06,
+      "loss": 2.5319,
+      "step": 3192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.350784253031202e-06,
+      "loss": 2.5352,
+      "step": 3193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.349012133686748e-06,
+      "loss": 2.4875,
+      "step": 3194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3472396356143806e-06,
+      "loss": 2.7119,
+      "step": 3195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.345466759099875e-06,
+      "loss": 2.5444,
+      "step": 3196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.343693504429071e-06,
+      "loss": 2.424,
+      "step": 3197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.341919871887868e-06,
+      "loss": 2.6463,
+      "step": 3198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.340145861762226e-06,
+      "loss": 2.5197,
+      "step": 3199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3383714743381664e-06,
+      "loss": 2.4189,
+      "step": 3200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3365967099017725e-06,
+      "loss": 2.4497,
+      "step": 3201
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3348215687391844e-06,
+      "loss": 2.5978,
+      "step": 3202
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.333046051136606e-06,
+      "loss": 2.7241,
+      "step": 3203
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.331270157380304e-06,
+      "loss": 2.4933,
+      "step": 3204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.329493887756599e-06,
+      "loss": 2.6976,
+      "step": 3205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.327717242551882e-06,
+      "loss": 2.6631,
+      "step": 3206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.325940222052594e-06,
+      "loss": 2.6559,
+      "step": 3207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.324162826545245e-06,
+      "loss": 2.6597,
+      "step": 3208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.3223850563163985e-06,
+      "loss": 2.6216,
+      "step": 3209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.3206069116526855e-06,
+      "loss": 2.5557,
+      "step": 3210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.318828392840791e-06,
+      "loss": 2.435,
+      "step": 3211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.317049500167466e-06,
+      "loss": 2.4956,
+      "step": 3212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.315270233919516e-06,
+      "loss": 2.4526,
+      "step": 3213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.313490594383812e-06,
+      "loss": 2.544,
+      "step": 3214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.3117105818472786e-06,
+      "loss": 2.726,
+      "step": 3215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.309930196596909e-06,
+      "loss": 2.7468,
+      "step": 3216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.308149438919752e-06,
+      "loss": 2.7192,
+      "step": 3217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.306368309102914e-06,
+      "loss": 2.4523,
+      "step": 3218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.304586807433567e-06,
+      "loss": 2.6558,
+      "step": 3219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.302804934198937e-06,
+      "loss": 2.4579,
+      "step": 3220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.301022689686313e-06,
+      "loss": 2.6612,
+      "step": 3221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.299240074183047e-06,
+      "loss": 2.5375,
+      "step": 3222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.297457087976546e-06,
+      "loss": 2.5869,
+      "step": 3223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.295673731354278e-06,
+      "loss": 2.6784,
+      "step": 3224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2938900046037705e-06,
+      "loss": 2.6059,
+      "step": 3225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.292105908012612e-06,
+      "loss": 2.416,
+      "step": 3226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.29032144186845e-06,
+      "loss": 2.5658,
+      "step": 3227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.28853660645899e-06,
+      "loss": 2.6764,
+      "step": 3228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.286751402072002e-06,
+      "loss": 2.6524,
+      "step": 3229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2849658289953084e-06,
+      "loss": 2.6203,
+      "step": 3230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.283179887516797e-06,
+      "loss": 2.7371,
+      "step": 3231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.28139357792441e-06,
+      "loss": 2.513,
+      "step": 3232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2796069005061545e-06,
+      "loss": 2.6752,
+      "step": 3233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.277819855550091e-06,
+      "loss": 2.514,
+      "step": 3234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2760324433443455e-06,
+      "loss": 2.6619,
+      "step": 3235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2742446641770985e-06,
+      "loss": 2.6035,
+      "step": 3236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.272456518336589e-06,
+      "loss": 2.6032,
+      "step": 3237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.270668006111121e-06,
+      "loss": 2.6293,
+      "step": 3238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.268879127789051e-06,
+      "loss": 2.7696,
+      "step": 3239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.267089883658798e-06,
+      "loss": 2.5116,
+      "step": 3240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.265300274008841e-06,
+      "loss": 2.4573,
+      "step": 3241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.263510299127714e-06,
+      "loss": 2.5243,
+      "step": 3242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.261719959304012e-06,
+      "loss": 2.5355,
+      "step": 3243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.259929254826393e-06,
+      "loss": 2.6284,
+      "step": 3244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.258138185983563e-06,
+      "loss": 2.4836,
+      "step": 3245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.256346753064299e-06,
+      "loss": 2.6707,
+      "step": 3246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.254554956357431e-06,
+      "loss": 2.5784,
+      "step": 3247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.252762796151843e-06,
+      "loss": 2.5519,
+      "step": 3248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.250970272736489e-06,
+      "loss": 2.3794,
+      "step": 3249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.249177386400369e-06,
+      "loss": 2.5234,
+      "step": 3250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2473841374325516e-06,
+      "loss": 2.5567,
+      "step": 3251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2455905261221585e-06,
+      "loss": 2.4524,
+      "step": 3252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.243796552758371e-06,
+      "loss": 2.7544,
+      "step": 3253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.242002217630428e-06,
+      "loss": 2.5569,
+      "step": 3254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2402075210276304e-06,
+      "loss": 2.521,
+      "step": 3255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.238412463239332e-06,
+      "loss": 2.5745,
+      "step": 3256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.236617044554948e-06,
+      "loss": 2.5726,
+      "step": 3257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.234821265263952e-06,
+      "loss": 2.7212,
+      "step": 3258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2330251256558724e-06,
+      "loss": 2.5722,
+      "step": 3259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.231228626020303e-06,
+      "loss": 2.6385,
+      "step": 3260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2294317666468875e-06,
+      "loss": 2.546,
+      "step": 3261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.227634547825332e-06,
+      "loss": 2.512,
+      "step": 3262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.225836969845398e-06,
+      "loss": 2.5028,
+      "step": 3263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.224039032996909e-06,
+      "loss": 2.4587,
+      "step": 3264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.222240737569741e-06,
+      "loss": 2.6003,
+      "step": 3265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.220442083853834e-06,
+      "loss": 2.6241,
+      "step": 3266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2186430721391796e-06,
+      "loss": 2.5696,
+      "step": 3267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.216843702715831e-06,
+      "loss": 2.4726,
+      "step": 3268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.215043975873899e-06,
+      "loss": 2.593,
+      "step": 3269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.213243891903549e-06,
+      "loss": 2.6046,
+      "step": 3270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.211443451095007e-06,
+      "loss": 2.4617,
+      "step": 3271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.209642653738555e-06,
+      "loss": 2.3496,
+      "step": 3272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.207841500124535e-06,
+      "loss": 2.4933,
+      "step": 3273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.206039990543341e-06,
+      "loss": 2.5011,
+      "step": 3274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.204238125285432e-06,
+      "loss": 2.4039,
+      "step": 3275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.202435904641316e-06,
+      "loss": 2.6036,
+      "step": 3276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.200633328901565e-06,
+      "loss": 2.5698,
+      "step": 3277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.198830398356807e-06,
+      "loss": 2.5325,
+      "step": 3278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.197027113297722e-06,
+      "loss": 2.4972,
+      "step": 3279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.195223474015054e-06,
+      "loss": 2.4004,
+      "step": 3280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.193419480799601e-06,
+      "loss": 2.5698,
+      "step": 3281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.191615133942215e-06,
+      "loss": 2.5352,
+      "step": 3282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1898104337338145e-06,
+      "loss": 2.7194,
+      "step": 3283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.188005380465365e-06,
+      "loss": 2.5107,
+      "step": 3284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.186199974427892e-06,
+      "loss": 2.537,
+      "step": 3285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.18439421591248e-06,
+      "loss": 2.4788,
+      "step": 3286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1825881052102675e-06,
+      "loss": 2.7171,
+      "step": 3287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.180781642612453e-06,
+      "loss": 2.4893,
+      "step": 3288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.17897482841029e-06,
+      "loss": 2.6121,
+      "step": 3289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.177167662895087e-06,
+      "loss": 2.68,
+      "step": 3290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.175360146358212e-06,
+      "loss": 2.5491,
+      "step": 3291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.173552279091087e-06,
+      "loss": 2.5489,
+      "step": 3292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1717440613851916e-06,
+      "loss": 2.3796,
+      "step": 3293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.169935493532065e-06,
+      "loss": 2.5513,
+      "step": 3294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.168126575823299e-06,
+      "loss": 2.6148,
+      "step": 3295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.166317308550541e-06,
+      "loss": 2.4997,
+      "step": 3296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.164507692005498e-06,
+      "loss": 2.5133,
+      "step": 3297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.162697726479931e-06,
+      "loss": 2.426,
+      "step": 3298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.16088741226566e-06,
+      "loss": 2.5029,
+      "step": 3299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.159076749654559e-06,
+      "loss": 2.566,
+      "step": 3300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.157265738938557e-06,
+      "loss": 2.5352,
+      "step": 3301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1554543804096425e-06,
+      "loss": 2.5181,
+      "step": 3302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.153642674359855e-06,
+      "loss": 2.5259,
+      "step": 3303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.151830621081297e-06,
+      "loss": 2.5003,
+      "step": 3304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1500182208661214e-06,
+      "loss": 2.4168,
+      "step": 3305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.148205474006542e-06,
+      "loss": 2.5064,
+      "step": 3306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.146392380794821e-06,
+      "loss": 2.51,
+      "step": 3307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.144578941523283e-06,
+      "loss": 2.6186,
+      "step": 3308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.142765156484306e-06,
+      "loss": 2.7698,
+      "step": 3309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.140951025970324e-06,
+      "loss": 2.5073,
+      "step": 3310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.139136550273828e-06,
+      "loss": 2.5513,
+      "step": 3311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.137321729687361e-06,
+      "loss": 2.6306,
+      "step": 3312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.135506564503525e-06,
+      "loss": 2.6023,
+      "step": 3313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1336910550149775e-06,
+      "loss": 2.5843,
+      "step": 3314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.131875201514429e-06,
+      "loss": 2.5588,
+      "step": 3315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.130059004294647e-06,
+      "loss": 2.5233,
+      "step": 3316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.128242463648456e-06,
+      "loss": 2.3727,
+      "step": 3317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.126425579868733e-06,
+      "loss": 2.4892,
+      "step": 3318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.124608353248412e-06,
+      "loss": 2.5658,
+      "step": 3319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.122790784080482e-06,
+      "loss": 2.6595,
+      "step": 3320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.120972872657986e-06,
+      "loss": 2.5787,
+      "step": 3321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.119154619274027e-06,
+      "loss": 2.6978,
+      "step": 3322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.117336024221755e-06,
+      "loss": 2.4182,
+      "step": 3323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.115517087794381e-06,
+      "loss": 2.5343,
+      "step": 3324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.11369781028517e-06,
+      "loss": 2.5621,
+      "step": 3325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.111878191987441e-06,
+      "loss": 2.5746,
+      "step": 3326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.110058233194569e-06,
+      "loss": 2.5858,
+      "step": 3327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.108237934199983e-06,
+      "loss": 2.5498,
+      "step": 3328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.106417295297169e-06,
+      "loss": 2.5795,
+      "step": 3329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.104596316779663e-06,
+      "loss": 2.4783,
+      "step": 3330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.102774998941061e-06,
+      "loss": 2.6812,
+      "step": 3331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.10095334207501e-06,
+      "loss": 2.3636,
+      "step": 3332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.099131346475213e-06,
+      "loss": 2.606,
+      "step": 3333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.097309012435431e-06,
+      "loss": 2.7828,
+      "step": 3334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.095486340249473e-06,
+      "loss": 2.3757,
+      "step": 3335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.093663330211208e-06,
+      "loss": 2.3866,
+      "step": 3336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.091839982614555e-06,
+      "loss": 2.6594,
+      "step": 3337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.090016297753492e-06,
+      "loss": 2.5154,
+      "step": 3338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.088192275922046e-06,
+      "loss": 2.4907,
+      "step": 3339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.086367917414307e-06,
+      "loss": 2.5544,
+      "step": 3340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.084543222524408e-06,
+      "loss": 2.3842,
+      "step": 3341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.082718191546546e-06,
+      "loss": 2.5327,
+      "step": 3342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.080892824774965e-06,
+      "loss": 2.5057,
+      "step": 3343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.079067122503969e-06,
+      "loss": 2.7276,
+      "step": 3344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.07724108502791e-06,
+      "loss": 2.6554,
+      "step": 3345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.075414712641203e-06,
+      "loss": 2.6792,
+      "step": 3346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.073588005638308e-06,
+      "loss": 2.5875,
+      "step": 3347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.071760964313739e-06,
+      "loss": 2.5652,
+      "step": 3348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0699335889620746e-06,
+      "loss": 2.5507,
+      "step": 3349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0681058798779335e-06,
+      "loss": 2.3991,
+      "step": 3350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.066277837355999e-06,
+      "loss": 2.5742,
+      "step": 3351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.064449461691003e-06,
+      "loss": 2.5156,
+      "step": 3352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0626207531777305e-06,
+      "loss": 2.3594,
+      "step": 3353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.060791712111021e-06,
+      "loss": 2.5027,
+      "step": 3354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0589623387857705e-06,
+      "loss": 2.4226,
+      "step": 3355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.057132633496924e-06,
+      "loss": 2.5909,
+      "step": 3356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.055302596539485e-06,
+      "loss": 2.5679,
+      "step": 3357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.053472228208506e-06,
+      "loss": 2.5209,
+      "step": 3358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0516415287990945e-06,
+      "loss": 2.6139,
+      "step": 3359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.049810498606413e-06,
+      "loss": 2.622,
+      "step": 3360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.047979137925674e-06,
+      "loss": 2.65,
+      "step": 3361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.046147447052146e-06,
+      "loss": 2.574,
+      "step": 3362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.044315426281151e-06,
+      "loss": 2.4838,
+      "step": 3363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.042483075908062e-06,
+      "loss": 2.5026,
+      "step": 3364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.040650396228306e-06,
+      "loss": 2.4315,
+      "step": 3365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.038817387537364e-06,
+      "loss": 2.6086,
+      "step": 3366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.036984050130771e-06,
+      "loss": 2.5921,
+      "step": 3367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.03515038430411e-06,
+      "loss": 2.6474,
+      "step": 3368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.033316390353022e-06,
+      "loss": 2.4734,
+      "step": 3369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.031482068573201e-06,
+      "loss": 2.418,
+      "step": 3370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.029647419260388e-06,
+      "loss": 2.4721,
+      "step": 3371
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.027812442710385e-06,
+      "loss": 2.595,
+      "step": 3372
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0259771392190405e-06,
+      "loss": 2.3401,
+      "step": 3373
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.024141509082259e-06,
+      "loss": 2.529,
+      "step": 3374
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.022305552595996e-06,
+      "loss": 2.5277,
+      "step": 3375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.020469270056258e-06,
+      "loss": 2.5605,
+      "step": 3376
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.018632661759112e-06,
+      "loss": 2.5286,
+      "step": 3377
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.016795728000666e-06,
+      "loss": 2.6061,
+      "step": 3378
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.014958469077087e-06,
+      "loss": 2.5439,
+      "step": 3379
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.013120885284599e-06,
+      "loss": 2.6058,
+      "step": 3380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.011282976919467e-06,
+      "loss": 2.494,
+      "step": 3381
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0094447442780156e-06,
+      "loss": 2.7405,
+      "step": 3382
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.007606187656623e-06,
+      "loss": 2.4808,
+      "step": 3383
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.005767307351714e-06,
+      "loss": 2.56,
+      "step": 3384
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.003928103659771e-06,
+      "loss": 2.4512,
+      "step": 3385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.002088576877325e-06,
+      "loss": 2.3884,
+      "step": 3386
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.00024872730096e-06,
+      "loss": 2.5553,
+      "step": 3387
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9984085552273136e-06,
+      "loss": 2.4752,
+      "step": 3388
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.996568060953075e-06,
+      "loss": 2.5007,
+      "step": 3389
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.994727244774981e-06,
+      "loss": 2.438,
+      "step": 3390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.992886106989827e-06,
+      "loss": 2.7247,
+      "step": 3391
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.991044647894456e-06,
+      "loss": 2.5372,
+      "step": 3392
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.989202867785762e-06,
+      "loss": 2.5679,
+      "step": 3393
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.987360766960697e-06,
+      "loss": 2.7698,
+      "step": 3394
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.985518345716256e-06,
+      "loss": 2.3488,
+      "step": 3395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.983675604349492e-06,
+      "loss": 2.544,
+      "step": 3396
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.981832543157507e-06,
+      "loss": 2.4098,
+      "step": 3397
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9799891624374565e-06,
+      "loss": 2.4566,
+      "step": 3398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.978145462486544e-06,
+      "loss": 2.5263,
+      "step": 3399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9763014436020285e-06,
+      "loss": 2.5483,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9744571060812185e-06,
+      "loss": 2.5378,
+      "step": 3401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.972612450221471e-06,
+      "loss": 2.6029,
+      "step": 3402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9707674763202e-06,
+      "loss": 2.471,
+      "step": 3403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.968922184674868e-06,
+      "loss": 2.489,
+      "step": 3404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.967076575582989e-06,
+      "loss": 2.4791,
+      "step": 3405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.965230649342127e-06,
+      "loss": 2.4615,
+      "step": 3406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.963384406249898e-06,
+      "loss": 2.5226,
+      "step": 3407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.96153784660397e-06,
+      "loss": 2.4778,
+      "step": 3408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.959690970702059e-06,
+      "loss": 2.5375,
+      "step": 3409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.957843778841937e-06,
+      "loss": 2.6204,
+      "step": 3410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.955996271321423e-06,
+      "loss": 2.4206,
+      "step": 3411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.954148448438389e-06,
+      "loss": 2.5351,
+      "step": 3412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9523003104907565e-06,
+      "loss": 2.6069,
+      "step": 3413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.950451857776495e-06,
+      "loss": 2.7543,
+      "step": 3414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.948603090593632e-06,
+      "loss": 2.5575,
+      "step": 3415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.946754009240241e-06,
+      "loss": 2.4693,
+      "step": 3416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.944904614014445e-06,
+      "loss": 2.5934,
+      "step": 3417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.943054905214421e-06,
+      "loss": 2.5002,
+      "step": 3418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.941204883138395e-06,
+      "loss": 2.6119,
+      "step": 3419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9393545480846405e-06,
+      "loss": 2.4744,
+      "step": 3420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9375039003514876e-06,
+      "loss": 2.4601,
+      "step": 3421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.935652940237313e-06,
+      "loss": 2.6223,
+      "step": 3422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.933801668040544e-06,
+      "loss": 2.6702,
+      "step": 3423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9319500840596596e-06,
+      "loss": 2.4057,
+      "step": 3424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.930098188593185e-06,
+      "loss": 2.7331,
+      "step": 3425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9282459819397016e-06,
+      "loss": 2.4977,
+      "step": 3426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.926393464397838e-06,
+      "loss": 2.5082,
+      "step": 3427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.924540636266272e-06,
+      "loss": 2.4777,
+      "step": 3428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.922687497843734e-06,
+      "loss": 2.5447,
+      "step": 3429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.920834049429002e-06,
+      "loss": 2.4544,
+      "step": 3430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.918980291320902e-06,
+      "loss": 2.6058,
+      "step": 3431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.917126223818315e-06,
+      "loss": 2.4188,
+      "step": 3432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9152718472201715e-06,
+      "loss": 2.5214,
+      "step": 3433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.913417161825449e-06,
+      "loss": 2.438,
+      "step": 3434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.911562167933176e-06,
+      "loss": 2.537,
+      "step": 3435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.909706865842429e-06,
+      "loss": 2.5084,
+      "step": 3436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.907851255852336e-06,
+      "loss": 2.5394,
+      "step": 3437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.905995338262075e-06,
+      "loss": 2.627,
+      "step": 3438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.904139113370875e-06,
+      "loss": 2.4779,
+      "step": 3439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.902282581478009e-06,
+      "loss": 2.5839,
+      "step": 3440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9004257428828044e-06,
+      "loss": 2.6905,
+      "step": 3441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8985685978846385e-06,
+      "loss": 2.436,
+      "step": 3442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.896711146782933e-06,
+      "loss": 2.4503,
+      "step": 3443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.894853389877163e-06,
+      "loss": 2.7182,
+      "step": 3444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.892995327466854e-06,
+      "loss": 2.394,
+      "step": 3445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.891136959851576e-06,
+      "loss": 2.3424,
+      "step": 3446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.889278287330955e-06,
+      "loss": 2.4091,
+      "step": 3447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.887419310204657e-06,
+      "loss": 2.4229,
+      "step": 3448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.885560028772406e-06,
+      "loss": 2.56,
+      "step": 3449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.88370044333397e-06,
+      "loss": 2.4882,
+      "step": 3450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.881840554189167e-06,
+      "loss": 2.6487,
+      "step": 3451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.879980361637865e-06,
+      "loss": 2.4626,
+      "step": 3452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.878119865979982e-06,
+      "loss": 2.5608,
+      "step": 3453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.876259067515478e-06,
+      "loss": 2.6365,
+      "step": 3454
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.874397966544371e-06,
+      "loss": 2.6155,
+      "step": 3455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.872536563366724e-06,
+      "loss": 2.5183,
+      "step": 3456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.870674858282647e-06,
+      "loss": 2.5987,
+      "step": 3457
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.868812851592299e-06,
+      "loss": 2.5776,
+      "step": 3458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.866950543595891e-06,
+      "loss": 2.3885,
+      "step": 3459
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.86508793459368e-06,
+      "loss": 2.4621,
+      "step": 3460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.863225024885968e-06,
+      "loss": 2.4538,
+      "step": 3461
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.861361814773114e-06,
+      "loss": 2.6471,
+      "step": 3462
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.85949830455552e-06,
+      "loss": 2.5987,
+      "step": 3463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.857634494533636e-06,
+      "loss": 2.4264,
+      "step": 3464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8557703850079605e-06,
+      "loss": 2.577,
+      "step": 3465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8539059762790425e-06,
+      "loss": 2.598,
+      "step": 3466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.852041268647478e-06,
+      "loss": 2.5505,
+      "step": 3467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8501762624139125e-06,
+      "loss": 2.539,
+      "step": 3468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.848310957879036e-06,
+      "loss": 2.6697,
+      "step": 3469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.846445355343591e-06,
+      "loss": 2.5349,
+      "step": 3470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.844579455108362e-06,
+      "loss": 2.4591,
+      "step": 3471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.842713257474191e-06,
+      "loss": 2.505,
+      "step": 3472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8408467627419575e-06,
+      "loss": 2.5075,
+      "step": 3473
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.838979971212598e-06,
+      "loss": 2.7163,
+      "step": 3474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.837112883187089e-06,
+      "loss": 2.4738,
+      "step": 3475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.835245498966461e-06,
+      "loss": 2.4947,
+      "step": 3476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.833377818851789e-06,
+      "loss": 2.5098,
+      "step": 3477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8315098431441955e-06,
+      "loss": 2.4884,
+      "step": 3478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.829641572144854e-06,
+      "loss": 2.5255,
+      "step": 3479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8277730061549806e-06,
+      "loss": 2.6656,
+      "step": 3480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.825904145475842e-06,
+      "loss": 2.5295,
+      "step": 3481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.824034990408753e-06,
+      "loss": 2.4591,
+      "step": 3482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8221655412550746e-06,
+      "loss": 2.5537,
+      "step": 3483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.820295798316214e-06,
+      "loss": 2.5295,
+      "step": 3484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.81842576189363e-06,
+      "loss": 2.4957,
+      "step": 3485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.816555432288824e-06,
+      "loss": 2.4792,
+      "step": 3486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.814684809803347e-06,
+      "loss": 2.7786,
+      "step": 3487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8128138947387966e-06,
+      "loss": 2.5126,
+      "step": 3488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.810942687396818e-06,
+      "loss": 2.5993,
+      "step": 3489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8090711880791025e-06,
+      "loss": 2.4943,
+      "step": 3490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.807199397087391e-06,
+      "loss": 2.6227,
+      "step": 3491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.805327314723469e-06,
+      "loss": 2.4049,
+      "step": 3492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.803454941289168e-06,
+      "loss": 2.4009,
+      "step": 3493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.801582277086371e-06,
+      "loss": 2.6056,
+      "step": 3494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7997093224170045e-06,
+      "loss": 2.3961,
+      "step": 3495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.79783607758304e-06,
+      "loss": 2.4373,
+      "step": 3496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.795962542886501e-06,
+      "loss": 2.5217,
+      "step": 3497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.794088718629451e-06,
+      "loss": 2.4587,
+      "step": 3498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.792214605114009e-06,
+      "loss": 2.3403,
+      "step": 3499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.790340202642333e-06,
+      "loss": 2.6014,
+      "step": 3500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7884655115166285e-06,
+      "loss": 2.5895,
+      "step": 3501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.786590532039152e-06,
+      "loss": 2.4506,
+      "step": 3502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.784715264512205e-06,
+      "loss": 2.7009,
+      "step": 3503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.78283970923813e-06,
+      "loss": 2.5024,
+      "step": 3504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7809638665193235e-06,
+      "loss": 2.6888,
+      "step": 3505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7790877366582255e-06,
+      "loss": 2.4187,
+      "step": 3506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.777211319957318e-06,
+      "loss": 2.4443,
+      "step": 3507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.775334616719136e-06,
+      "loss": 2.5614,
+      "step": 3508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.773457627246256e-06,
+      "loss": 2.5991,
+      "step": 3509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.771580351841304e-06,
+      "loss": 2.5444,
+      "step": 3510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.769702790806949e-06,
+      "loss": 2.5674,
+      "step": 3511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.76782494444591e-06,
+      "loss": 2.5935,
+      "step": 3512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.765946813060946e-06,
+      "loss": 2.5608,
+      "step": 3513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.764068396954868e-06,
+      "loss": 2.7615,
+      "step": 3514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.76218969643053e-06,
+      "loss": 2.612,
+      "step": 3515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.760310711790831e-06,
+      "loss": 2.5248,
+      "step": 3516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.75843144333872e-06,
+      "loss": 2.4135,
+      "step": 3517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.756551891377185e-06,
+      "loss": 2.525,
+      "step": 3518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7546720562092664e-06,
+      "loss": 2.4553,
+      "step": 3519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.752791938138046e-06,
+      "loss": 2.5535,
+      "step": 3520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7509115374666515e-06,
+      "loss": 2.4303,
+      "step": 3521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.74903085449826e-06,
+      "loss": 2.5413,
+      "step": 3522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.747149889536089e-06,
+      "loss": 2.5923,
+      "step": 3523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7452686428834045e-06,
+      "loss": 2.6194,
+      "step": 3524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.743387114843518e-06,
+      "loss": 2.4902,
+      "step": 3525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.741505305719785e-06,
+      "loss": 2.511,
+      "step": 3526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.739623215815607e-06,
+      "loss": 2.5482,
+      "step": 3527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.737740845434432e-06,
+      "loss": 2.4535,
+      "step": 3528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7358581948797495e-06,
+      "loss": 2.4784,
+      "step": 3529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.733975264455097e-06,
+      "loss": 2.3545,
+      "step": 3530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7320920544640585e-06,
+      "loss": 2.4496,
+      "step": 3531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.73020856521026e-06,
+      "loss": 2.5349,
+      "step": 3532
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7283247969973756e-06,
+      "loss": 2.5711,
+      "step": 3533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7264407501291195e-06,
+      "loss": 2.5519,
+      "step": 3534
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.724556424909256e-06,
+      "loss": 2.5586,
+      "step": 3535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.722671821641591e-06,
+      "loss": 2.4885,
+      "step": 3536
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.720786940629977e-06,
+      "loss": 2.6116,
+      "step": 3537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.718901782178312e-06,
+      "loss": 2.7089,
+      "step": 3538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.717016346590536e-06,
+      "loss": 2.5655,
+      "step": 3539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.715130634170636e-06,
+      "loss": 2.5051,
+      "step": 3540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.7132446452226405e-06,
+      "loss": 2.5019,
+      "step": 3541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.711358380050628e-06,
+      "loss": 2.6083,
+      "step": 3542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.709471838958715e-06,
+      "loss": 2.7503,
+      "step": 3543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.707585022251068e-06,
+      "loss": 2.6002,
+      "step": 3544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.705697930231898e-06,
+      "loss": 2.4639,
+      "step": 3545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.703810563205454e-06,
+      "loss": 2.5349,
+      "step": 3546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.7019229214760365e-06,
+      "loss": 2.6932,
+      "step": 3547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.700035005347983e-06,
+      "loss": 2.3157,
+      "step": 3548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.698146815125685e-06,
+      "loss": 2.4562,
+      "step": 3549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.696258351113569e-06,
+      "loss": 2.5796,
+      "step": 3550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.694369613616113e-06,
+      "loss": 2.6003,
+      "step": 3551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.692480602937832e-06,
+      "loss": 2.4744,
+      "step": 3552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.690591319383289e-06,
+      "loss": 2.6278,
+      "step": 3553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.68870176325709e-06,
+      "loss": 2.5808,
+      "step": 3554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.68681193486389e-06,
+      "loss": 2.5646,
+      "step": 3555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6849218345083785e-06,
+      "loss": 2.4373,
+      "step": 3556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.683031462495296e-06,
+      "loss": 2.4639,
+      "step": 3557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.681140819129424e-06,
+      "loss": 2.5158,
+      "step": 3558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.679249904715588e-06,
+      "loss": 2.4793,
+      "step": 3559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.677358719558655e-06,
+      "loss": 2.6138,
+      "step": 3560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.675467263963544e-06,
+      "loss": 2.7053,
+      "step": 3561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6735755382352074e-06,
+      "loss": 2.5928,
+      "step": 3562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.671683542678646e-06,
+      "loss": 2.589,
+      "step": 3563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6697912775989045e-06,
+      "loss": 2.73,
+      "step": 3564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.66789874330107e-06,
+      "loss": 2.5841,
+      "step": 3565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.666005940090271e-06,
+      "loss": 2.5886,
+      "step": 3566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.664112868271686e-06,
+      "loss": 2.4859,
+      "step": 3567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.662219528150529e-06,
+      "loss": 2.5956,
+      "step": 3568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.660325920032061e-06,
+      "loss": 2.5993,
+      "step": 3569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.658432044221584e-06,
+      "loss": 2.3985,
+      "step": 3570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.656537901024448e-06,
+      "loss": 2.4631,
+      "step": 3571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.654643490746042e-06,
+      "loss": 2.4616,
+      "step": 3572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6527488136917995e-06,
+      "loss": 2.6352,
+      "step": 3573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.650853870167196e-06,
+      "loss": 2.5351,
+      "step": 3574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.648958660477751e-06,
+      "loss": 2.5779,
+      "step": 3575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.647063184929025e-06,
+      "loss": 2.5388,
+      "step": 3576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.645167443826624e-06,
+      "loss": 2.6739,
+      "step": 3577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.643271437476196e-06,
+      "loss": 2.5951,
+      "step": 3578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.641375166183432e-06,
+      "loss": 2.4106,
+      "step": 3579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6394786302540645e-06,
+      "loss": 2.6302,
+      "step": 3580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6375818299938685e-06,
+      "loss": 2.5257,
+      "step": 3581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.635684765708662e-06,
+      "loss": 2.4935,
+      "step": 3582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.633787437704308e-06,
+      "loss": 2.7234,
+      "step": 3583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6318898462867086e-06,
+      "loss": 2.6244,
+      "step": 3584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.629991991761811e-06,
+      "loss": 2.4376,
+      "step": 3585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6280938744356025e-06,
+      "loss": 2.6211,
+      "step": 3586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6261954946141134e-06,
+      "loss": 2.616,
+      "step": 3587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.624296852603419e-06,
+      "loss": 2.6237,
+      "step": 3588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.622397948709632e-06,
+      "loss": 2.4744,
+      "step": 3589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6204987832389115e-06,
+      "loss": 2.3844,
+      "step": 3590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6185993564974585e-06,
+      "loss": 2.4213,
+      "step": 3591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.616699668791512e-06,
+      "loss": 2.5316,
+      "step": 3592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.614799720427357e-06,
+      "loss": 2.4745,
+      "step": 3593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.612899511711322e-06,
+      "loss": 2.4829,
+      "step": 3594
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.610999042949771e-06,
+      "loss": 2.505,
+      "step": 3595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.609098314449116e-06,
+      "loss": 2.676,
+      "step": 3596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.607197326515808e-06,
+      "loss": 2.7384,
+      "step": 3597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.605296079456342e-06,
+      "loss": 2.532,
+      "step": 3598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.603394573577253e-06,
+      "loss": 2.3399,
+      "step": 3599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.601492809185118e-06,
+      "loss": 2.5882,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.599590786586554e-06,
+      "loss": 2.5883,
+      "step": 3601
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.597688506088224e-06,
+      "loss": 2.4541,
+      "step": 3602
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.595785967996828e-06,
+      "loss": 2.5639,
+      "step": 3603
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.593883172619111e-06,
+      "loss": 2.4487,
+      "step": 3604
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.591980120261857e-06,
+      "loss": 2.6249,
+      "step": 3605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5900768112318936e-06,
+      "loss": 2.6483,
+      "step": 3606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.588173245836087e-06,
+      "loss": 2.5078,
+      "step": 3607
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.586269424381349e-06,
+      "loss": 2.6789,
+      "step": 3608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.584365347174626e-06,
+      "loss": 2.6752,
+      "step": 3609
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.582461014522913e-06,
+      "loss": 2.6031,
+      "step": 3610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.580556426733242e-06,
+      "loss": 2.4531,
+      "step": 3611
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.578651584112687e-06,
+      "loss": 2.5375,
+      "step": 3612
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.576746486968363e-06,
+      "loss": 2.5404,
+      "step": 3613
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.574841135607426e-06,
+      "loss": 2.6264,
+      "step": 3614
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.572935530337073e-06,
+      "loss": 2.5931,
+      "step": 3615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.571029671464542e-06,
+      "loss": 2.6492,
+      "step": 3616
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.569123559297113e-06,
+      "loss": 2.5023,
+      "step": 3617
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.567217194142104e-06,
+      "loss": 2.4458,
+      "step": 3618
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.565310576306877e-06,
+      "loss": 2.4768,
+      "step": 3619
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.563403706098833e-06,
+      "loss": 2.531,
+      "step": 3620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.561496583825413e-06,
+      "loss": 2.3396,
+      "step": 3621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.559589209794102e-06,
+      "loss": 2.5596,
+      "step": 3622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5576815843124205e-06,
+      "loss": 2.5697,
+      "step": 3623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5557737076879335e-06,
+      "loss": 2.6553,
+      "step": 3624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.553865580228246e-06,
+      "loss": 2.6455,
+      "step": 3625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.551957202241001e-06,
+      "loss": 2.4026,
+      "step": 3626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.550048574033885e-06,
+      "loss": 2.623,
+      "step": 3627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5481396959146225e-06,
+      "loss": 2.5947,
+      "step": 3628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.54623056819098e-06,
+      "loss": 2.5972,
+      "step": 3629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.544321191170764e-06,
+      "loss": 2.4611,
+      "step": 3630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5424115651618195e-06,
+      "loss": 2.6117,
+      "step": 3631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.540501690472032e-06,
+      "loss": 2.4188,
+      "step": 3632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5385915674093295e-06,
+      "loss": 2.6753,
+      "step": 3633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.536681196281678e-06,
+      "loss": 2.5817,
+      "step": 3634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.534770577397084e-06,
+      "loss": 2.5957,
+      "step": 3635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.532859711063594e-06,
+      "loss": 2.622,
+      "step": 3636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.530948597589293e-06,
+      "loss": 2.6513,
+      "step": 3637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.529037237282309e-06,
+      "loss": 2.4837,
+      "step": 3638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.527125630450808e-06,
+      "loss": 2.3508,
+      "step": 3639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5252137774029955e-06,
+      "loss": 2.5426,
+      "step": 3640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.523301678447113e-06,
+      "loss": 2.7029,
+      "step": 3641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5213893338914515e-06,
+      "loss": 2.4754,
+      "step": 3642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5194767440443315e-06,
+      "loss": 2.5594,
+      "step": 3643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.517563909214119e-06,
+      "loss": 2.6044,
+      "step": 3644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.515650829709217e-06,
+      "loss": 2.4836,
+      "step": 3645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.513737505838069e-06,
+      "loss": 2.4863,
+      "step": 3646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.511823937909158e-06,
+      "loss": 2.4536,
+      "step": 3647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.509910126231003e-06,
+      "loss": 2.6277,
+      "step": 3648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.507996071112168e-06,
+      "loss": 2.6093,
+      "step": 3649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5060817728612544e-06,
+      "loss": 2.6891,
+      "step": 3650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5041672317868984e-06,
+      "loss": 2.5324,
+      "step": 3651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.502252448197782e-06,
+      "loss": 2.625,
+      "step": 3652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.500337422402619e-06,
+      "loss": 2.6434,
+      "step": 3653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.498422154710171e-06,
+      "loss": 2.5062,
+      "step": 3654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.49650664542923e-06,
+      "loss": 2.7233,
+      "step": 3655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4945908948686355e-06,
+      "loss": 2.7268,
+      "step": 3656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.492674903337257e-06,
+      "loss": 2.5823,
+      "step": 3657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.490758671144006e-06,
+      "loss": 2.4805,
+      "step": 3658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.48884219859784e-06,
+      "loss": 2.5523,
+      "step": 3659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.486925486007743e-06,
+      "loss": 2.5479,
+      "step": 3660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.485008533682748e-06,
+      "loss": 2.5667,
+      "step": 3661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4830913419319205e-06,
+      "loss": 2.3752,
+      "step": 3662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.481173911064367e-06,
+      "loss": 2.597,
+      "step": 3663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.479256241389231e-06,
+      "loss": 2.6501,
+      "step": 3664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4773383332156985e-06,
+      "loss": 2.5663,
+      "step": 3665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.47542018685299e-06,
+      "loss": 2.6456,
+      "step": 3666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.473501802610364e-06,
+      "loss": 2.5619,
+      "step": 3667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.471583180797121e-06,
+      "loss": 2.6434,
+      "step": 3668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.469664321722595e-06,
+      "loss": 2.6267,
+      "step": 3669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.467745225696164e-06,
+      "loss": 2.4389,
+      "step": 3670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.465825893027238e-06,
+      "loss": 2.517,
+      "step": 3671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.463906324025272e-06,
+      "loss": 2.537,
+      "step": 3672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4619865189997536e-06,
+      "loss": 2.4526,
+      "step": 3673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.460066478260209e-06,
+      "loss": 2.6329,
+      "step": 3674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.458146202116206e-06,
+      "loss": 2.5594,
+      "step": 3675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.456225690877345e-06,
+      "loss": 2.5028,
+      "step": 3676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.454304944853269e-06,
+      "loss": 2.4956,
+      "step": 3677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.452383964353659e-06,
+      "loss": 2.4796,
+      "step": 3678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4504627496882284e-06,
+      "loss": 2.7503,
+      "step": 3679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.448541301166737e-06,
+      "loss": 2.4559,
+      "step": 3680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.44661961909897e-06,
+      "loss": 2.5203,
+      "step": 3681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4446977037947635e-06,
+      "loss": 2.5451,
+      "step": 3682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.442775555563984e-06,
+      "loss": 2.4295,
+      "step": 3683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.440853174716535e-06,
+      "loss": 2.7933,
+      "step": 3684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.438930561562361e-06,
+      "loss": 2.4571,
+      "step": 3685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.437007716411441e-06,
+      "loss": 2.5555,
+      "step": 3686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.435084639573792e-06,
+      "loss": 2.6315,
+      "step": 3687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.43316133135947e-06,
+      "loss": 2.6183,
+      "step": 3688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4312377920785685e-06,
+      "loss": 2.5398,
+      "step": 3689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.429314022041216e-06,
+      "loss": 2.3666,
+      "step": 3690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.427390021557579e-06,
+      "loss": 2.4387,
+      "step": 3691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4254657909378615e-06,
+      "loss": 2.5695,
+      "step": 3692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4235413304923025e-06,
+      "loss": 2.4248,
+      "step": 3693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.421616640531183e-06,
+      "loss": 2.633,
+      "step": 3694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.419691721364817e-06,
+      "loss": 2.621,
+      "step": 3695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.417766573303557e-06,
+      "loss": 2.5487,
+      "step": 3696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.415841196657792e-06,
+      "loss": 2.4969,
+      "step": 3697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4139155917379445e-06,
+      "loss": 2.5071,
+      "step": 3698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.411989758854482e-06,
+      "loss": 2.7043,
+      "step": 3699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.410063698317901e-06,
+      "loss": 2.6105,
+      "step": 3700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.408137410438738e-06,
+      "loss": 2.5549,
+      "step": 3701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.406210895527567e-06,
+      "loss": 2.629,
+      "step": 3702
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.404284153894995e-06,
+      "loss": 2.2953,
+      "step": 3703
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.402357185851669e-06,
+      "loss": 2.3908,
+      "step": 3704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.40042999170827e-06,
+      "loss": 2.5741,
+      "step": 3705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.39850257177552e-06,
+      "loss": 2.601,
+      "step": 3706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.396574926364172e-06,
+      "loss": 2.5252,
+      "step": 3707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.394647055785017e-06,
+      "loss": 2.5862,
+      "step": 3708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3927189603488835e-06,
+      "loss": 2.6775,
+      "step": 3709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3907906403666355e-06,
+      "loss": 2.6005,
+      "step": 3710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.388862096149173e-06,
+      "loss": 2.4173,
+      "step": 3711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.386933328007434e-06,
+      "loss": 2.5111,
+      "step": 3712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.385004336252392e-06,
+      "loss": 2.5264,
+      "step": 3713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.383075121195051e-06,
+      "loss": 2.5429,
+      "step": 3714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.381145683146459e-06,
+      "loss": 2.5776,
+      "step": 3715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.379216022417695e-06,
+      "loss": 2.4665,
+      "step": 3716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.377286139319878e-06,
+      "loss": 2.3581,
+      "step": 3717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.375356034164159e-06,
+      "loss": 2.5014,
+      "step": 3718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.373425707261725e-06,
+      "loss": 2.7837,
+      "step": 3719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.371495158923801e-06,
+      "loss": 2.7078,
+      "step": 3720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.369564389461644e-06,
+      "loss": 2.4752,
+      "step": 3721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.367633399186554e-06,
+      "loss": 2.4045,
+      "step": 3722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.365702188409859e-06,
+      "loss": 2.5954,
+      "step": 3723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.363770757442927e-06,
+      "loss": 2.5643,
+      "step": 3724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3618391065971565e-06,
+      "loss": 2.4706,
+      "step": 3725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.359907236183988e-06,
+      "loss": 2.466,
+      "step": 3726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.357975146514893e-06,
+      "loss": 2.6247,
+      "step": 3727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3560428379013795e-06,
+      "loss": 2.5249,
+      "step": 3728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.354110310654992e-06,
+      "loss": 2.4566,
+      "step": 3729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.352177565087308e-06,
+      "loss": 2.4077,
+      "step": 3730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.350244601509943e-06,
+      "loss": 2.7088,
+      "step": 3731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.348311420234542e-06,
+      "loss": 2.7253,
+      "step": 3732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.346378021572795e-06,
+      "loss": 2.5939,
+      "step": 3733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.344444405836414e-06,
+      "loss": 2.8125,
+      "step": 3734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.34251057333716e-06,
+      "loss": 2.4783,
+      "step": 3735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.340576524386819e-06,
+      "loss": 2.4292,
+      "step": 3736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.338642259297214e-06,
+      "loss": 2.5913,
+      "step": 3737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.336707778380207e-06,
+      "loss": 2.4744,
+      "step": 3738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3347730819476875e-06,
+      "loss": 2.5118,
+      "step": 3739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.332838170311586e-06,
+      "loss": 2.6141,
+      "step": 3740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.330903043783867e-06,
+      "loss": 2.5305,
+      "step": 3741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.328967702676528e-06,
+      "loss": 2.4725,
+      "step": 3742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.327032147301596e-06,
+      "loss": 2.5415,
+      "step": 3743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.325096377971145e-06,
+      "loss": 2.4754,
+      "step": 3744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.323160394997273e-06,
+      "loss": 2.5013,
+      "step": 3745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.321224198692115e-06,
+      "loss": 2.5,
+      "step": 3746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.319287789367844e-06,
+      "loss": 2.5858,
+      "step": 3747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.31735116733666e-06,
+      "loss": 2.6736,
+      "step": 3748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.315414332910807e-06,
+      "loss": 2.6039,
+      "step": 3749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.313477286402557e-06,
+      "loss": 2.2934,
+      "step": 3750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.311540028124214e-06,
+      "loss": 2.4773,
+      "step": 3751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.309602558388122e-06,
+      "loss": 2.5347,
+      "step": 3752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.307664877506657e-06,
+      "loss": 2.5361,
+      "step": 3753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.305726985792227e-06,
+      "loss": 2.6407,
+      "step": 3754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.303788883557276e-06,
+      "loss": 2.5262,
+      "step": 3755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.301850571114282e-06,
+      "loss": 2.5746,
+      "step": 3756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2999120487757535e-06,
+      "loss": 2.5742,
+      "step": 3757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.297973316854241e-06,
+      "loss": 2.7387,
+      "step": 3758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.296034375662321e-06,
+      "loss": 2.5202,
+      "step": 3759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.294095225512604e-06,
+      "loss": 2.5415,
+      "step": 3760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.29215586671774e-06,
+      "loss": 2.7613,
+      "step": 3761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.290216299590407e-06,
+      "loss": 2.5604,
+      "step": 3762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.288276524443319e-06,
+      "loss": 2.4239,
+      "step": 3763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.286336541589224e-06,
+      "loss": 2.5269,
+      "step": 3764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2843963513409015e-06,
+      "loss": 2.5552,
+      "step": 3765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.282455954011166e-06,
+      "loss": 2.4145,
+      "step": 3766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.280515349912866e-06,
+      "loss": 2.4981,
+      "step": 3767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2785745393588815e-06,
+      "loss": 2.6869,
+      "step": 3768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.276633522662127e-06,
+      "loss": 2.6129,
+      "step": 3769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.27469230013555e-06,
+      "loss": 2.7346,
+      "step": 3770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.272750872092128e-06,
+      "loss": 2.6123,
+      "step": 3771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.270809238844881e-06,
+      "loss": 2.7158,
+      "step": 3772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.26886740070685e-06,
+      "loss": 2.643,
+      "step": 3773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.266925357991118e-06,
+      "loss": 2.61,
+      "step": 3774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.264983111010796e-06,
+      "loss": 2.6215,
+      "step": 3775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.263040660079031e-06,
+      "loss": 2.4995,
+      "step": 3776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.261098005509002e-06,
+      "loss": 2.513,
+      "step": 3777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.25915514761392e-06,
+      "loss": 2.5109,
+      "step": 3778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2572120867070286e-06,
+      "loss": 2.6455,
+      "step": 3779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.255268823101604e-06,
+      "loss": 2.3787,
+      "step": 3780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.253325357110957e-06,
+      "loss": 2.3545,
+      "step": 3781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.25138168904843e-06,
+      "loss": 2.5868,
+      "step": 3782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2494378192273995e-06,
+      "loss": 2.4098,
+      "step": 3783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.24749374796127e-06,
+      "loss": 2.5499,
+      "step": 3784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2455494755634825e-06,
+      "loss": 2.5285,
+      "step": 3785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.243605002347508e-06,
+      "loss": 2.5452,
+      "step": 3786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.241660328626853e-06,
+      "loss": 2.3822,
+      "step": 3787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.239715454715054e-06,
+      "loss": 2.6895,
+      "step": 3788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.237770380925682e-06,
+      "loss": 2.7445,
+      "step": 3789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.235825107572335e-06,
+      "loss": 2.518,
+      "step": 3790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.233879634968649e-06,
+      "loss": 2.6076,
+      "step": 3791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.231933963428288e-06,
+      "loss": 2.4786,
+      "step": 3792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.2299880932649515e-06,
+      "loss": 2.6228,
+      "step": 3793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.22804202479237e-06,
+      "loss": 2.5041,
+      "step": 3794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.226095758324304e-06,
+      "loss": 2.4939,
+      "step": 3795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.224149294174549e-06,
+      "loss": 2.4139,
+      "step": 3796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.222202632656927e-06,
+      "loss": 2.3342,
+      "step": 3797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.220255774085297e-06,
+      "loss": 2.4234,
+      "step": 3798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.218308718773551e-06,
+      "loss": 2.7604,
+      "step": 3799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.216361467035608e-06,
+      "loss": 2.5816,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.214414019185421e-06,
+      "loss": 2.58,
+      "step": 3801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.212466375536974e-06,
+      "loss": 2.5587,
+      "step": 3802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.210518536404282e-06,
+      "loss": 2.4652,
+      "step": 3803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.208570502101393e-06,
+      "loss": 2.6887,
+      "step": 3804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.206622272942386e-06,
+      "loss": 2.5344,
+      "step": 3805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.204673849241372e-06,
+      "loss": 2.56,
+      "step": 3806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.202725231312492e-06,
+      "loss": 2.4485,
+      "step": 3807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.200776419469918e-06,
+      "loss": 2.675,
+      "step": 3808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.198827414027856e-06,
+      "loss": 2.5281,
+      "step": 3809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.196878215300538e-06,
+      "loss": 2.598,
+      "step": 3810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.194928823602235e-06,
+      "loss": 2.6079,
+      "step": 3811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.192979239247243e-06,
+      "loss": 2.6066,
+      "step": 3812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1910294625498904e-06,
+      "loss": 2.5422,
+      "step": 3813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1890794938245355e-06,
+      "loss": 2.7632,
+      "step": 3814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.187129333385571e-06,
+      "loss": 2.5497,
+      "step": 3815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1851789815474175e-06,
+      "loss": 2.5933,
+      "step": 3816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.183228438624528e-06,
+      "loss": 2.3955,
+      "step": 3817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.181277704931386e-06,
+      "loss": 2.4543,
+      "step": 3818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1793267807825056e-06,
+      "loss": 2.5337,
+      "step": 3819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.177375666492431e-06,
+      "loss": 2.5746,
+      "step": 3820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.175424362375737e-06,
+      "loss": 2.4997,
+      "step": 3821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.17347286874703e-06,
+      "loss": 2.551,
+      "step": 3822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.171521185920948e-06,
+      "loss": 2.5846,
+      "step": 3823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.169569314212157e-06,
+      "loss": 2.5286,
+      "step": 3824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1676172539353555e-06,
+      "loss": 2.4873,
+      "step": 3825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.165665005405268e-06,
+      "loss": 2.6157,
+      "step": 3826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.163712568936657e-06,
+      "loss": 2.6965,
+      "step": 3827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.161759944844308e-06,
+      "loss": 2.6114,
+      "step": 3828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.159807133443042e-06,
+      "loss": 2.6759,
+      "step": 3829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1578541350477076e-06,
+      "loss": 2.6991,
+      "step": 3830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.155900949973184e-06,
+      "loss": 2.5984,
+      "step": 3831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.153947578534378e-06,
+      "loss": 2.4548,
+      "step": 3832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.151994021046233e-06,
+      "loss": 2.7431,
+      "step": 3833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.150040277823715e-06,
+      "loss": 2.4875,
+      "step": 3834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.148086349181824e-06,
+      "loss": 2.5336,
+      "step": 3835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.146132235435591e-06,
+      "loss": 2.5493,
+      "step": 3836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.144177936900073e-06,
+      "loss": 2.4407,
+      "step": 3837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.14222345389036e-06,
+      "loss": 2.5409,
+      "step": 3838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.140268786721571e-06,
+      "loss": 2.6212,
+      "step": 3839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.138313935708852e-06,
+      "loss": 2.7249,
+      "step": 3840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1363589011673815e-06,
+      "loss": 2.6356,
+      "step": 3841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1344036834123695e-06,
+      "loss": 2.5843,
+      "step": 3842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1324482827590495e-06,
+      "loss": 2.5712,
+      "step": 3843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1304926995226895e-06,
+      "loss": 2.5164,
+      "step": 3844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.128536934018587e-06,
+      "loss": 2.6558,
+      "step": 3845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.126580986562065e-06,
+      "loss": 2.5482,
+      "step": 3846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.12462485746848e-06,
+      "loss": 2.4661,
+      "step": 3847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1226685470532125e-06,
+      "loss": 2.4729,
+      "step": 3848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.12071205563168e-06,
+      "loss": 2.467,
+      "step": 3849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.118755383519323e-06,
+      "loss": 2.3403,
+      "step": 3850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.116798531031612e-06,
+      "loss": 2.4577,
+      "step": 3851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.114841498484049e-06,
+      "loss": 2.5683,
+      "step": 3852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.112884286192163e-06,
+      "loss": 2.5354,
+      "step": 3853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1109268944715125e-06,
+      "loss": 2.4382,
+      "step": 3854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1089693236376854e-06,
+      "loss": 2.3584,
+      "step": 3855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1070115740062985e-06,
+      "loss": 2.6541,
+      "step": 3856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.105053645892995e-06,
+      "loss": 2.5337,
+      "step": 3857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.103095539613452e-06,
+      "loss": 2.692,
+      "step": 3858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1011372554833696e-06,
+      "loss": 2.5418,
+      "step": 3859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.099178793818479e-06,
+      "loss": 2.7404,
+      "step": 3860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.097220154934543e-06,
+      "loss": 2.5488,
+      "step": 3861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.095261339147347e-06,
+      "loss": 2.5016,
+      "step": 3862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.093302346772711e-06,
+      "loss": 2.5127,
+      "step": 3863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.091343178126478e-06,
+      "loss": 2.6184,
+      "step": 3864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.089383833524524e-06,
+      "loss": 2.4993,
+      "step": 3865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0874243132827505e-06,
+      "loss": 2.6617,
+      "step": 3866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.085464617717088e-06,
+      "loss": 2.4842,
+      "step": 3867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.083504747143496e-06,
+      "loss": 2.5359,
+      "step": 3868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.08154470187796e-06,
+      "loss": 2.6675,
+      "step": 3869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.079584482236499e-06,
+      "loss": 2.4625,
+      "step": 3870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.077624088535152e-06,
+      "loss": 2.4221,
+      "step": 3871
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.075663521089994e-06,
+      "loss": 2.3978,
+      "step": 3872
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.073702780217124e-06,
+      "loss": 2.5835,
+      "step": 3873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.071741866232665e-06,
+      "loss": 2.7314,
+      "step": 3874
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.069780779452779e-06,
+      "loss": 2.6017,
+      "step": 3875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0678195201936455e-06,
+      "loss": 2.6736,
+      "step": 3876
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.065858088771477e-06,
+      "loss": 2.5366,
+      "step": 3877
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.06389648550251e-06,
+      "loss": 2.7111,
+      "step": 3878
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.061934710703015e-06,
+      "loss": 2.4499,
+      "step": 3879
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.059972764689283e-06,
+      "loss": 2.4882,
+      "step": 3880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.058010647777636e-06,
+      "loss": 2.7229,
+      "step": 3881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.056048360284424e-06,
+      "loss": 2.467,
+      "step": 3882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0540859025260245e-06,
+      "loss": 2.6057,
+      "step": 3883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0521232748188416e-06,
+      "loss": 2.3461,
+      "step": 3884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.050160477479305e-06,
+      "loss": 2.6968,
+      "step": 3885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.048197510823876e-06,
+      "loss": 2.3131,
+      "step": 3886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.046234375169041e-06,
+      "loss": 2.4788,
+      "step": 3887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.044271070831312e-06,
+      "loss": 2.3937,
+      "step": 3888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0423075981272315e-06,
+      "loss": 2.5292,
+      "step": 3889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.040343957373367e-06,
+      "loss": 2.5041,
+      "step": 3890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.038380148886312e-06,
+      "loss": 2.3524,
+      "step": 3891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0364161729826905e-06,
+      "loss": 2.6107,
+      "step": 3892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.03445202997915e-06,
+      "loss": 2.5269,
+      "step": 3893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.032487720192367e-06,
+      "loss": 2.7897,
+      "step": 3894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.030523243939046e-06,
+      "loss": 2.6203,
+      "step": 3895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.028558601535915e-06,
+      "loss": 2.3367,
+      "step": 3896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.02659379329973e-06,
+      "loss": 2.6044,
+      "step": 3897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.024628819547274e-06,
+      "loss": 2.5391,
+      "step": 3898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.022663680595361e-06,
+      "loss": 2.527,
+      "step": 3899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.020698376760824e-06,
+      "loss": 2.6466,
+      "step": 3900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.018732908360527e-06,
+      "loss": 2.5457,
+      "step": 3901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.016767275711359e-06,
+      "loss": 2.4356,
+      "step": 3902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.014801479130237e-06,
+      "loss": 2.6068,
+      "step": 3903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.012835518934102e-06,
+      "loss": 2.3668,
+      "step": 3904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.010869395439926e-06,
+      "loss": 2.53,
+      "step": 3905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.008903108964702e-06,
+      "loss": 2.5431,
+      "step": 3906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.006936659825453e-06,
+      "loss": 2.5259,
+      "step": 3907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0049700483392256e-06,
+      "loss": 2.4428,
+      "step": 3908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.003003274823094e-06,
+      "loss": 2.558,
+      "step": 3909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0010363395941585e-06,
+      "loss": 2.5492,
+      "step": 3910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.999069242969546e-06,
+      "loss": 2.6465,
+      "step": 3911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.997101985266408e-06,
+      "loss": 2.5352,
+      "step": 3912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.995134566801923e-06,
+      "loss": 2.553,
+      "step": 3913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.993166987893294e-06,
+      "loss": 2.5015,
+      "step": 3914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.991199248857752e-06,
+      "loss": 2.5744,
+      "step": 3915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9892313500125545e-06,
+      "loss": 2.5855,
+      "step": 3916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9872632916749796e-06,
+      "loss": 2.5242,
+      "step": 3917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.985295074162337e-06,
+      "loss": 2.6281,
+      "step": 3918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.983326697791959e-06,
+      "loss": 2.4923,
+      "step": 3919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.981358162881202e-06,
+      "loss": 2.4754,
+      "step": 3920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.979389469747453e-06,
+      "loss": 2.5251,
+      "step": 3921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.977420618708122e-06,
+      "loss": 2.525,
+      "step": 3922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.975451610080643e-06,
+      "loss": 2.6192,
+      "step": 3923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9734824441824745e-06,
+      "loss": 2.6711,
+      "step": 3924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.971513121331105e-06,
+      "loss": 2.5914,
+      "step": 3925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.969543641844044e-06,
+      "loss": 2.5018,
+      "step": 3926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.967574006038829e-06,
+      "loss": 2.4823,
+      "step": 3927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.965604214233022e-06,
+      "loss": 2.5358,
+      "step": 3928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9636342667442095e-06,
+      "loss": 2.4957,
+      "step": 3929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.961664163890002e-06,
+      "loss": 2.5533,
+      "step": 3930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.959693905988038e-06,
+      "loss": 2.5552,
+      "step": 3931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.957723493355977e-06,
+      "loss": 2.5967,
+      "step": 3932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.955752926311508e-06,
+      "loss": 2.5695,
+      "step": 3933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.953782205172343e-06,
+      "loss": 2.5538,
+      "step": 3934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.951811330256218e-06,
+      "loss": 2.5676,
+      "step": 3935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.949840301880894e-06,
+      "loss": 2.74,
+      "step": 3936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.947869120364157e-06,
+      "loss": 2.5113,
+      "step": 3937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.945897786023817e-06,
+      "loss": 2.4869,
+      "step": 3938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943926299177713e-06,
+      "loss": 2.6041,
+      "step": 3939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.941954660143703e-06,
+      "loss": 2.4952,
+      "step": 3940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.93998286923967e-06,
+      "loss": 2.5617,
+      "step": 3941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.938010926783525e-06,
+      "loss": 2.5811,
+      "step": 3942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.936038833093199e-06,
+      "loss": 2.5142,
+      "step": 3943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9340665884866535e-06,
+      "loss": 2.4873,
+      "step": 3944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.932094193281868e-06,
+      "loss": 2.4744,
+      "step": 3945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.930121647796852e-06,
+      "loss": 2.5832,
+      "step": 3946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.928148952349631e-06,
+      "loss": 2.5365,
+      "step": 3947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.926176107258265e-06,
+      "loss": 2.7332,
+      "step": 3948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.924203112840832e-06,
+      "loss": 2.5406,
+      "step": 3949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.922229969415432e-06,
+      "loss": 2.5556,
+      "step": 3950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.920256677300195e-06,
+      "loss": 2.5723,
+      "step": 3951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.918283236813272e-06,
+      "loss": 2.5269,
+      "step": 3952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9163096482728385e-06,
+      "loss": 2.6579,
+      "step": 3953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.914335911997091e-06,
+      "loss": 2.4878,
+      "step": 3954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.912362028304254e-06,
+      "loss": 2.5012,
+      "step": 3955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.910387997512573e-06,
+      "loss": 2.6394,
+      "step": 3956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9084138199403195e-06,
+      "loss": 2.4923,
+      "step": 3957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.906439495905787e-06,
+      "loss": 2.5159,
+      "step": 3958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.904465025727293e-06,
+      "loss": 2.5627,
+      "step": 3959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9024904097231765e-06,
+      "loss": 2.5482,
+      "step": 3960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9005156482118045e-06,
+      "loss": 2.4057,
+      "step": 3961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.898540741511564e-06,
+      "loss": 2.3859,
+      "step": 3962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.896565689940866e-06,
+      "loss": 2.445,
+      "step": 3963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.894590493818149e-06,
+      "loss": 2.6075,
+      "step": 3964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8926151534618646e-06,
+      "loss": 2.3935,
+      "step": 3965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.890639669190499e-06,
+      "loss": 2.5498,
+      "step": 3966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.888664041322555e-06,
+      "loss": 2.5568,
+      "step": 3967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8866882701765605e-06,
+      "loss": 2.4742,
+      "step": 3968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.884712356071067e-06,
+      "loss": 2.6151,
+      "step": 3969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.882736299324648e-06,
+      "loss": 2.537,
+      "step": 3970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.880760100255899e-06,
+      "loss": 2.4412,
+      "step": 3971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8787837591834415e-06,
+      "loss": 2.6351,
+      "step": 3972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.876807276425918e-06,
+      "loss": 2.4655,
+      "step": 3973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8748306523019925e-06,
+      "loss": 2.7257,
+      "step": 3974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.872853887130356e-06,
+      "loss": 2.6065,
+      "step": 3975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.870876981229716e-06,
+      "loss": 2.6387,
+      "step": 3976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.86889993491881e-06,
+      "loss": 2.4923,
+      "step": 3977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.866922748516392e-06,
+      "loss": 2.6374,
+      "step": 3978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.864945422341242e-06,
+      "loss": 2.4061,
+      "step": 3979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.86296795671216e-06,
+      "loss": 2.4448,
+      "step": 3980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.860990351947973e-06,
+      "loss": 2.5263,
+      "step": 3981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.859012608367523e-06,
+      "loss": 2.4387,
+      "step": 3982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8570347262896835e-06,
+      "loss": 2.5369,
+      "step": 3983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.855056706033343e-06,
+      "loss": 2.3639,
+      "step": 3984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.853078547917415e-06,
+      "loss": 2.5387,
+      "step": 3985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.851100252260835e-06,
+      "loss": 2.4705,
+      "step": 3986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.849121819382562e-06,
+      "loss": 2.4171,
+      "step": 3987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.847143249601575e-06,
+      "loss": 2.2717,
+      "step": 3988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.845164543236876e-06,
+      "loss": 2.5776,
+      "step": 3989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.84318570060749e-06,
+      "loss": 2.5797,
+      "step": 3990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.841206722032463e-06,
+      "loss": 2.5699,
+      "step": 3991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.839227607830862e-06,
+      "loss": 2.5221,
+      "step": 3992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.837248358321775e-06,
+      "loss": 2.3332,
+      "step": 3993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.835268973824319e-06,
+      "loss": 2.5413,
+      "step": 3994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.833289454657623e-06,
+      "loss": 2.6294,
+      "step": 3995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.831309801140841e-06,
+      "loss": 2.4566,
+      "step": 3996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.829330013593154e-06,
+      "loss": 2.6098,
+      "step": 3997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.827350092333758e-06,
+      "loss": 2.5004,
+      "step": 3998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.825370037681872e-06,
+      "loss": 2.6004,
+      "step": 3999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.82338984995674e-06,
+      "loss": 2.4355,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.821409529477624e-06,
+      "loss": 2.4033,
+      "step": 4001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.819429076563805e-06,
+      "loss": 2.3989,
+      "step": 4002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.817448491534594e-06,
+      "loss": 2.5317,
+      "step": 4003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.815467774709314e-06,
+      "loss": 2.556,
+      "step": 4004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.813486926407315e-06,
+      "loss": 2.5343,
+      "step": 4005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.811505946947967e-06,
+      "loss": 2.5064,
+      "step": 4006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.809524836650658e-06,
+      "loss": 2.4374,
+      "step": 4007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.807543595834799e-06,
+      "loss": 2.5041,
+      "step": 4008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.805562224819827e-06,
+      "loss": 2.4426,
+      "step": 4009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.803580723925193e-06,
+      "loss": 2.459,
+      "step": 4010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.801599093470372e-06,
+      "loss": 2.5747,
+      "step": 4011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.799617333774861e-06,
+      "loss": 2.5783,
+      "step": 4012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7976354451581715e-06,
+      "loss": 2.5484,
+      "step": 4013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.795653427939846e-06,
+      "loss": 2.4644,
+      "step": 4014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793671282439439e-06,
+      "loss": 2.5731,
+      "step": 4015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791689008976531e-06,
+      "loss": 2.4201,
+      "step": 4016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.78970660787072e-06,
+      "loss": 2.7292,
+      "step": 4017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787724079441626e-06,
+      "loss": 2.5944,
+      "step": 4018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7857414240088895e-06,
+      "loss": 2.5921,
+      "step": 4019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.783758641892172e-06,
+      "loss": 2.5541,
+      "step": 4020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7817757334111526e-06,
+      "loss": 2.5021,
+      "step": 4021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.779792698885534e-06,
+      "loss": 2.6369,
+      "step": 4022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.777809538635039e-06,
+      "loss": 2.415,
+      "step": 4023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.775826252979409e-06,
+      "loss": 2.5949,
+      "step": 4024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.773842842238404e-06,
+      "loss": 2.5123,
+      "step": 4025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.771859306731809e-06,
+      "loss": 2.6106,
+      "step": 4026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.769875646779428e-06,
+      "loss": 2.4634,
+      "step": 4027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.767891862701081e-06,
+      "loss": 2.5433,
+      "step": 4028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.765907954816612e-06,
+      "loss": 2.5971,
+      "step": 4029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7639239234458846e-06,
+      "loss": 2.5318,
+      "step": 4030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7619397689087775e-06,
+      "loss": 2.6506,
+      "step": 4031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.759955491525196e-06,
+      "loss": 2.64,
+      "step": 4032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7579710916150645e-06,
+      "loss": 2.5451,
+      "step": 4033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.755986569498321e-06,
+      "loss": 2.587,
+      "step": 4034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.754001925494929e-06,
+      "loss": 2.5761,
+      "step": 4035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7520171599248704e-06,
+      "loss": 2.6311,
+      "step": 4036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.750032273108145e-06,
+      "loss": 2.4299,
+      "step": 4037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.748047265364773e-06,
+      "loss": 2.536,
+      "step": 4038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.746062137014795e-06,
+      "loss": 2.4225,
+      "step": 4039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.744076888378272e-06,
+      "loss": 2.557,
+      "step": 4040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7420915197752815e-06,
+      "loss": 2.5416,
+      "step": 4041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.740106031525919e-06,
+      "loss": 2.4537,
+      "step": 4042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.738120423950306e-06,
+      "loss": 2.5181,
+      "step": 4043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.73613469736858e-06,
+      "loss": 2.635,
+      "step": 4044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.734148852100892e-06,
+      "loss": 2.4326,
+      "step": 4045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.732162888467421e-06,
+      "loss": 2.5944,
+      "step": 4046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.73017680678836e-06,
+      "loss": 2.463,
+      "step": 4047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.728190607383921e-06,
+      "loss": 2.5032,
+      "step": 4048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.726204290574336e-06,
+      "loss": 2.3934,
+      "step": 4049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.724217856679859e-06,
+      "loss": 2.5912,
+      "step": 4050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.722231306020758e-06,
+      "loss": 2.6152,
+      "step": 4051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7202446389173225e-06,
+      "loss": 2.5514,
+      "step": 4052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.718257855689859e-06,
+      "loss": 2.5303,
+      "step": 4053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7162709566586925e-06,
+      "loss": 2.496,
+      "step": 4054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.71428394214417e-06,
+      "loss": 2.6053,
+      "step": 4055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.712296812466657e-06,
+      "loss": 2.6862,
+      "step": 4056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.710309567946531e-06,
+      "loss": 2.4729,
+      "step": 4057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.708322208904196e-06,
+      "loss": 2.516,
+      "step": 4058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.706334735660069e-06,
+      "loss": 2.63,
+      "step": 4059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.704347148534589e-06,
+      "loss": 2.5354,
+      "step": 4060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.702359447848212e-06,
+      "loss": 2.4654,
+      "step": 4061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.70037163392141e-06,
+      "loss": 2.5069,
+      "step": 4062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6983837070746785e-06,
+      "loss": 2.6844,
+      "step": 4063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.696395667628526e-06,
+      "loss": 2.5196,
+      "step": 4064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.694407515903482e-06,
+      "loss": 2.734,
+      "step": 4065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.692419252220091e-06,
+      "loss": 2.6266,
+      "step": 4066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.690430876898923e-06,
+      "loss": 2.5426,
+      "step": 4067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.688442390260559e-06,
+      "loss": 2.5827,
+      "step": 4068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.686453792625598e-06,
+      "loss": 2.4421,
+      "step": 4069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6844650843146595e-06,
+      "loss": 2.6865,
+      "step": 4070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.68247626564838e-06,
+      "loss": 2.6414,
+      "step": 4071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6804873369474155e-06,
+      "loss": 2.6358,
+      "step": 4072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.678498298532437e-06,
+      "loss": 2.6106,
+      "step": 4073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6765091507241345e-06,
+      "loss": 2.5209,
+      "step": 4074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.674519893843216e-06,
+      "loss": 2.6547,
+      "step": 4075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.672530528210405e-06,
+      "loss": 2.5692,
+      "step": 4076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6705410541464455e-06,
+      "loss": 2.5473,
+      "step": 4077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.668551471972097e-06,
+      "loss": 2.6338,
+      "step": 4078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.666561782008137e-06,
+      "loss": 2.5355,
+      "step": 4079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.664571984575362e-06,
+      "loss": 2.69,
+      "step": 4080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.662582079994583e-06,
+      "loss": 2.3956,
+      "step": 4081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.660592068586629e-06,
+      "loss": 2.4715,
+      "step": 4082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.658601950672347e-06,
+      "loss": 2.4634,
+      "step": 4083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.656611726572601e-06,
+      "loss": 2.4539,
+      "step": 4084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.654621396608272e-06,
+      "loss": 2.6542,
+      "step": 4085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.65263096110026e-06,
+      "loss": 2.5361,
+      "step": 4086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.650640420369475e-06,
+      "loss": 2.5117,
+      "step": 4087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.648649774736855e-06,
+      "loss": 2.4335,
+      "step": 4088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.646659024523346e-06,
+      "loss": 2.4316,
+      "step": 4089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.644668170049913e-06,
+      "loss": 2.6245,
+      "step": 4090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.642677211637542e-06,
+      "loss": 2.4428,
+      "step": 4091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.640686149607228e-06,
+      "loss": 2.6339,
+      "step": 4092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.638694984279991e-06,
+      "loss": 2.5714,
+      "step": 4093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6367037159768625e-06,
+      "loss": 2.6661,
+      "step": 4094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.63471234501889e-06,
+      "loss": 2.5311,
+      "step": 4095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.632720871727142e-06,
+      "loss": 2.5472,
+      "step": 4096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.630729296422701e-06,
+      "loss": 2.5339,
+      "step": 4097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.628737619426664e-06,
+      "loss": 2.6301,
+      "step": 4098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.626745841060146e-06,
+      "loss": 2.4696,
+      "step": 4099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.624753961644281e-06,
+      "loss": 2.4119,
+      "step": 4100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.622761981500215e-06,
+      "loss": 2.4492,
+      "step": 4101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.620769900949113e-06,
+      "loss": 2.5032,
+      "step": 4102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.618777720312156e-06,
+      "loss": 2.7037,
+      "step": 4103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.61678543991054e-06,
+      "loss": 2.4918,
+      "step": 4104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6147930600654755e-06,
+      "loss": 2.4369,
+      "step": 4105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.612800581098193e-06,
+      "loss": 2.4236,
+      "step": 4106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.610808003329937e-06,
+      "loss": 2.6601,
+      "step": 4107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.608815327081969e-06,
+      "loss": 2.5355,
+      "step": 4108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.606822552675564e-06,
+      "loss": 2.5028,
+      "step": 4109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.604829680432012e-06,
+      "loss": 2.3269,
+      "step": 4110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.602836710672627e-06,
+      "loss": 2.5745,
+      "step": 4111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.600843643718728e-06,
+      "loss": 2.4346,
+      "step": 4112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.598850479891653e-06,
+      "loss": 2.5822,
+      "step": 4113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.596857219512762e-06,
+      "loss": 2.4729,
+      "step": 4114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.594863862903422e-06,
+      "loss": 2.3652,
+      "step": 4115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.592870410385021e-06,
+      "loss": 2.4995,
+      "step": 4116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.590876862278959e-06,
+      "loss": 2.6704,
+      "step": 4117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.588883218906653e-06,
+      "loss": 2.5942,
+      "step": 4118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.586889480589537e-06,
+      "loss": 2.6051,
+      "step": 4119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.584895647649058e-06,
+      "loss": 2.6485,
+      "step": 4120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.582901720406677e-06,
+      "loss": 2.5641,
+      "step": 4121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.580907699183875e-06,
+      "loss": 2.4983,
+      "step": 4122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.578913584302144e-06,
+      "loss": 2.5445,
+      "step": 4123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.57691937608299e-06,
+      "loss": 2.5767,
+      "step": 4124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.574925074847939e-06,
+      "loss": 2.6114,
+      "step": 4125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.572930680918529e-06,
+      "loss": 2.4994,
+      "step": 4126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.570936194616313e-06,
+      "loss": 2.5801,
+      "step": 4127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.568941616262861e-06,
+      "loss": 2.5797,
+      "step": 4128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.566946946179753e-06,
+      "loss": 2.5963,
+      "step": 4129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.564952184688588e-06,
+      "loss": 2.5391,
+      "step": 4130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.56295733211098e-06,
+      "loss": 2.5562,
+      "step": 4131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.560962388768554e-06,
+      "loss": 2.4871,
+      "step": 4132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5589673549829525e-06,
+      "loss": 2.4754,
+      "step": 4133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.556972231075834e-06,
+      "loss": 2.4606,
+      "step": 4134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.554977017368868e-06,
+      "loss": 2.5276,
+      "step": 4135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.552981714183738e-06,
+      "loss": 2.5456,
+      "step": 4136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.550986321842147e-06,
+      "loss": 2.46,
+      "step": 4137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.548990840665809e-06,
+      "loss": 2.5926,
+      "step": 4138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5469952709764515e-06,
+      "loss": 2.5604,
+      "step": 4139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5449996130958185e-06,
+      "loss": 2.5762,
+      "step": 4140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5430038673456655e-06,
+      "loss": 2.4994,
+      "step": 4141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5410080340477634e-06,
+      "loss": 2.4543,
+      "step": 4142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.539012113523899e-06,
+      "loss": 2.5115,
+      "step": 4143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5370161060958726e-06,
+      "loss": 2.5635,
+      "step": 4144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.535020012085497e-06,
+      "loss": 2.7138,
+      "step": 4145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.533023831814599e-06,
+      "loss": 2.6107,
+      "step": 4146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5310275656050195e-06,
+      "loss": 2.4846,
+      "step": 4147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.529031213778615e-06,
+      "loss": 2.6026,
+      "step": 4148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5270347766572515e-06,
+      "loss": 2.645,
+      "step": 4149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.525038254562817e-06,
+      "loss": 2.5704,
+      "step": 4150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.523041647817204e-06,
+      "loss": 2.5021,
+      "step": 4151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.521044956742324e-06,
+      "loss": 2.5612,
+      "step": 4152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5190481816601e-06,
+      "loss": 2.5784,
+      "step": 4153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.517051322892468e-06,
+      "loss": 2.5125,
+      "step": 4154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.515054380761382e-06,
+      "loss": 2.6093,
+      "step": 4155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.513057355588804e-06,
+      "loss": 2.4796,
+      "step": 4156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.511060247696713e-06,
+      "loss": 2.4179,
+      "step": 4157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.509063057407098e-06,
+      "loss": 2.6826,
+      "step": 4158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.507065785041963e-06,
+      "loss": 2.652,
+      "step": 4159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.505068430923327e-06,
+      "loss": 2.5544,
+      "step": 4160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.503070995373221e-06,
+      "loss": 2.4317,
+      "step": 4161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5010734787136865e-06,
+      "loss": 2.5095,
+      "step": 4162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.499075881266784e-06,
+      "loss": 2.5806,
+      "step": 4163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.497078203354577e-06,
+      "loss": 2.6062,
+      "step": 4164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4950804452991545e-06,
+      "loss": 2.5624,
+      "step": 4165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4930826074226085e-06,
+      "loss": 2.6041,
+      "step": 4166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.49108469004705e-06,
+      "loss": 2.5914,
+      "step": 4167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4890866934946e-06,
+      "loss": 2.7266,
+      "step": 4168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.487088618087391e-06,
+      "loss": 2.4253,
+      "step": 4169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.485090464147571e-06,
+      "loss": 2.5052,
+      "step": 4170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.483092231997299e-06,
+      "loss": 2.4201,
+      "step": 4171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.481093921958749e-06,
+      "loss": 2.5636,
+      "step": 4172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.479095534354103e-06,
+      "loss": 2.6252,
+      "step": 4173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.47709706950556e-06,
+      "loss": 2.4417,
+      "step": 4174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.475098527735329e-06,
+      "loss": 2.4963,
+      "step": 4175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.473099909365633e-06,
+      "loss": 2.5189,
+      "step": 4176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.471101214718704e-06,
+      "loss": 2.5357,
+      "step": 4177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.469102444116791e-06,
+      "loss": 2.698,
+      "step": 4178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.467103597882154e-06,
+      "loss": 2.5239,
+      "step": 4179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4651046763370615e-06,
+      "loss": 2.3852,
+      "step": 4180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.463105679803798e-06,
+      "loss": 2.434,
+      "step": 4181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.461106608604658e-06,
+      "loss": 2.5047,
+      "step": 4182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.459107463061949e-06,
+      "loss": 2.5734,
+      "step": 4183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.457108243497993e-06,
+      "loss": 2.3627,
+      "step": 4184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.455108950235119e-06,
+      "loss": 2.5102,
+      "step": 4185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.45310958359567e-06,
+      "loss": 2.5064,
+      "step": 4186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4511101439020006e-06,
+      "loss": 2.5779,
+      "step": 4187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.449110631476481e-06,
+      "loss": 2.6646,
+      "step": 4188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4471110466414855e-06,
+      "loss": 2.3701,
+      "step": 4189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.445111389719408e-06,
+      "loss": 2.3831,
+      "step": 4190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.443111661032648e-06,
+      "loss": 2.5394,
+      "step": 4191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.44111186090362e-06,
+      "loss": 2.6391,
+      "step": 4192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.439111989654748e-06,
+      "loss": 2.6628,
+      "step": 4193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.43711204760847e-06,
+      "loss": 2.6402,
+      "step": 4194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.435112035087232e-06,
+      "loss": 2.441,
+      "step": 4195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.433111952413496e-06,
+      "loss": 2.5556,
+      "step": 4196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.431111799909728e-06,
+      "loss": 2.568,
+      "step": 4197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.429111577898415e-06,
+      "loss": 2.6705,
+      "step": 4198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.427111286702044e-06,
+      "loss": 2.5791,
+      "step": 4199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.425110926643123e-06,
+      "loss": 2.6414,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.42311049804417e-06,
+      "loss": 2.5155,
+      "step": 4201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.421110001227705e-06,
+      "loss": 2.4702,
+      "step": 4202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.41910943651627e-06,
+      "loss": 2.3799,
+      "step": 4203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.417108804232409e-06,
+      "loss": 2.586,
+      "step": 4204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.415108104698687e-06,
+      "loss": 2.6274,
+      "step": 4205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.413107338237667e-06,
+      "loss": 2.672,
+      "step": 4206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.411106505171936e-06,
+      "loss": 2.4756,
+      "step": 4207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.409105605824082e-06,
+      "loss": 2.6223,
+      "step": 4208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.40710464051671e-06,
+      "loss": 2.4368,
+      "step": 4209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.405103609572427e-06,
+      "loss": 2.5212,
+      "step": 4210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.403102513313862e-06,
+      "loss": 2.3784,
+      "step": 4211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.4011013520636466e-06,
+      "loss": 2.521,
+      "step": 4212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.399100126144426e-06,
+      "loss": 2.3947,
+      "step": 4213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3970988358788565e-06,
+      "loss": 2.5648,
+      "step": 4214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.395097481589599e-06,
+      "loss": 2.4612,
+      "step": 4215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.393096063599333e-06,
+      "loss": 2.5685,
+      "step": 4216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.391094582230744e-06,
+      "loss": 2.6409,
+      "step": 4217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.389093037806524e-06,
+      "loss": 2.3948,
+      "step": 4218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.387091430649385e-06,
+      "loss": 2.4082,
+      "step": 4219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.385089761082039e-06,
+      "loss": 2.4952,
+      "step": 4220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.383088029427214e-06,
+      "loss": 2.5713,
+      "step": 4221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.381086236007647e-06,
+      "loss": 2.4298,
+      "step": 4222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.379084381146082e-06,
+      "loss": 2.4903,
+      "step": 4223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.377082465165277e-06,
+      "loss": 2.6372,
+      "step": 4224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.375080488387998e-06,
+      "loss": 2.6019,
+      "step": 4225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3730784511370204e-06,
+      "loss": 2.5179,
+      "step": 4226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.371076353735132e-06,
+      "loss": 2.6131,
+      "step": 4227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3690741965051255e-06,
+      "loss": 2.516,
+      "step": 4228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.367071979769805e-06,
+      "loss": 2.6027,
+      "step": 4229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3650697038519885e-06,
+      "loss": 2.4422,
+      "step": 4230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.363067369074498e-06,
+      "loss": 2.4694,
+      "step": 4231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.361064975760166e-06,
+      "loss": 2.6596,
+      "step": 4232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.35906252423184e-06,
+      "loss": 2.5504,
+      "step": 4233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.357060014812367e-06,
+      "loss": 2.5195,
+      "step": 4234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.355057447824612e-06,
+      "loss": 2.4864,
+      "step": 4235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.353054823591446e-06,
+      "loss": 2.5911,
+      "step": 4236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.351052142435747e-06,
+      "loss": 2.5579,
+      "step": 4237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.349049404680407e-06,
+      "loss": 2.5381,
+      "step": 4238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.347046610648324e-06,
+      "loss": 2.4927,
+      "step": 4239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.345043760662406e-06,
+      "loss": 2.5695,
+      "step": 4240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.343040855045567e-06,
+      "loss": 2.6301,
+      "step": 4241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.341037894120737e-06,
+      "loss": 2.5352,
+      "step": 4242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.339034878210846e-06,
+      "loss": 2.4195,
+      "step": 4243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3370318076388405e-06,
+      "loss": 2.6437,
+      "step": 4244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.335028682727672e-06,
+      "loss": 2.734,
+      "step": 4245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.333025503800301e-06,
+      "loss": 2.4667,
+      "step": 4246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3310222711796975e-06,
+      "loss": 2.4553,
+      "step": 4247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.329018985188841e-06,
+      "loss": 2.4828,
+      "step": 4248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.327015646150716e-06,
+      "loss": 2.4915,
+      "step": 4249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.32501225438832e-06,
+      "loss": 2.4959,
+      "step": 4250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.323008810224657e-06,
+      "loss": 2.4616,
+      "step": 4251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3210053139827374e-06,
+      "loss": 2.3947,
+      "step": 4252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.319001765985585e-06,
+      "loss": 2.5313,
+      "step": 4253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3169981665562275e-06,
+      "loss": 2.3548,
+      "step": 4254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.314994516017703e-06,
+      "loss": 2.7625,
+      "step": 4255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3129908146930565e-06,
+      "loss": 2.6501,
+      "step": 4256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.310987062905343e-06,
+      "loss": 2.5438,
+      "step": 4257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.308983260977622e-06,
+      "loss": 2.5125,
+      "step": 4258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.306979409232967e-06,
+      "loss": 2.4539,
+      "step": 4259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.304975507994453e-06,
+      "loss": 2.6234,
+      "step": 4260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.302971557585169e-06,
+      "loss": 2.4804,
+      "step": 4261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.300967558328208e-06,
+      "loss": 2.6117,
+      "step": 4262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.298963510546673e-06,
+      "loss": 2.5833,
+      "step": 4263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.296959414563671e-06,
+      "loss": 2.5299,
+      "step": 4264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2949552707023215e-06,
+      "loss": 2.5245,
+      "step": 4265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.29295107928575e-06,
+      "loss": 2.6001,
+      "step": 4266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.290946840637089e-06,
+      "loss": 2.4249,
+      "step": 4267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.288942555079479e-06,
+      "loss": 2.5154,
+      "step": 4268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.286938222936069e-06,
+      "loss": 2.4015,
+      "step": 4269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.284933844530013e-06,
+      "loss": 2.6337,
+      "step": 4270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.282929420184474e-06,
+      "loss": 2.6177,
+      "step": 4271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2809249502226255e-06,
+      "loss": 2.5195,
+      "step": 4272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.278920434967643e-06,
+      "loss": 2.5401,
+      "step": 4273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2769158747427115e-06,
+      "loss": 2.5632,
+      "step": 4274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.274911269871024e-06,
+      "loss": 2.5878,
+      "step": 4275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.27290662067578e-06,
+      "loss": 2.3144,
+      "step": 4276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.270901927480186e-06,
+      "loss": 2.4555,
+      "step": 4277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.268897190607457e-06,
+      "loss": 2.5036,
+      "step": 4278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.266892410380814e-06,
+      "loss": 2.3945,
+      "step": 4279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.264887587123483e-06,
+      "loss": 2.3634,
+      "step": 4280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.262882721158699e-06,
+      "loss": 2.5458,
+      "step": 4281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.260877812809704e-06,
+      "loss": 2.4633,
+      "step": 4282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2588728623997485e-06,
+      "loss": 2.6636,
+      "step": 4283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.256867870252087e-06,
+      "loss": 2.5437,
+      "step": 4284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.25486283668998e-06,
+      "loss": 2.6446,
+      "step": 4285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2528577620366986e-06,
+      "loss": 2.5654,
+      "step": 4286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.250852646615514e-06,
+      "loss": 2.5356,
+      "step": 4287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.248847490749711e-06,
+      "loss": 2.5357,
+      "step": 4288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.246842294762581e-06,
+      "loss": 2.5901,
+      "step": 4289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.244837058977413e-06,
+      "loss": 2.5917,
+      "step": 4290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.242831783717514e-06,
+      "loss": 2.5189,
+      "step": 4291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.240826469306187e-06,
+      "loss": 2.4242,
+      "step": 4292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2388211160667475e-06,
+      "loss": 2.508,
+      "step": 4293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.236815724322516e-06,
+      "loss": 2.4792,
+      "step": 4294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.23481029439682e-06,
+      "loss": 2.725,
+      "step": 4295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.232804826612993e-06,
+      "loss": 2.4685,
+      "step": 4296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.23079932129437e-06,
+      "loss": 2.5457,
+      "step": 4297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.228793778764297e-06,
+      "loss": 2.5421,
+      "step": 4298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.226788199346127e-06,
+      "loss": 2.5129,
+      "step": 4299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.224782583363215e-06,
+      "loss": 2.5694,
+      "step": 4300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.222776931138926e-06,
+      "loss": 2.5762,
+      "step": 4301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.220771242996624e-06,
+      "loss": 2.5664,
+      "step": 4302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2187655192596875e-06,
+      "loss": 2.4692,
+      "step": 4303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.216759760251493e-06,
+      "loss": 2.5931,
+      "step": 4304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.214753966295429e-06,
+      "loss": 2.5956,
+      "step": 4305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2127481377148845e-06,
+      "loss": 2.4225,
+      "step": 4306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.210742274833259e-06,
+      "loss": 2.4239,
+      "step": 4307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.208736377973954e-06,
+      "loss": 2.3057,
+      "step": 4308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.206730447460377e-06,
+      "loss": 2.5247,
+      "step": 4309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.204724483615941e-06,
+      "loss": 2.3896,
+      "step": 4310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.202718486764065e-06,
+      "loss": 2.3684,
+      "step": 4311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.200712457228173e-06,
+      "loss": 2.5044,
+      "step": 4312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.198706395331695e-06,
+      "loss": 2.5967,
+      "step": 4313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1967003013980655e-06,
+      "loss": 2.5187,
+      "step": 4314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.194694175750723e-06,
+      "loss": 2.6281,
+      "step": 4315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1926880187131134e-06,
+      "loss": 2.3195,
+      "step": 4316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.190681830608686e-06,
+      "loss": 2.5504,
+      "step": 4317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.188675611760896e-06,
+      "loss": 2.3847,
+      "step": 4318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.186669362493204e-06,
+      "loss": 2.5551,
+      "step": 4319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.184663083129073e-06,
+      "loss": 2.5166,
+      "step": 4320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.182656773991973e-06,
+      "loss": 2.4939,
+      "step": 4321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.18065043540538e-06,
+      "loss": 2.5114,
+      "step": 4322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.178644067692771e-06,
+      "loss": 2.4324,
+      "step": 4323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.176637671177631e-06,
+      "loss": 2.5323,
+      "step": 4324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.174631246183449e-06,
+      "loss": 2.5382,
+      "step": 4325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.172624793033717e-06,
+      "loss": 2.5209,
+      "step": 4326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.170618312051934e-06,
+      "loss": 2.435,
+      "step": 4327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.168611803561599e-06,
+      "loss": 2.577,
+      "step": 4328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.166605267886222e-06,
+      "loss": 2.3858,
+      "step": 4329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.164598705349313e-06,
+      "loss": 2.429,
+      "step": 4330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.162592116274386e-06,
+      "loss": 2.5314,
+      "step": 4331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.160585500984962e-06,
+      "loss": 2.548,
+      "step": 4332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.158578859804562e-06,
+      "loss": 2.5639,
+      "step": 4333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.156572193056718e-06,
+      "loss": 2.6438,
+      "step": 4334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1545655010649605e-06,
+      "loss": 2.5023,
+      "step": 4335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.152558784152824e-06,
+      "loss": 2.6907,
+      "step": 4336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.150552042643851e-06,
+      "loss": 2.496,
+      "step": 4337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.148545276861583e-06,
+      "loss": 2.7109,
+      "step": 4338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.146538487129571e-06,
+      "loss": 2.3729,
+      "step": 4339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.144531673771364e-06,
+      "loss": 2.55,
+      "step": 4340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.14252483711052e-06,
+      "loss": 2.4161,
+      "step": 4341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.140517977470597e-06,
+      "loss": 2.4703,
+      "step": 4342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.138511095175158e-06,
+      "loss": 2.5521,
+      "step": 4343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.136504190547772e-06,
+      "loss": 2.4457,
+      "step": 4344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.134497263912007e-06,
+      "loss": 2.6243,
+      "step": 4345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.132490315591437e-06,
+      "loss": 2.6425,
+      "step": 4346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.130483345909642e-06,
+      "loss": 2.5647,
+      "step": 4347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1284763551901995e-06,
+      "loss": 2.499,
+      "step": 4348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.126469343756695e-06,
+      "loss": 2.6119,
+      "step": 4349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.124462311932719e-06,
+      "loss": 2.4121,
+      "step": 4350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1224552600418566e-06,
+      "loss": 2.5365,
+      "step": 4351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1204481884077075e-06,
+      "loss": 2.4647,
+      "step": 4352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.118441097353867e-06,
+      "loss": 2.6164,
+      "step": 4353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.116433987203934e-06,
+      "loss": 2.4972,
+      "step": 4354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.114426858281515e-06,
+      "loss": 2.5168,
+      "step": 4355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.112419710910213e-06,
+      "loss": 2.5557,
+      "step": 4356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.110412545413641e-06,
+      "loss": 2.5062,
+      "step": 4357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.10840536211541e-06,
+      "loss": 2.7639,
+      "step": 4358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.106398161339135e-06,
+      "loss": 2.5096,
+      "step": 4359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.104390943408434e-06,
+      "loss": 2.4705,
+      "step": 4360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.102383708646929e-06,
+      "loss": 2.6436,
+      "step": 4361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.100376457378243e-06,
+      "loss": 2.453,
+      "step": 4362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.098369189926003e-06,
+      "loss": 2.6943,
+      "step": 4363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.096361906613836e-06,
+      "loss": 2.3764,
+      "step": 4364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.094354607765374e-06,
+      "loss": 2.6486,
+      "step": 4365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.092347293704253e-06,
+      "loss": 2.6558,
+      "step": 4366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.090339964754108e-06,
+      "loss": 2.4995,
+      "step": 4367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0883326212385775e-06,
+      "loss": 2.4483,
+      "step": 4368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.086325263481303e-06,
+      "loss": 2.6501,
+      "step": 4369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.084317891805928e-06,
+      "loss": 2.5364,
+      "step": 4370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0823105065360976e-06,
+      "loss": 2.3887,
+      "step": 4371
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.080303107995461e-06,
+      "loss": 2.6551,
+      "step": 4372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.078295696507668e-06,
+      "loss": 2.5312,
+      "step": 4373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.076288272396369e-06,
+      "loss": 2.6157,
+      "step": 4374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.07428083598522e-06,
+      "loss": 2.6937,
+      "step": 4375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.072273387597877e-06,
+      "loss": 2.6244,
+      "step": 4376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.070265927557997e-06,
+      "loss": 2.6403,
+      "step": 4377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.068258456189241e-06,
+      "loss": 2.5331,
+      "step": 4378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.066250973815271e-06,
+      "loss": 2.6135,
+      "step": 4379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.064243480759749e-06,
+      "loss": 2.5462,
+      "step": 4380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0622359773463416e-06,
+      "loss": 2.407,
+      "step": 4381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0602284638987145e-06,
+      "loss": 2.2708,
+      "step": 4382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.058220940740537e-06,
+      "loss": 2.505,
+      "step": 4383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.05621340819548e-06,
+      "loss": 2.5065,
+      "step": 4384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0542058665872155e-06,
+      "loss": 2.3628,
+      "step": 4385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.052198316239414e-06,
+      "loss": 2.6275,
+      "step": 4386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.050190757475751e-06,
+      "loss": 2.6426,
+      "step": 4387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.048183190619904e-06,
+      "loss": 2.5916,
+      "step": 4388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.04617561599555e-06,
+      "loss": 2.6117,
+      "step": 4389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0441680339263655e-06,
+      "loss": 2.5506,
+      "step": 4390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.042160444736032e-06,
+      "loss": 2.3557,
+      "step": 4391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.040152848748228e-06,
+      "loss": 2.4903,
+      "step": 4392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.038145246286636e-06,
+      "loss": 2.4843,
+      "step": 4393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.036137637674943e-06,
+      "loss": 2.6772,
+      "step": 4394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.034130023236827e-06,
+      "loss": 2.5635,
+      "step": 4395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.032122403295977e-06,
+      "loss": 2.6972,
+      "step": 4396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.030114778176077e-06,
+      "loss": 2.4733,
+      "step": 4397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.028107148200812e-06,
+      "loss": 2.4518,
+      "step": 4398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.026099513693873e-06,
+      "loss": 2.5072,
+      "step": 4399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0240918749789455e-06,
+      "loss": 2.5627,
+      "step": 4400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.022084232379719e-06,
+      "loss": 2.6159,
+      "step": 4401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.020076586219884e-06,
+      "loss": 2.6684,
+      "step": 4402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.018068936823127e-06,
+      "loss": 2.4774,
+      "step": 4403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.016061284513142e-06,
+      "loss": 2.6553,
+      "step": 4404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.014053629613619e-06,
+      "loss": 2.5731,
+      "step": 4405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.01204597244825e-06,
+      "loss": 2.5659,
+      "step": 4406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.010038313340725e-06,
+      "loss": 2.3407,
+      "step": 4407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.008030652614737e-06,
+      "loss": 2.5823,
+      "step": 4408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.006022990593977e-06,
+      "loss": 2.6064,
+      "step": 4409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.004015327602139e-06,
+      "loss": 2.7604,
+      "step": 4410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0020076639629165e-06,
+      "loss": 2.4973,
+      "step": 4411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5e-06,
+      "loss": 2.5491,
+      "step": 4412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.997992336037085e-06,
+      "loss": 2.513,
+      "step": 4413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.995984672397863e-06,
+      "loss": 2.4393,
+      "step": 4414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.993977009406024e-06,
+      "loss": 2.6177,
+      "step": 4415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.991969347385266e-06,
+      "loss": 2.4957,
+      "step": 4416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.989961686659278e-06,
+      "loss": 2.3424,
+      "step": 4417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.987954027551751e-06,
+      "loss": 2.484,
+      "step": 4418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.985946370386381e-06,
+      "loss": 2.562,
+      "step": 4419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.983938715486858e-06,
+      "loss": 2.4753,
+      "step": 4420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9819310631768734e-06,
+      "loss": 2.6078,
+      "step": 4421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.979923413780117e-06,
+      "loss": 2.3461,
+      "step": 4422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.977915767620282e-06,
+      "loss": 2.5036,
+      "step": 4423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.975908125021055e-06,
+      "loss": 2.3665,
+      "step": 4424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.973900486306129e-06,
+      "loss": 2.5325,
+      "step": 4425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9718928517991885e-06,
+      "loss": 2.3529,
+      "step": 4426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.969885221823925e-06,
+      "loss": 2.6802,
+      "step": 4427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.967877596704026e-06,
+      "loss": 2.4688,
+      "step": 4428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.965869976763173e-06,
+      "loss": 2.6919,
+      "step": 4429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.963862362325058e-06,
+      "loss": 2.4719,
+      "step": 4430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.961854753713364e-06,
+      "loss": 2.4536,
+      "step": 4431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.959847151251774e-06,
+      "loss": 2.4291,
+      "step": 4432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.95783955526397e-06,
+      "loss": 2.6364,
+      "step": 4433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.955831966073636e-06,
+      "loss": 2.5107,
+      "step": 4434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.953824384004451e-06,
+      "loss": 2.5304,
+      "step": 4435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.951816809380098e-06,
+      "loss": 2.4195,
+      "step": 4436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.949809242524251e-06,
+      "loss": 2.4153,
+      "step": 4437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.947801683760588e-06,
+      "loss": 2.4473,
+      "step": 4438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.945794133412787e-06,
+      "loss": 2.4882,
+      "step": 4439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.94378659180452e-06,
+      "loss": 2.5923,
+      "step": 4440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9417790592594625e-06,
+      "loss": 2.4903,
+      "step": 4441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.939771536101286e-06,
+      "loss": 2.5608,
+      "step": 4442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.93776402265366e-06,
+      "loss": 2.5445,
+      "step": 4443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.935756519240253e-06,
+      "loss": 2.4062,
+      "step": 4444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.933749026184731e-06,
+      "loss": 2.5928,
+      "step": 4445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.93174154381076e-06,
+      "loss": 2.6551,
+      "step": 4446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.929734072442004e-06,
+      "loss": 2.589,
+      "step": 4447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9277266124021245e-06,
+      "loss": 2.4513,
+      "step": 4448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925719164014781e-06,
+      "loss": 2.575,
+      "step": 4449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923711727603632e-06,
+      "loss": 2.5154,
+      "step": 4450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9217043034923335e-06,
+      "loss": 2.4726,
+      "step": 4451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.919696892004539e-06,
+      "loss": 2.4604,
+      "step": 4452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.917689493463902e-06,
+      "loss": 2.5113,
+      "step": 4453
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.915682108194073e-06,
+      "loss": 2.5333,
+      "step": 4454
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.913674736518699e-06,
+      "loss": 2.5274,
+      "step": 4455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.911667378761423e-06,
+      "loss": 2.5939,
+      "step": 4456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909660035245893e-06,
+      "loss": 2.4214,
+      "step": 4457
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.907652706295748e-06,
+      "loss": 2.449,
+      "step": 4458
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.905645392234628e-06,
+      "loss": 2.6625,
+      "step": 4459
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.903638093386167e-06,
+      "loss": 2.3883,
+      "step": 4460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.901630810074e-06,
+      "loss": 2.5502,
+      "step": 4461
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.899623542621759e-06,
+      "loss": 2.4975,
+      "step": 4462
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8976162913530715e-06,
+      "loss": 2.5358,
+      "step": 4463
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.895609056591566e-06,
+      "loss": 2.5447,
+      "step": 4464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.893601838660866e-06,
+      "loss": 2.4736,
+      "step": 4465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.891594637884591e-06,
+      "loss": 2.4252,
+      "step": 4466
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8895874545863604e-06,
+      "loss": 2.5349,
+      "step": 4467
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.887580289089788e-06,
+      "loss": 2.3846,
+      "step": 4468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.885573141718487e-06,
+      "loss": 2.4806,
+      "step": 4469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.883566012796068e-06,
+      "loss": 2.6025,
+      "step": 4470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8815589026461356e-06,
+      "loss": 2.7083,
+      "step": 4471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.879551811592295e-06,
+      "loss": 2.7444,
+      "step": 4472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.877544739958145e-06,
+      "loss": 2.467,
+      "step": 4473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.875537688067284e-06,
+      "loss": 2.3173,
+      "step": 4474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8735306562433045e-06,
+      "loss": 2.4096,
+      "step": 4475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.871523644809802e-06,
+      "loss": 2.4586,
+      "step": 4476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.869516654090361e-06,
+      "loss": 2.4228,
+      "step": 4477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.867509684408564e-06,
+      "loss": 2.6349,
+      "step": 4478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.865502736087995e-06,
+      "loss": 2.3917,
+      "step": 4479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.86349580945223e-06,
+      "loss": 2.6994,
+      "step": 4480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.861488904824843e-06,
+      "loss": 2.6121,
+      "step": 4481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.859482022529406e-06,
+      "loss": 2.4559,
+      "step": 4482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.857475162889483e-06,
+      "loss": 2.4448,
+      "step": 4483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.855468326228638e-06,
+      "loss": 2.5122,
+      "step": 4484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.85346151287043e-06,
+      "loss": 2.5293,
+      "step": 4485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8514547231384164e-06,
+      "loss": 2.6188,
+      "step": 4486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.84944795735615e-06,
+      "loss": 2.5477,
+      "step": 4487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.847441215847177e-06,
+      "loss": 2.5933,
+      "step": 4488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.845434498935041e-06,
+      "loss": 2.4475,
+      "step": 4489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.843427806943283e-06,
+      "loss": 2.3603,
+      "step": 4490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.841421140195438e-06,
+      "loss": 2.7032,
+      "step": 4491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.839414499015041e-06,
+      "loss": 2.4853,
+      "step": 4492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.837407883725616e-06,
+      "loss": 2.4529,
+      "step": 4493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.83540129465069e-06,
+      "loss": 2.4536,
+      "step": 4494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.833394732113781e-06,
+      "loss": 2.3853,
+      "step": 4495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.831388196438402e-06,
+      "loss": 2.5134,
+      "step": 4496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.829381687948068e-06,
+      "loss": 2.4522,
+      "step": 4497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.827375206966284e-06,
+      "loss": 2.5825,
+      "step": 4498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.825368753816552e-06,
+      "loss": 2.5482,
+      "step": 4499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.82336232882237e-06,
+      "loss": 2.4532,
+      "step": 4500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.821355932307231e-06,
+      "loss": 2.4695,
+      "step": 4501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.819349564594622e-06,
+      "loss": 2.5487,
+      "step": 4502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8173432260080285e-06,
+      "loss": 2.6335,
+      "step": 4503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.815336916870929e-06,
+      "loss": 2.4739,
+      "step": 4504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.813330637506799e-06,
+      "loss": 2.572,
+      "step": 4505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.811324388239106e-06,
+      "loss": 2.6269,
+      "step": 4506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.809318169391314e-06,
+      "loss": 2.5877,
+      "step": 4507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.807311981286888e-06,
+      "loss": 2.622,
+      "step": 4508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.805305824249277e-06,
+      "loss": 2.4769,
+      "step": 4509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.803299698601935e-06,
+      "loss": 2.5003,
+      "step": 4510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.801293604668306e-06,
+      "loss": 2.4463,
+      "step": 4511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.799287542771828e-06,
+      "loss": 2.4727,
+      "step": 4512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.797281513235937e-06,
+      "loss": 2.6627,
+      "step": 4513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.79527551638406e-06,
+      "loss": 2.5732,
+      "step": 4514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.793269552539625e-06,
+      "loss": 2.4735,
+      "step": 4515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.791263622026048e-06,
+      "loss": 2.5869,
+      "step": 4516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7892577251667426e-06,
+      "loss": 2.633,
+      "step": 4517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7872518622851155e-06,
+      "loss": 2.612,
+      "step": 4518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.785246033704572e-06,
+      "loss": 2.6092,
+      "step": 4519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.783240239748509e-06,
+      "loss": 2.5165,
+      "step": 4520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.781234480740314e-06,
+      "loss": 2.6296,
+      "step": 4521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.779228757003377e-06,
+      "loss": 2.5957,
+      "step": 4522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.777223068861077e-06,
+      "loss": 2.4699,
+      "step": 4523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.775217416636786e-06,
+      "loss": 2.4819,
+      "step": 4524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.773211800653875e-06,
+      "loss": 2.5784,
+      "step": 4525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7712062212357045e-06,
+      "loss": 2.3416,
+      "step": 4526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.769200678705633e-06,
+      "loss": 2.4512,
+      "step": 4527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.76719517338701e-06,
+      "loss": 2.3488,
+      "step": 4528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.76518970560318e-06,
+      "loss": 2.6528,
+      "step": 4529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.763184275677484e-06,
+      "loss": 2.7052,
+      "step": 4530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.761178883933254e-06,
+      "loss": 2.521,
+      "step": 4531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7591735306938144e-06,
+      "loss": 2.5912,
+      "step": 4532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.757168216282488e-06,
+      "loss": 2.4897,
+      "step": 4533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.755162941022588e-06,
+      "loss": 2.4647,
+      "step": 4534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7531577052374215e-06,
+      "loss": 2.4967,
+      "step": 4535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.75115250925029e-06,
+      "loss": 2.3062,
+      "step": 4536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.749147353384487e-06,
+      "loss": 2.4538,
+      "step": 4537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.747142237963305e-06,
+      "loss": 2.3927,
+      "step": 4538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7451371633100224e-06,
+      "loss": 2.5665,
+      "step": 4539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7431321297479135e-06,
+      "loss": 2.5967,
+      "step": 4540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7411271376002515e-06,
+      "loss": 2.7084,
+      "step": 4541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.739122187190296e-06,
+      "loss": 2.5987,
+      "step": 4542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.737117278841303e-06,
+      "loss": 2.3484,
+      "step": 4543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7351124128765185e-06,
+      "loss": 2.6684,
+      "step": 4544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.733107589619188e-06,
+      "loss": 2.4695,
+      "step": 4545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7311028093925445e-06,
+      "loss": 2.6274,
+      "step": 4546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729098072519816e-06,
+      "loss": 2.4477,
+      "step": 4547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727093379324222e-06,
+      "loss": 2.4511,
+      "step": 4548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.725088730128978e-06,
+      "loss": 2.6093,
+      "step": 4549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.723084125257291e-06,
+      "loss": 2.5035,
+      "step": 4550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.721079565032357e-06,
+      "loss": 2.5606,
+      "step": 4551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.719075049777375e-06,
+      "loss": 2.5483,
+      "step": 4552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.717070579815526e-06,
+      "loss": 2.4872,
+      "step": 4553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.715066155469989e-06,
+      "loss": 2.4046,
+      "step": 4554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.713061777063933e-06,
+      "loss": 2.6931,
+      "step": 4555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.711057444920522e-06,
+      "loss": 2.5915,
+      "step": 4556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7090531593629125e-06,
+      "loss": 2.6586,
+      "step": 4557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7070489207142525e-06,
+      "loss": 2.4936,
+      "step": 4558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.705044729297681e-06,
+      "loss": 2.7177,
+      "step": 4559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.703040585436331e-06,
+      "loss": 2.488,
+      "step": 4560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7010364894533304e-06,
+      "loss": 2.5838,
+      "step": 4561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.699032441671794e-06,
+      "loss": 2.5047,
+      "step": 4562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.697028442414831e-06,
+      "loss": 2.456,
+      "step": 4563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6950244920055475e-06,
+      "loss": 2.3813,
+      "step": 4564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.693020590767035e-06,
+      "loss": 2.4847,
+      "step": 4565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.69101673902238e-06,
+      "loss": 2.7934,
+      "step": 4566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6890129370946595e-06,
+      "loss": 2.5712,
+      "step": 4567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.687009185306945e-06,
+      "loss": 2.6099,
+      "step": 4568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.685005483982299e-06,
+      "loss": 2.573,
+      "step": 4569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.683001833443774e-06,
+      "loss": 2.4833,
+      "step": 4570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.680998234014416e-06,
+      "loss": 2.7227,
+      "step": 4571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.678994686017263e-06,
+      "loss": 2.6348,
+      "step": 4572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6769911897753456e-06,
+      "loss": 2.3681,
+      "step": 4573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.67498774561168e-06,
+      "loss": 2.5644,
+      "step": 4574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.672984353849285e-06,
+      "loss": 2.604,
+      "step": 4575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.670981014811161e-06,
+      "loss": 2.5612,
+      "step": 4576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.668977728820303e-06,
+      "loss": 2.4682,
+      "step": 4577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.666974496199701e-06,
+      "loss": 2.5893,
+      "step": 4578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.664971317272329e-06,
+      "loss": 2.5033,
+      "step": 4579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.662968192361161e-06,
+      "loss": 2.5951,
+      "step": 4580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.660965121789156e-06,
+      "loss": 2.5869,
+      "step": 4581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.658962105879266e-06,
+      "loss": 2.6145,
+      "step": 4582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.656959144954434e-06,
+      "loss": 2.361,
+      "step": 4583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6549562393375965e-06,
+      "loss": 2.4994,
+      "step": 4584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.652953389351676e-06,
+      "loss": 2.5968,
+      "step": 4585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.650950595319593e-06,
+      "loss": 2.4734,
+      "step": 4586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6489478575642535e-06,
+      "loss": 2.493,
+      "step": 4587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.646945176408555e-06,
+      "loss": 2.5463,
+      "step": 4588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.64494255217539e-06,
+      "loss": 2.4834,
+      "step": 4589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.642939985187634e-06,
+      "loss": 2.5488,
+      "step": 4590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.640937475768162e-06,
+      "loss": 2.6648,
+      "step": 4591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6389350242398354e-06,
+      "loss": 2.6388,
+      "step": 4592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.636932630925505e-06,
+      "loss": 2.5063,
+      "step": 4593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.634930296148014e-06,
+      "loss": 2.485,
+      "step": 4594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.632928020230196e-06,
+      "loss": 2.6323,
+      "step": 4595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.630925803494877e-06,
+      "loss": 2.5163,
+      "step": 4596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.628923646264869e-06,
+      "loss": 2.4747,
+      "step": 4597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6269215488629795e-06,
+      "loss": 2.6745,
+      "step": 4598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.624919511612003e-06,
+      "loss": 2.5249,
+      "step": 4599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.622917534834724e-06,
+      "loss": 2.4862,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.62091561885392e-06,
+      "loss": 2.4027,
+      "step": 4601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.618913763992355e-06,
+      "loss": 2.5152,
+      "step": 4602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.616911970572788e-06,
+      "loss": 2.4695,
+      "step": 4603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.614910238917963e-06,
+      "loss": 2.5852,
+      "step": 4604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.612908569350618e-06,
+      "loss": 2.566,
+      "step": 4605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6109069621934765e-06,
+      "loss": 2.6288,
+      "step": 4606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.608905417769258e-06,
+      "loss": 2.5299,
+      "step": 4607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.606903936400667e-06,
+      "loss": 2.7037,
+      "step": 4608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.604902518410402e-06,
+      "loss": 2.6379,
+      "step": 4609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.602901164121145e-06,
+      "loss": 2.4958,
+      "step": 4610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6008998738555746e-06,
+      "loss": 2.5981,
+      "step": 4611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.598898647936354e-06,
+      "loss": 2.3219,
+      "step": 4612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5968974866861394e-06,
+      "loss": 2.5228,
+      "step": 4613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.594896390427574e-06,
+      "loss": 2.4718,
+      "step": 4614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.592895359483294e-06,
+      "loss": 2.459,
+      "step": 4615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.59089439417592e-06,
+      "loss": 2.4922,
+      "step": 4616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.588893494828066e-06,
+      "loss": 2.5695,
+      "step": 4617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.586892661762333e-06,
+      "loss": 2.5564,
+      "step": 4618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.584891895301315e-06,
+      "loss": 2.53,
+      "step": 4619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.582891195767591e-06,
+      "loss": 2.4467,
+      "step": 4620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5808905634837314e-06,
+      "loss": 2.4923,
+      "step": 4621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.578889998772296e-06,
+      "loss": 2.511,
+      "step": 4622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.576889501955833e-06,
+      "loss": 2.5366,
+      "step": 4623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5748890733568775e-06,
+      "loss": 2.6332,
+      "step": 4624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.572888713297957e-06,
+      "loss": 2.6805,
+      "step": 4625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.570888422101588e-06,
+      "loss": 2.6554,
+      "step": 4626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.568888200090274e-06,
+      "loss": 2.5627,
+      "step": 4627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5668880475865074e-06,
+      "loss": 2.538,
+      "step": 4628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.564887964912768e-06,
+      "loss": 2.4616,
+      "step": 4629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.562887952391531e-06,
+      "loss": 2.5926,
+      "step": 4630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5608880103452525e-06,
+      "loss": 2.4931,
+      "step": 4631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5588881390963804e-06,
+      "loss": 2.3971,
+      "step": 4632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5568883389673525e-06,
+      "loss": 2.5567,
+      "step": 4633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.554888610280593e-06,
+      "loss": 2.5392,
+      "step": 4634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.552888953358515e-06,
+      "loss": 2.505,
+      "step": 4635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.55088936852352e-06,
+      "loss": 2.4855,
+      "step": 4636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.548889856098e-06,
+      "loss": 2.4606,
+      "step": 4637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5468904164043324e-06,
+      "loss": 2.5365,
+      "step": 4638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.544891049764884e-06,
+      "loss": 2.4926,
+      "step": 4639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.542891756502008e-06,
+      "loss": 2.5628,
+      "step": 4640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.540892536938051e-06,
+      "loss": 2.6502,
+      "step": 4641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5388933913953434e-06,
+      "loss": 2.5149,
+      "step": 4642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.536894320196203e-06,
+      "loss": 2.6246,
+      "step": 4643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.534895323662939e-06,
+      "loss": 2.4287,
+      "step": 4644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.532896402117847e-06,
+      "loss": 2.5406,
+      "step": 4645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5308975558832095e-06,
+      "loss": 2.6834,
+      "step": 4646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.528898785281297e-06,
+      "loss": 2.6195,
+      "step": 4647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.526900090634368e-06,
+      "loss": 2.514,
+      "step": 4648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.524901472264672e-06,
+      "loss": 2.5704,
+      "step": 4649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.522902930494442e-06,
+      "loss": 2.4946,
+      "step": 4650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.520904465645897e-06,
+      "loss": 2.613,
+      "step": 4651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.518906078041252e-06,
+      "loss": 2.5978,
+      "step": 4652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.516907768002701e-06,
+      "loss": 2.4853,
+      "step": 4653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.51490953585243e-06,
+      "loss": 2.4376,
+      "step": 4654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5129113819126095e-06,
+      "loss": 2.5481,
+      "step": 4655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.510913306505401e-06,
+      "loss": 2.4397,
+      "step": 4656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.508915309952951e-06,
+      "loss": 2.6523,
+      "step": 4657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.506917392577393e-06,
+      "loss": 2.4729,
+      "step": 4658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.504919554700846e-06,
+      "loss": 2.4852,
+      "step": 4659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.502921796645424e-06,
+      "loss": 2.4196,
+      "step": 4660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.50092411873322e-06,
+      "loss": 2.5954,
+      "step": 4661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.498926521286315e-06,
+      "loss": 2.4166,
+      "step": 4662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.49692900462678e-06,
+      "loss": 2.6521,
+      "step": 4663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.494931569076674e-06,
+      "loss": 2.6387,
+      "step": 4664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.492934214958038e-06,
+      "loss": 2.3802,
+      "step": 4665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.490936942592904e-06,
+      "loss": 2.516,
+      "step": 4666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.488939752303288e-06,
+      "loss": 2.5003,
+      "step": 4667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.486942644411197e-06,
+      "loss": 2.5952,
+      "step": 4668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.484945619238619e-06,
+      "loss": 2.4896,
+      "step": 4669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.482948677107533e-06,
+      "loss": 2.5625,
+      "step": 4670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.480951818339902e-06,
+      "loss": 2.7392,
+      "step": 4671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.478955043257679e-06,
+      "loss": 2.4575,
+      "step": 4672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.476958352182798e-06,
+      "loss": 2.4935,
+      "step": 4673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.474961745437183e-06,
+      "loss": 2.6235,
+      "step": 4674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.472965223342748e-06,
+      "loss": 2.46,
+      "step": 4675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4709687862213866e-06,
+      "loss": 2.5234,
+      "step": 4676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.468972434394982e-06,
+      "loss": 2.5119,
+      "step": 4677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4669761681854025e-06,
+      "loss": 2.6826,
+      "step": 4678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.464979987914504e-06,
+      "loss": 2.5457,
+      "step": 4679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.462983893904128e-06,
+      "loss": 2.628,
+      "step": 4680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4609878864761024e-06,
+      "loss": 2.4302,
+      "step": 4681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.458991965952238e-06,
+      "loss": 2.5833,
+      "step": 4682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.456996132654337e-06,
+      "loss": 2.6866,
+      "step": 4683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.455000386904185e-06,
+      "loss": 2.4904,
+      "step": 4684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4530047290235484e-06,
+      "loss": 2.6688,
+      "step": 4685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.451009159334191e-06,
+      "loss": 2.5853,
+      "step": 4686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.449013678157853e-06,
+      "loss": 2.4731,
+      "step": 4687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.447018285816263e-06,
+      "loss": 2.3941,
+      "step": 4688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.445022982631134e-06,
+      "loss": 2.5579,
+      "step": 4689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.443027768924168e-06,
+      "loss": 2.2959,
+      "step": 4690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.441032645017048e-06,
+      "loss": 2.4759,
+      "step": 4691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.439037611231448e-06,
+      "loss": 2.5521,
+      "step": 4692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.437042667889024e-06,
+      "loss": 2.4141,
+      "step": 4693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.435047815311414e-06,
+      "loss": 2.422,
+      "step": 4694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4330530538202496e-06,
+      "loss": 2.3777,
+      "step": 4695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.43105838373714e-06,
+      "loss": 2.4974,
+      "step": 4696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.429063805383687e-06,
+      "loss": 2.5579,
+      "step": 4697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4270693190814716e-06,
+      "loss": 2.5238,
+      "step": 4698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4250749251520625e-06,
+      "loss": 2.5379,
+      "step": 4699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.423080623917012e-06,
+      "loss": 2.3751,
+      "step": 4700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.421086415697858e-06,
+      "loss": 2.463,
+      "step": 4701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.419092300816127e-06,
+      "loss": 2.5634,
+      "step": 4702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.417098279593324e-06,
+      "loss": 2.6424,
+      "step": 4703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.415104352350944e-06,
+      "loss": 2.6202,
+      "step": 4704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.413110519410465e-06,
+      "loss": 2.6464,
+      "step": 4705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.411116781093348e-06,
+      "loss": 2.4334,
+      "step": 4706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.409123137721042e-06,
+      "loss": 2.707,
+      "step": 4707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.40712958961498e-06,
+      "loss": 2.5717,
+      "step": 4708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.405136137096579e-06,
+      "loss": 2.6126,
+      "step": 4709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.40314278048724e-06,
+      "loss": 2.4792,
+      "step": 4710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.401149520108348e-06,
+      "loss": 2.493,
+      "step": 4711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.399156356281274e-06,
+      "loss": 2.4694,
+      "step": 4712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.397163289327375e-06,
+      "loss": 2.5274,
+      "step": 4713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.395170319567988e-06,
+      "loss": 2.2966,
+      "step": 4714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.393177447324439e-06,
+      "loss": 2.4796,
+      "step": 4715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.391184672918034e-06,
+      "loss": 2.5127,
+      "step": 4716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3891919966700645e-06,
+      "loss": 2.6744,
+      "step": 4717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3871994189018075e-06,
+      "loss": 2.5703,
+      "step": 4718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3852069399345245e-06,
+      "loss": 2.6238,
+      "step": 4719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.383214560089461e-06,
+      "loss": 2.4545,
+      "step": 4720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.381222279687845e-06,
+      "loss": 2.6328,
+      "step": 4721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.379230099050888e-06,
+      "loss": 2.5598,
+      "step": 4722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.377238018499786e-06,
+      "loss": 2.4774,
+      "step": 4723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3752460383557195e-06,
+      "loss": 2.579,
+      "step": 4724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.373254158939855e-06,
+      "loss": 2.6955,
+      "step": 4725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.371262380573338e-06,
+      "loss": 2.4979,
+      "step": 4726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.369270703577302e-06,
+      "loss": 2.5012,
+      "step": 4727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.36727912827286e-06,
+      "loss": 2.4457,
+      "step": 4728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3652876549811105e-06,
+      "loss": 2.4972,
+      "step": 4729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.363296284023139e-06,
+      "loss": 2.6137,
+      "step": 4730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3613050157200095e-06,
+      "loss": 2.5875,
+      "step": 4731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3593138503927725e-06,
+      "loss": 2.3217,
+      "step": 4732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3573227883624605e-06,
+      "loss": 2.4437,
+      "step": 4733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.355331829950088e-06,
+      "loss": 2.3815,
+      "step": 4734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3533409754766555e-06,
+      "loss": 2.6557,
+      "step": 4735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.351350225263147e-06,
+      "loss": 2.4646,
+      "step": 4736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.349359579630526e-06,
+      "loss": 2.4214,
+      "step": 4737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.347369038899744e-06,
+      "loss": 2.4911,
+      "step": 4738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.34537860339173e-06,
+      "loss": 2.4061,
+      "step": 4739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3433882734274e-06,
+      "loss": 2.6639,
+      "step": 4740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.341398049327654e-06,
+      "loss": 2.5669,
+      "step": 4741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.339407931413372e-06,
+      "loss": 2.4022,
+      "step": 4742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.337417920005418e-06,
+      "loss": 2.532,
+      "step": 4743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3354280154246385e-06,
+      "loss": 2.5382,
+      "step": 4744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.333438217991864e-06,
+      "loss": 2.7162,
+      "step": 4745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.331448528027905e-06,
+      "loss": 2.4285,
+      "step": 4746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.329458945853555e-06,
+      "loss": 2.4565,
+      "step": 4747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.327469471789597e-06,
+      "loss": 2.4436,
+      "step": 4748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.325480106156787e-06,
+      "loss": 2.5075,
+      "step": 4749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.323490849275868e-06,
+      "loss": 2.415,
+      "step": 4750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.321501701467564e-06,
+      "loss": 2.2453,
+      "step": 4751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.319512663052585e-06,
+      "loss": 2.4952,
+      "step": 4752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.317523734351621e-06,
+      "loss": 2.5837,
+      "step": 4753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.315534915685341e-06,
+      "loss": 2.5487,
+      "step": 4754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.313546207374404e-06,
+      "loss": 2.4302,
+      "step": 4755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.311557609739442e-06,
+      "loss": 2.4719,
+      "step": 4756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.309569123101077e-06,
+      "loss": 2.6649,
+      "step": 4757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.30758074777991e-06,
+      "loss": 2.4569,
+      "step": 4758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3055924840965204e-06,
+      "loss": 2.5502,
+      "step": 4759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.303604332371476e-06,
+      "loss": 2.3928,
+      "step": 4760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.301616292925324e-06,
+      "loss": 2.3291,
+      "step": 4761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.299628366078592e-06,
+      "loss": 2.4832,
+      "step": 4762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.297640552151789e-06,
+      "loss": 2.5048,
+      "step": 4763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.295652851465412e-06,
+      "loss": 2.462,
+      "step": 4764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2936652643399315e-06,
+      "loss": 2.5956,
+      "step": 4765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.291677791095805e-06,
+      "loss": 2.5795,
+      "step": 4766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2896904320534704e-06,
+      "loss": 2.6372,
+      "step": 4767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.287703187533346e-06,
+      "loss": 2.5446,
+      "step": 4768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.285716057855831e-06,
+      "loss": 2.5414,
+      "step": 4769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.283729043341308e-06,
+      "loss": 2.3886,
+      "step": 4770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.281742144310144e-06,
+      "loss": 2.4189,
+      "step": 4771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.27975536108268e-06,
+      "loss": 2.3943,
+      "step": 4772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.277768693979244e-06,
+      "loss": 2.5218,
+      "step": 4773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.275782143320141e-06,
+      "loss": 2.5302,
+      "step": 4774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.273795709425664e-06,
+      "loss": 2.5052,
+      "step": 4775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.271809392616081e-06,
+      "loss": 2.6473,
+      "step": 4776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.269823193211641e-06,
+      "loss": 2.5813,
+      "step": 4777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.26783711153258e-06,
+      "loss": 2.5097,
+      "step": 4778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.265851147899109e-06,
+      "loss": 2.4427,
+      "step": 4779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.263865302631423e-06,
+      "loss": 2.5225,
+      "step": 4780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.261879576049695e-06,
+      "loss": 2.5665,
+      "step": 4781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.259893968474082e-06,
+      "loss": 2.5526,
+      "step": 4782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.257908480224722e-06,
+      "loss": 2.5995,
+      "step": 4783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2559231116217305e-06,
+      "loss": 2.4624,
+      "step": 4784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.253937862985205e-06,
+      "loss": 2.5096,
+      "step": 4785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.251952734635228e-06,
+      "loss": 2.5136,
+      "step": 4786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.249967726891857e-06,
+      "loss": 2.3949,
+      "step": 4787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.24798284007513e-06,
+      "loss": 2.6419,
+      "step": 4788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.245998074505072e-06,
+      "loss": 2.4746,
+      "step": 4789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.244013430501681e-06,
+      "loss": 2.5046,
+      "step": 4790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.242028908384937e-06,
+      "loss": 2.5395,
+      "step": 4791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.240044508474805e-06,
+      "loss": 2.5844,
+      "step": 4792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.238060231091224e-06,
+      "loss": 2.4795,
+      "step": 4793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.236076076554119e-06,
+      "loss": 2.5622,
+      "step": 4794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.23409204518339e-06,
+      "loss": 2.6337,
+      "step": 4795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2321081372989195e-06,
+      "loss": 2.5716,
+      "step": 4796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2301243532205726e-06,
+      "loss": 2.4417,
+      "step": 4797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.228140693268191e-06,
+      "loss": 2.47,
+      "step": 4798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.226157157761598e-06,
+      "loss": 2.5784,
+      "step": 4799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.224173747020593e-06,
+      "loss": 2.5336,
+      "step": 4800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.222190461364963e-06,
+      "loss": 2.3404,
+      "step": 4801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.220207301114467e-06,
+      "loss": 2.5454,
+      "step": 4802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.21822426658885e-06,
+      "loss": 2.7036,
+      "step": 4803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.216241358107831e-06,
+      "loss": 2.5493,
+      "step": 4804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.214258575991111e-06,
+      "loss": 2.4355,
+      "step": 4805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.212275920558375e-06,
+      "loss": 2.6433,
+      "step": 4806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.21029339212928e-06,
+      "loss": 2.4853,
+      "step": 4807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.208310991023469e-06,
+      "loss": 2.7157,
+      "step": 4808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.206328717560562e-06,
+      "loss": 2.6732,
+      "step": 4809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.204346572060156e-06,
+      "loss": 2.4377,
+      "step": 4810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.20236455484183e-06,
+      "loss": 2.6815,
+      "step": 4811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.200382666225141e-06,
+      "loss": 2.5664,
+      "step": 4812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1984009065296285e-06,
+      "loss": 2.7265,
+      "step": 4813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1964192760748085e-06,
+      "loss": 2.5134,
+      "step": 4814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1944377751801744e-06,
+      "loss": 2.3888,
+      "step": 4815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.192456404165202e-06,
+      "loss": 2.5797,
+      "step": 4816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.190475163349345e-06,
+      "loss": 2.5108,
+      "step": 4817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.188494053052034e-06,
+      "loss": 2.4257,
+      "step": 4818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.186513073592686e-06,
+      "loss": 2.5396,
+      "step": 4819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.184532225290687e-06,
+      "loss": 2.446,
+      "step": 4820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.182551508465408e-06,
+      "loss": 2.5204,
+      "step": 4821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.180570923436196e-06,
+      "loss": 2.4695,
+      "step": 4822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1785904705223776e-06,
+      "loss": 2.4724,
+      "step": 4823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.176610150043261e-06,
+      "loss": 2.4487,
+      "step": 4824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.174629962318129e-06,
+      "loss": 2.3325,
+      "step": 4825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.172649907666244e-06,
+      "loss": 2.4992,
+      "step": 4826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1706699864068485e-06,
+      "loss": 2.6275,
+      "step": 4827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.16869019885916e-06,
+      "loss": 2.4178,
+      "step": 4828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.166710545342379e-06,
+      "loss": 2.618,
+      "step": 4829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.164731026175682e-06,
+      "loss": 2.587,
+      "step": 4830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.162751641678224e-06,
+      "loss": 2.4629,
+      "step": 4831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.16077239216914e-06,
+      "loss": 2.7639,
+      "step": 4832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.158793277967539e-06,
+      "loss": 2.5509,
+      "step": 4833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.156814299392512e-06,
+      "loss": 2.559,
+      "step": 4834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1548354567631245e-06,
+      "loss": 2.502,
+      "step": 4835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.152856750398426e-06,
+      "loss": 2.6424,
+      "step": 4836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.15087818061744e-06,
+      "loss": 2.5216,
+      "step": 4837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.148899747739168e-06,
+      "loss": 2.5336,
+      "step": 4838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.146921452082587e-06,
+      "loss": 2.5291,
+      "step": 4839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.144943293966659e-06,
+      "loss": 2.5277,
+      "step": 4840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.142965273710317e-06,
+      "loss": 2.4815,
+      "step": 4841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.140987391632477e-06,
+      "loss": 2.4966,
+      "step": 4842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.139009648052029e-06,
+      "loss": 2.3893,
+      "step": 4843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.137032043287841e-06,
+      "loss": 2.6514,
+      "step": 4844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.135054577658761e-06,
+      "loss": 2.4987,
+      "step": 4845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.133077251483609e-06,
+      "loss": 2.6475,
+      "step": 4846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.131100065081191e-06,
+      "loss": 2.4836,
+      "step": 4847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.129123018770285e-06,
+      "loss": 2.3492,
+      "step": 4848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1271461128696464e-06,
+      "loss": 2.5578,
+      "step": 4849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.125169347698009e-06,
+      "loss": 2.5241,
+      "step": 4850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.123192723574083e-06,
+      "loss": 2.7434,
+      "step": 4851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.121216240816559e-06,
+      "loss": 2.5684,
+      "step": 4852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.119239899744101e-06,
+      "loss": 2.4261,
+      "step": 4853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.117263700675353e-06,
+      "loss": 2.6018,
+      "step": 4854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.115287643928934e-06,
+      "loss": 2.529,
+      "step": 4855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.11331172982344e-06,
+      "loss": 2.4285,
+      "step": 4856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.111335958677447e-06,
+      "loss": 2.5076,
+      "step": 4857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1093603308095025e-06,
+      "loss": 2.6507,
+      "step": 4858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.107384846538137e-06,
+      "loss": 2.6842,
+      "step": 4859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.105409506181855e-06,
+      "loss": 2.3477,
+      "step": 4860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.103434310059135e-06,
+      "loss": 2.5301,
+      "step": 4861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.101459258488438e-06,
+      "loss": 2.5024,
+      "step": 4862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.099484351788197e-06,
+      "loss": 2.5225,
+      "step": 4863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.097509590276824e-06,
+      "loss": 2.5166,
+      "step": 4864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.095534974272709e-06,
+      "loss": 2.4833,
+      "step": 4865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.093560504094215e-06,
+      "loss": 2.4851,
+      "step": 4866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.091586180059682e-06,
+      "loss": 2.4024,
+      "step": 4867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.089612002487428e-06,
+      "loss": 2.4603,
+      "step": 4868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.087637971695747e-06,
+      "loss": 2.4883,
+      "step": 4869
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.085664088002911e-06,
+      "loss": 2.6685,
+      "step": 4870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.083690351727164e-06,
+      "loss": 2.5464,
+      "step": 4871
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0817167631867294e-06,
+      "loss": 2.5826,
+      "step": 4872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.079743322699807e-06,
+      "loss": 2.4273,
+      "step": 4873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.077770030584569e-06,
+      "loss": 2.3708,
+      "step": 4874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0757968871591705e-06,
+      "loss": 2.5209,
+      "step": 4875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0738238927417354e-06,
+      "loss": 2.4891,
+      "step": 4876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.07185104765037e-06,
+      "loss": 2.3561,
+      "step": 4877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.069878352203151e-06,
+      "loss": 2.479,
+      "step": 4878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.067905806718133e-06,
+      "loss": 2.5887,
+      "step": 4879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.065933411513349e-06,
+      "loss": 2.4512,
+      "step": 4880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0639611669068025e-06,
+      "loss": 2.4799,
+      "step": 4881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.061989073216478e-06,
+      "loss": 2.562,
+      "step": 4882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0600171307603325e-06,
+      "loss": 2.6175,
+      "step": 4883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0580453398563005e-06,
+      "loss": 2.7528,
+      "step": 4884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.056073700822288e-06,
+      "loss": 2.5791,
+      "step": 4885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.054102213976184e-06,
+      "loss": 2.5083,
+      "step": 4886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.052130879635845e-06,
+      "loss": 2.5872,
+      "step": 4887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.050159698119107e-06,
+      "loss": 2.5085,
+      "step": 4888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.048188669743783e-06,
+      "loss": 2.6802,
+      "step": 4889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.046217794827659e-06,
+      "loss": 2.6018,
+      "step": 4890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.044247073688494e-06,
+      "loss": 2.5118,
+      "step": 4891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.042276506644024e-06,
+      "loss": 2.4972,
+      "step": 4892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.040306094011965e-06,
+      "loss": 2.3447,
+      "step": 4893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.038335836110001e-06,
+      "loss": 2.6201,
+      "step": 4894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.036365733255794e-06,
+      "loss": 2.5011,
+      "step": 4895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.034395785766978e-06,
+      "loss": 2.6591,
+      "step": 4896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.032425993961171e-06,
+      "loss": 2.523,
+      "step": 4897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.030456358155957e-06,
+      "loss": 2.6904,
+      "step": 4898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.028486878668896e-06,
+      "loss": 2.4741,
+      "step": 4899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026517555817527e-06,
+      "loss": 2.6222,
+      "step": 4900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02454838991936e-06,
+      "loss": 2.6375,
+      "step": 4901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02257938129188e-06,
+      "loss": 2.4083,
+      "step": 4902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020610530252548e-06,
+      "loss": 2.5175,
+      "step": 4903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018641837118799e-06,
+      "loss": 2.4851,
+      "step": 4904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016673302208044e-06,
+      "loss": 2.4044,
+      "step": 4905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0147049258376654e-06,
+      "loss": 2.4443,
+      "step": 4906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01273670832502e-06,
+      "loss": 2.6252,
+      "step": 4907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.010768649987446e-06,
+      "loss": 2.3682,
+      "step": 4908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0088007511422485e-06,
+      "loss": 2.3978,
+      "step": 4909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.006833012106707e-06,
+      "loss": 2.5544,
+      "step": 4910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.004865433198078e-06,
+      "loss": 2.5584,
+      "step": 4911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.002898014733593e-06,
+      "loss": 2.4182,
+      "step": 4912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.000930757030456e-06,
+      "loss": 2.3943,
+      "step": 4913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.998963660405843e-06,
+      "loss": 2.6366,
+      "step": 4914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.996996725176908e-06,
+      "loss": 2.6088,
+      "step": 4915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.995029951660777e-06,
+      "loss": 2.5193,
+      "step": 4916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.993063340174549e-06,
+      "loss": 2.5376,
+      "step": 4917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991096891035298e-06,
+      "loss": 2.595,
+      "step": 4918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.989130604560074e-06,
+      "loss": 2.7174,
+      "step": 4919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.987164481065899e-06,
+      "loss": 2.5327,
+      "step": 4920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.985198520869765e-06,
+      "loss": 2.4284,
+      "step": 4921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.983232724288642e-06,
+      "loss": 2.6153,
+      "step": 4922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.981267091639475e-06,
+      "loss": 2.531,
+      "step": 4923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.979301623239177e-06,
+      "loss": 2.6541,
+      "step": 4924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9773363194046405e-06,
+      "loss": 2.4796,
+      "step": 4925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9753711804527266e-06,
+      "loss": 2.5867,
+      "step": 4926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9734062067002716e-06,
+      "loss": 2.467,
+      "step": 4927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.971441398464088e-06,
+      "loss": 2.438,
+      "step": 4928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.969476756060954e-06,
+      "loss": 2.6418,
+      "step": 4929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.967512279807633e-06,
+      "loss": 2.4639,
+      "step": 4930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.965547970020851e-06,
+      "loss": 2.6308,
+      "step": 4931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.963583827017311e-06,
+      "loss": 2.6424,
+      "step": 4932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.961619851113689e-06,
+      "loss": 2.6717,
+      "step": 4933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.959656042626634e-06,
+      "loss": 2.6414,
+      "step": 4934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.957692401872769e-06,
+      "loss": 2.5456,
+      "step": 4935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.955728929168689e-06,
+      "loss": 2.5684,
+      "step": 4936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9537656248309605e-06,
+      "loss": 2.4174,
+      "step": 4937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.951802489176126e-06,
+      "loss": 2.5022,
+      "step": 4938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9498395225206964e-06,
+      "loss": 2.5,
+      "step": 4939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.94787672518116e-06,
+      "loss": 2.6546,
+      "step": 4940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9459140974739755e-06,
+      "loss": 2.5153,
+      "step": 4941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.943951639715576e-06,
+      "loss": 2.5803,
+      "step": 4942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.941989352222366e-06,
+      "loss": 2.5764,
+      "step": 4943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.940027235310719e-06,
+      "loss": 2.4846,
+      "step": 4944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.938065289296986e-06,
+      "loss": 2.4249,
+      "step": 4945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9361035144974905e-06,
+      "loss": 2.4133,
+      "step": 4946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.934141911228525e-06,
+      "loss": 2.543,
+      "step": 4947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.932180479806357e-06,
+      "loss": 2.6736,
+      "step": 4948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.930219220547223e-06,
+      "loss": 2.4867,
+      "step": 4949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.928258133767336e-06,
+      "loss": 2.4432,
+      "step": 4950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.92629721978288e-06,
+      "loss": 2.3723,
+      "step": 4951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.924336478910007e-06,
+      "loss": 2.6019,
+      "step": 4952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.922375911464849e-06,
+      "loss": 2.5161,
+      "step": 4953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.920415517763503e-06,
+      "loss": 2.5733,
+      "step": 4954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.918455298122041e-06,
+      "loss": 2.4679,
+      "step": 4955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.916495252856506e-06,
+      "loss": 2.4922,
+      "step": 4956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.914535382282913e-06,
+      "loss": 2.4435,
+      "step": 4957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.912575686717251e-06,
+      "loss": 2.4271,
+      "step": 4958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.910616166475477e-06,
+      "loss": 2.5095,
+      "step": 4959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.908656821873523e-06,
+      "loss": 2.5107,
+      "step": 4960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.906697653227291e-06,
+      "loss": 2.4476,
+      "step": 4961
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.904738660852654e-06,
+      "loss": 2.4865,
+      "step": 4962
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9027798450654585e-06,
+      "loss": 2.5297,
+      "step": 4963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.900821206181521e-06,
+      "loss": 2.674,
+      "step": 4964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.898862744516631e-06,
+      "loss": 2.5417,
+      "step": 4965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.89690446038655e-06,
+      "loss": 2.4093,
+      "step": 4966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8949463541070065e-06,
+      "loss": 2.4897,
+      "step": 4967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.892988425993703e-06,
+      "loss": 2.525,
+      "step": 4968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.891030676362315e-06,
+      "loss": 2.3902,
+      "step": 4969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.889073105528489e-06,
+      "loss": 2.5884,
+      "step": 4970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.887115713807839e-06,
+      "loss": 2.3973,
+      "step": 4971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.885158501515954e-06,
+      "loss": 2.3552,
+      "step": 4972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.88320146896839e-06,
+      "loss": 2.4136,
+      "step": 4973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.881244616480679e-06,
+      "loss": 2.47,
+      "step": 4974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.879287944368321e-06,
+      "loss": 2.5461,
+      "step": 4975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8773314529467875e-06,
+      "loss": 2.6127,
+      "step": 4976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8753751425315225e-06,
+      "loss": 2.6071,
+      "step": 4977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.873419013437937e-06,
+      "loss": 2.618,
+      "step": 4978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.871463065981415e-06,
+      "loss": 2.496,
+      "step": 4979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.869507300477311e-06,
+      "loss": 2.4832,
+      "step": 4980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.867551717240952e-06,
+      "loss": 2.545,
+      "step": 4981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.865596316587633e-06,
+      "loss": 2.593,
+      "step": 4982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.86364109883262e-06,
+      "loss": 2.4829,
+      "step": 4983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8616860642911516e-06,
+      "loss": 2.5589,
+      "step": 4984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.85973121327843e-06,
+      "loss": 2.5866,
+      "step": 4985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.857776546109641e-06,
+      "loss": 2.4887,
+      "step": 4986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.855822063099927e-06,
+      "loss": 2.4433,
+      "step": 4987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.853867764564409e-06,
+      "loss": 2.6315,
+      "step": 4988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.851913650818177e-06,
+      "loss": 2.5939,
+      "step": 4989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.849959722176287e-06,
+      "loss": 2.4589,
+      "step": 4990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.848005978953769e-06,
+      "loss": 2.6314,
+      "step": 4991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.846052421465623e-06,
+      "loss": 2.5799,
+      "step": 4992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.844099050026819e-06,
+      "loss": 2.5222,
+      "step": 4993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.842145864952295e-06,
+      "loss": 2.4196,
+      "step": 4994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.84019286655696e-06,
+      "loss": 2.413,
+      "step": 4995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.838240055155692e-06,
+      "loss": 2.595,
+      "step": 4996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.836287431063344e-06,
+      "loss": 2.5207,
+      "step": 4997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.834334994594733e-06,
+      "loss": 2.5144,
+      "step": 4998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.832382746064647e-06,
+      "loss": 2.3374,
+      "step": 4999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.830430685787844e-06,
+      "loss": 2.577,
+      "step": 5000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.828478814079054e-06,
+      "loss": 2.7151,
+      "step": 5001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8265271312529714e-06,
+      "loss": 2.5266,
+      "step": 5002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.824575637624265e-06,
+      "loss": 2.5121,
+      "step": 5003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8226243335075715e-06,
+      "loss": 2.4622,
+      "step": 5004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.820673219217497e-06,
+      "loss": 2.4387,
+      "step": 5005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.818722295068616e-06,
+      "loss": 2.5886,
+      "step": 5006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.816771561375473e-06,
+      "loss": 2.5406,
+      "step": 5007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.814821018452583e-06,
+      "loss": 2.592,
+      "step": 5008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.812870666614431e-06,
+      "loss": 2.5259,
+      "step": 5009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8109205061754657e-06,
+      "loss": 2.3736,
+      "step": 5010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8089705374501116e-06,
+      "loss": 2.462,
+      "step": 5011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8070207607527587e-06,
+      "loss": 2.5273,
+      "step": 5012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.805071176397766e-06,
+      "loss": 2.4739,
+      "step": 5013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8031217846994628e-06,
+      "loss": 2.6388,
+      "step": 5014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.801172585972146e-06,
+      "loss": 2.6525,
+      "step": 5015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7992235805300838e-06,
+      "loss": 2.4216,
+      "step": 5016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7972747686875104e-06,
+      "loss": 2.4764,
+      "step": 5017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7953261507586282e-06,
+      "loss": 2.6501,
+      "step": 5018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7933777270576145e-06,
+      "loss": 2.5571,
+      "step": 5019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7914294978986083e-06,
+      "loss": 2.5942,
+      "step": 5020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7894814635957188e-06,
+      "loss": 2.4701,
+      "step": 5021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7875336244630278e-06,
+      "loss": 2.4182,
+      "step": 5022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7855859808145808e-06,
+      "loss": 2.3399,
+      "step": 5023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7836385329643933e-06,
+      "loss": 2.3826,
+      "step": 5024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7816912812264506e-06,
+      "loss": 2.5337,
+      "step": 5025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.779744225914703e-06,
+      "loss": 2.4479,
+      "step": 5026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.777797367343076e-06,
+      "loss": 2.4938,
+      "step": 5027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7758507058254547e-06,
+      "loss": 2.5076,
+      "step": 5028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.773904241675696e-06,
+      "loss": 2.5482,
+      "step": 5029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.77195797520763e-06,
+      "loss": 2.4423,
+      "step": 5030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.770011906735049e-06,
+      "loss": 2.4936,
+      "step": 5031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.768066036571713e-06,
+      "loss": 2.5321,
+      "step": 5032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.766120365031352e-06,
+      "loss": 2.5735,
+      "step": 5033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.764174892427666e-06,
+      "loss": 2.5805,
+      "step": 5034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.76222961907432e-06,
+      "loss": 2.4771,
+      "step": 5035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.760284545284947e-06,
+      "loss": 2.4682,
+      "step": 5036
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7583396713731486e-06,
+      "loss": 2.456,
+      "step": 5037
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7563949976524933e-06,
+      "loss": 2.6913,
+      "step": 5038
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7544505244365204e-06,
+      "loss": 2.5072,
+      "step": 5039
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7525062520387305e-06,
+      "loss": 2.4621,
+      "step": 5040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.750562180772601e-06,
+      "loss": 2.4993,
+      "step": 5041
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7486183109515696e-06,
+      "loss": 2.4674,
+      "step": 5042
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7466746428890434e-06,
+      "loss": 2.5972,
+      "step": 5043
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.744731176898396e-06,
+      "loss": 2.5712,
+      "step": 5044
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7427879132929735e-06,
+      "loss": 2.5569,
+      "step": 5045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7408448523860817e-06,
+      "loss": 2.515,
+      "step": 5046
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.738901994490999e-06,
+      "loss": 2.5363,
+      "step": 5047
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7369593399209704e-06,
+      "loss": 2.4126,
+      "step": 5048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7350168889892046e-06,
+      "loss": 2.5044,
+      "step": 5049
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7330746420088837e-06,
+      "loss": 2.354,
+      "step": 5050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7311325992931517e-06,
+      "loss": 2.3892,
+      "step": 5051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7291907611551197e-06,
+      "loss": 2.5166,
+      "step": 5052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7272491279078716e-06,
+      "loss": 2.7588,
+      "step": 5053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.725307699864452e-06,
+      "loss": 2.5537,
+      "step": 5054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.723366477337875e-06,
+      "loss": 2.5741,
+      "step": 5055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7214254606411193e-06,
+      "loss": 2.5606,
+      "step": 5056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7194846500871353e-06,
+      "loss": 2.5653,
+      "step": 5057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.717544045988835e-06,
+      "loss": 2.4784,
+      "step": 5058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7156036486591006e-06,
+      "loss": 2.5439,
+      "step": 5059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7136634584107787e-06,
+      "loss": 2.461,
+      "step": 5060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.711723475556682e-06,
+      "loss": 2.5265,
+      "step": 5061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.709783700409595e-06,
+      "loss": 2.5148,
+      "step": 5062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.70784413328226e-06,
+      "loss": 2.5461,
+      "step": 5063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.705904774487396e-06,
+      "loss": 2.7058,
+      "step": 5064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.703965624337681e-06,
+      "loss": 2.5522,
+      "step": 5065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7020266831457598e-06,
+      "loss": 2.562,
+      "step": 5066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7000879512242473e-06,
+      "loss": 2.6152,
+      "step": 5067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.69814942888572e-06,
+      "loss": 2.5508,
+      "step": 5068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6962111164427263e-06,
+      "loss": 2.5404,
+      "step": 5069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6942730142077756e-06,
+      "loss": 2.3514,
+      "step": 5070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.692335122493346e-06,
+      "loss": 2.5645,
+      "step": 5071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6903974416118786e-06,
+      "loss": 2.6065,
+      "step": 5072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6884599718757873e-06,
+      "loss": 2.5013,
+      "step": 5073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6865227135974436e-06,
+      "loss": 2.4169,
+      "step": 5074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.684585667089192e-06,
+      "loss": 2.4233,
+      "step": 5075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6826488326633393e-06,
+      "loss": 2.7467,
+      "step": 5076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.680712210632158e-06,
+      "loss": 2.4093,
+      "step": 5077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6787758013078863e-06,
+      "loss": 2.4178,
+      "step": 5078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6768396050027284e-06,
+      "loss": 2.6067,
+      "step": 5079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6749036220288563e-06,
+      "loss": 2.5035,
+      "step": 5080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6729678526984046e-06,
+      "loss": 2.4545,
+      "step": 5081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6710322973234758e-06,
+      "loss": 2.5253,
+      "step": 5082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.669096956216135e-06,
+      "loss": 2.5167,
+      "step": 5083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6671618296884147e-06,
+      "loss": 2.6043,
+      "step": 5084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6652269180523137e-06,
+      "loss": 2.6464,
+      "step": 5085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.663292221619794e-06,
+      "loss": 2.5744,
+      "step": 5086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.661357740702787e-06,
+      "loss": 2.5735,
+      "step": 5087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6594234756131826e-06,
+      "loss": 2.5746,
+      "step": 5088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6574894266628415e-06,
+      "loss": 2.4488,
+      "step": 5089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.655555594163587e-06,
+      "loss": 2.7364,
+      "step": 5090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6536219784272074e-06,
+      "loss": 2.5308,
+      "step": 5091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6516885797654593e-06,
+      "loss": 2.5674,
+      "step": 5092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.64975539849006e-06,
+      "loss": 2.3874,
+      "step": 5093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6478224349126944e-06,
+      "loss": 2.4535,
+      "step": 5094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6458896893450104e-06,
+      "loss": 2.4008,
+      "step": 5095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6439571620986213e-06,
+      "loss": 2.622,
+      "step": 5096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6420248534851075e-06,
+      "loss": 2.5958,
+      "step": 5097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.640092763816013e-06,
+      "loss": 2.5485,
+      "step": 5098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6381608934028447e-06,
+      "loss": 2.5861,
+      "step": 5099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6362292425570754e-06,
+      "loss": 2.4628,
+      "step": 5100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.634297811590143e-06,
+      "loss": 2.5686,
+      "step": 5101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6323666008134465e-06,
+      "loss": 2.4375,
+      "step": 5102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6304356105383566e-06,
+      "loss": 2.3743,
+      "step": 5103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6285048410762018e-06,
+      "loss": 2.5021,
+      "step": 5104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6265742927382778e-06,
+      "loss": 2.4798,
+      "step": 5105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6246439658358446e-06,
+      "loss": 2.5,
+      "step": 5106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6227138606801225e-06,
+      "loss": 2.4777,
+      "step": 5107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.620783977582305e-06,
+      "loss": 2.5858,
+      "step": 5108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6188543168535416e-06,
+      "loss": 2.6944,
+      "step": 5109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.61692487880495e-06,
+      "loss": 2.479,
+      "step": 5110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.61499566374761e-06,
+      "loss": 2.2808,
+      "step": 5111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.613066671992567e-06,
+      "loss": 2.7281,
+      "step": 5112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6111379038508276e-06,
+      "loss": 2.4341,
+      "step": 5113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6092093596333654e-06,
+      "loss": 2.4488,
+      "step": 5114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6072810396511186e-06,
+      "loss": 2.2915,
+      "step": 5115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.605352944214986e-06,
+      "loss": 2.3949,
+      "step": 5116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.603425073635831e-06,
+      "loss": 2.525,
+      "step": 5117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.601497428224481e-06,
+      "loss": 2.3545,
+      "step": 5118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5995700082917307e-06,
+      "loss": 2.5431,
+      "step": 5119
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5976428141483326e-06,
+      "loss": 2.4549,
+      "step": 5120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5957158461050056e-06,
+      "loss": 2.4153,
+      "step": 5121
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5937891044724344e-06,
+      "loss": 2.5595,
+      "step": 5122
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5918625895612622e-06,
+      "loss": 2.4526,
+      "step": 5123
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5899363016821e-06,
+      "loss": 2.5859,
+      "step": 5124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5880102411455185e-06,
+      "loss": 2.5839,
+      "step": 5125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.586084408262056e-06,
+      "loss": 2.568,
+      "step": 5126
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5841588033422103e-06,
+      "loss": 2.5326,
+      "step": 5127
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5822334266964454e-06,
+      "loss": 2.4828,
+      "step": 5128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.580308278635183e-06,
+      "loss": 2.4173,
+      "step": 5129
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5783833594688177e-06,
+      "loss": 2.5672,
+      "step": 5130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5764586695076987e-06,
+      "loss": 2.4434,
+      "step": 5131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5745342090621406e-06,
+      "loss": 2.7237,
+      "step": 5132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.572609978442423e-06,
+      "loss": 2.5052,
+      "step": 5133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5706859779587856e-06,
+      "loss": 2.6683,
+      "step": 5134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5687622079214328e-06,
+      "loss": 2.3648,
+      "step": 5135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.566838668640531e-06,
+      "loss": 2.4437,
+      "step": 5136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5649153604262094e-06,
+      "loss": 2.5803,
+      "step": 5137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5629922835885618e-06,
+      "loss": 2.5502,
+      "step": 5138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.561069438437642e-06,
+      "loss": 2.5065,
+      "step": 5139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5591468252834654e-06,
+      "loss": 2.5241,
+      "step": 5140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.557224444436017e-06,
+      "loss": 2.3322,
+      "step": 5141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.555302296205237e-06,
+      "loss": 2.4034,
+      "step": 5142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.553380380901031e-06,
+      "loss": 2.4135,
+      "step": 5143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.551458698833265e-06,
+      "loss": 2.525,
+      "step": 5144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.549537250311772e-06,
+      "loss": 2.4441,
+      "step": 5145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.547616035646343e-06,
+      "loss": 2.5917,
+      "step": 5146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.545695055146733e-06,
+      "loss": 2.5828,
+      "step": 5147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.543774309122657e-06,
+      "loss": 2.5627,
+      "step": 5148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.541853797883797e-06,
+      "loss": 2.6128,
+      "step": 5149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5399335217397936e-06,
+      "loss": 2.6368,
+      "step": 5150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.53801348100025e-06,
+      "loss": 2.6401,
+      "step": 5151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.536093675974729e-06,
+      "loss": 2.4807,
+      "step": 5152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.534174106972762e-06,
+      "loss": 2.5955,
+      "step": 5153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.532254774303838e-06,
+      "loss": 2.4661,
+      "step": 5154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5303356782774058e-06,
+      "loss": 2.6481,
+      "step": 5155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.528416819202881e-06,
+      "loss": 2.6402,
+      "step": 5156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5264981973896374e-06,
+      "loss": 2.4807,
+      "step": 5157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5245798131470123e-06,
+      "loss": 2.4342,
+      "step": 5158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.522661666784303e-06,
+      "loss": 2.5033,
+      "step": 5159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5207437586107698e-06,
+      "loss": 2.5971,
+      "step": 5160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.518826088935635e-06,
+      "loss": 2.6164,
+      "step": 5161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5169086580680824e-06,
+      "loss": 2.7458,
+      "step": 5162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5149914663172524e-06,
+      "loss": 2.6041,
+      "step": 5163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5130745139922572e-06,
+      "loss": 2.5145,
+      "step": 5164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.511157801402162e-06,
+      "loss": 2.6434,
+      "step": 5165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5092413288559948e-06,
+      "loss": 2.4729,
+      "step": 5166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5073250966627447e-06,
+      "loss": 2.5423,
+      "step": 5167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5054091051313666e-06,
+      "loss": 2.3933,
+      "step": 5168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5034933545707704e-06,
+      "loss": 2.5371,
+      "step": 5169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5015778452898306e-06,
+      "loss": 2.6622,
+      "step": 5170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4996625775973812e-06,
+      "loss": 2.5032,
+      "step": 5171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.497747551802221e-06,
+      "loss": 2.492,
+      "step": 5172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4958327682131032e-06,
+      "loss": 2.5095,
+      "step": 5173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.493918227138746e-06,
+      "loss": 2.4737,
+      "step": 5174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4920039288878316e-06,
+      "loss": 2.5062,
+      "step": 5175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4900898737689973e-06,
+      "loss": 2.4692,
+      "step": 5176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4881760620908444e-06,
+      "loss": 2.3774,
+      "step": 5177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4862624941619316e-06,
+      "loss": 2.6447,
+      "step": 5178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.484349170290784e-06,
+      "loss": 2.3787,
+      "step": 5179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4824360907858824e-06,
+      "loss": 2.3746,
+      "step": 5180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.48052325595567e-06,
+      "loss": 2.4904,
+      "step": 5181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4786106661085506e-06,
+      "loss": 2.4811,
+      "step": 5182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4766983215528875e-06,
+      "loss": 2.404,
+      "step": 5183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4747862225970074e-06,
+      "loss": 2.4135,
+      "step": 5184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.472874369549192e-06,
+      "loss": 2.4078,
+      "step": 5185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.47096276271769e-06,
+      "loss": 2.4307,
+      "step": 5186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.469051402410707e-06,
+      "loss": 2.5864,
+      "step": 5187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.467140288936407e-06,
+      "loss": 2.5506,
+      "step": 5188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4652294226029177e-06,
+      "loss": 2.4612,
+      "step": 5189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.463318803718323e-06,
+      "loss": 2.555,
+      "step": 5190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4614084325906718e-06,
+      "loss": 2.5685,
+      "step": 5191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.45949830952797e-06,
+      "loss": 2.5255,
+      "step": 5192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.457588434838184e-06,
+      "loss": 2.3345,
+      "step": 5193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4556788088292393e-06,
+      "loss": 2.5304,
+      "step": 5194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4537694318090215e-06,
+      "loss": 2.549,
+      "step": 5195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4518603040853783e-06,
+      "loss": 2.5154,
+      "step": 5196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.449951425966116e-06,
+      "loss": 2.4802,
+      "step": 5197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4480427977590004e-06,
+      "loss": 2.501,
+      "step": 5198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4461344197717562e-06,
+      "loss": 2.6123,
+      "step": 5199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.444226292312068e-06,
+      "loss": 2.5901,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.442318415687581e-06,
+      "loss": 2.5017,
+      "step": 5201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4404107902059003e-06,
+      "loss": 2.6913,
+      "step": 5202
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.438503416174589e-06,
+      "loss": 2.4979,
+      "step": 5203
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.43659629390117e-06,
+      "loss": 2.5792,
+      "step": 5204
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.434689423693126e-06,
+      "loss": 2.3556,
+      "step": 5205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4327828058578984e-06,
+      "loss": 2.5302,
+      "step": 5206
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.430876440702889e-06,
+      "loss": 2.4373,
+      "step": 5207
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4289703285354587e-06,
+      "loss": 2.3931,
+      "step": 5208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4270644696629286e-06,
+      "loss": 2.4269,
+      "step": 5209
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4251588643925757e-06,
+      "loss": 2.4641,
+      "step": 5210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.423253513031639e-06,
+      "loss": 2.6967,
+      "step": 5211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.421348415887315e-06,
+      "loss": 2.5224,
+      "step": 5212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4194435732667593e-06,
+      "loss": 2.4448,
+      "step": 5213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4175389854770885e-06,
+      "loss": 2.351,
+      "step": 5214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.415634652825376e-06,
+      "loss": 2.4423,
+      "step": 5215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4137305756186543e-06,
+      "loss": 2.3759,
+      "step": 5216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.411826754163915e-06,
+      "loss": 2.3932,
+      "step": 5217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4099231887681073e-06,
+      "loss": 2.6536,
+      "step": 5218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4080198797381435e-06,
+      "loss": 2.5059,
+      "step": 5219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4061168273808896e-06,
+      "loss": 2.8095,
+      "step": 5220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4042140320031726e-06,
+      "loss": 2.4908,
+      "step": 5221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4023114939117773e-06,
+      "loss": 2.5286,
+      "step": 5222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4004092134134477e-06,
+      "loss": 2.4125,
+      "step": 5223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3985071908148836e-06,
+      "loss": 2.5578,
+      "step": 5224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3966054264227476e-06,
+      "loss": 2.5223,
+      "step": 5225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3947039205436593e-06,
+      "loss": 2.523,
+      "step": 5226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3928026734841935e-06,
+      "loss": 2.3373,
+      "step": 5227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.390901685550887e-06,
+      "loss": 2.3591,
+      "step": 5228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3890009570502304e-06,
+      "loss": 2.3541,
+      "step": 5229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3871004882886805e-06,
+      "loss": 2.6044,
+      "step": 5230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.385200279572643e-06,
+      "loss": 2.4239,
+      "step": 5231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.383300331208489e-06,
+      "loss": 2.5854,
+      "step": 5232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3814006435025436e-06,
+      "loss": 2.5308,
+      "step": 5233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3795012167610898e-06,
+      "loss": 2.3812,
+      "step": 5234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3776020512903697e-06,
+      "loss": 2.5382,
+      "step": 5235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3757031473965827e-06,
+      "loss": 2.6031,
+      "step": 5236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3738045053858882e-06,
+      "loss": 2.4826,
+      "step": 5237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3719061255643996e-06,
+      "loss": 2.7035,
+      "step": 5238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3700080082381913e-06,
+      "loss": 2.4976,
+      "step": 5239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3681101537132914e-06,
+      "loss": 2.5154,
+      "step": 5240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.366212562295693e-06,
+      "loss": 2.5132,
+      "step": 5241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.364315234291339e-06,
+      "loss": 2.6111,
+      "step": 5242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3624181700061328e-06,
+      "loss": 2.5742,
+      "step": 5243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.360521369745937e-06,
+      "loss": 2.6058,
+      "step": 5244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.358624833816569e-06,
+      "loss": 2.4172,
+      "step": 5245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.356728562523805e-06,
+      "loss": 2.564,
+      "step": 5246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3548325561733775e-06,
+      "loss": 2.5143,
+      "step": 5247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3529368150709762e-06,
+      "loss": 2.5733,
+      "step": 5248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.351041339522252e-06,
+      "loss": 2.7168,
+      "step": 5249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3491461298328064e-06,
+      "loss": 2.656,
+      "step": 5250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3472511863082034e-06,
+      "loss": 2.5923,
+      "step": 5251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3453565092539586e-06,
+      "loss": 2.5038,
+      "step": 5252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.343462098975553e-06,
+      "loss": 2.4233,
+      "step": 5253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3415679557784163e-06,
+      "loss": 2.5002,
+      "step": 5254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.339674079967942e-06,
+      "loss": 2.5497,
+      "step": 5255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3377804718494733e-06,
+      "loss": 2.5611,
+      "step": 5256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.335887131728316e-06,
+      "loss": 2.5799,
+      "step": 5257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.33399405990973e-06,
+      "loss": 2.3542,
+      "step": 5258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.332101256698932e-06,
+      "loss": 2.6261,
+      "step": 5259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.330208722401097e-06,
+      "loss": 2.5567,
+      "step": 5260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3283164573213557e-06,
+      "loss": 2.5271,
+      "step": 5261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.326424461764795e-06,
+      "loss": 2.4679,
+      "step": 5262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3245327360364566e-06,
+      "loss": 2.4132,
+      "step": 5263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3226412804413444e-06,
+      "loss": 2.4701,
+      "step": 5264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.320750095284414e-06,
+      "loss": 2.5796,
+      "step": 5265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3188591808705773e-06,
+      "loss": 2.511,
+      "step": 5266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3169685375047044e-06,
+      "loss": 2.297,
+      "step": 5267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.315078165491622e-06,
+      "loss": 2.5519,
+      "step": 5268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3131880651361115e-06,
+      "loss": 2.4593,
+      "step": 5269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3112982367429105e-06,
+      "loss": 2.5366,
+      "step": 5270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3094086806167126e-06,
+      "loss": 2.4915,
+      "step": 5271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3075193970621706e-06,
+      "loss": 2.321,
+      "step": 5272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3056303863838902e-06,
+      "loss": 2.4296,
+      "step": 5273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3037416488864303e-06,
+      "loss": 2.4635,
+      "step": 5274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3018531848743164e-06,
+      "loss": 2.6559,
+      "step": 5275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.299964994652017e-06,
+      "loss": 2.3666,
+      "step": 5276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2980770785239656e-06,
+      "loss": 2.591,
+      "step": 5277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2961894367945467e-06,
+      "loss": 2.3876,
+      "step": 5278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.294302069768104e-06,
+      "loss": 2.5892,
+      "step": 5279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.292414977748932e-06,
+      "loss": 2.5619,
+      "step": 5280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2905281610412866e-06,
+      "loss": 2.5996,
+      "step": 5281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.288641619949374e-06,
+      "loss": 2.5966,
+      "step": 5282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.286755354777361e-06,
+      "loss": 2.345,
+      "step": 5283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2848693658293675e-06,
+      "loss": 2.5066,
+      "step": 5284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2829836534094642e-06,
+      "loss": 2.6319,
+      "step": 5285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.281098217821689e-06,
+      "loss": 2.5479,
+      "step": 5286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2792130593700233e-06,
+      "loss": 2.5742,
+      "step": 5287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2773281783584104e-06,
+      "loss": 2.4955,
+      "step": 5288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2754435750907454e-06,
+      "loss": 2.6172,
+      "step": 5289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2735592498708818e-06,
+      "loss": 2.4367,
+      "step": 5290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2716752030026265e-06,
+      "loss": 2.4898,
+      "step": 5291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.269791434789741e-06,
+      "loss": 2.6084,
+      "step": 5292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.267907945535943e-06,
+      "loss": 2.4735,
+      "step": 5293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2660247355449037e-06,
+      "loss": 2.6067,
+      "step": 5294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.264141805120253e-06,
+      "loss": 2.5848,
+      "step": 5295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.262259154565569e-06,
+      "loss": 2.498,
+      "step": 5296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.260376784184393e-06,
+      "loss": 2.5366,
+      "step": 5297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2584946942802153e-06,
+      "loss": 2.4881,
+      "step": 5298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2566128851564833e-06,
+      "loss": 2.5459,
+      "step": 5299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.254731357116597e-06,
+      "loss": 2.3718,
+      "step": 5300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2528501104639126e-06,
+      "loss": 2.5318,
+      "step": 5301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.250969145501742e-06,
+      "loss": 2.4096,
+      "step": 5302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2490884625333506e-06,
+      "loss": 2.4562,
+      "step": 5303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.247208061861957e-06,
+      "loss": 2.6356,
+      "step": 5304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.245327943790736e-06,
+      "loss": 2.4381,
+      "step": 5305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2434481086228163e-06,
+      "loss": 2.3019,
+      "step": 5306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.24156855666128e-06,
+      "loss": 2.3692,
+      "step": 5307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2396892882091678e-06,
+      "loss": 2.6403,
+      "step": 5308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2378103035694708e-06,
+      "loss": 2.4894,
+      "step": 5309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2359316030451327e-06,
+      "loss": 2.4567,
+      "step": 5310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.234053186939055e-06,
+      "loss": 2.554,
+      "step": 5311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.232175055554091e-06,
+      "loss": 2.5812,
+      "step": 5312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2302972091930517e-06,
+      "loss": 2.6583,
+      "step": 5313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2284196481586984e-06,
+      "loss": 2.4835,
+      "step": 5314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2265423727537457e-06,
+      "loss": 2.7434,
+      "step": 5315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2246653832808674e-06,
+      "loss": 2.5405,
+      "step": 5316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2227886800426844e-06,
+      "loss": 2.5325,
+      "step": 5317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2209122633417766e-06,
+      "loss": 2.376,
+      "step": 5318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2190361334806765e-06,
+      "loss": 2.6829,
+      "step": 5319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2171602907618706e-06,
+      "loss": 2.6358,
+      "step": 5320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2152847354877963e-06,
+      "loss": 2.4313,
+      "step": 5321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2134094679608485e-06,
+      "loss": 2.5283,
+      "step": 5322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.211534488483373e-06,
+      "loss": 2.5206,
+      "step": 5323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.209659797357669e-06,
+      "loss": 2.3794,
+      "step": 5324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.207785394885993e-06,
+      "loss": 2.4439,
+      "step": 5325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2059112813705505e-06,
+      "loss": 2.4697,
+      "step": 5326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2040374571135024e-06,
+      "loss": 2.5367,
+      "step": 5327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2021639224169615e-06,
+      "loss": 2.3017,
+      "step": 5328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2002906775829967e-06,
+      "loss": 2.5431,
+      "step": 5329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1984177229136287e-06,
+      "loss": 2.6122,
+      "step": 5330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1965450587108317e-06,
+      "loss": 2.6673,
+      "step": 5331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1946726852765325e-06,
+      "loss": 2.3926,
+      "step": 5332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1928006029126103e-06,
+      "loss": 2.6189,
+      "step": 5333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.190928811920899e-06,
+      "loss": 2.6169,
+      "step": 5334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.189057312603183e-06,
+      "loss": 2.5706,
+      "step": 5335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.187186105261205e-06,
+      "loss": 2.4661,
+      "step": 5336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.185315190196655e-06,
+      "loss": 2.3463,
+      "step": 5337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.183444567711178e-06,
+      "loss": 2.4356,
+      "step": 5338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1815742381063718e-06,
+      "loss": 2.6496,
+      "step": 5339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.179704201683786e-06,
+      "loss": 2.4783,
+      "step": 5340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1778344587449267e-06,
+      "loss": 2.4083,
+      "step": 5341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1759650095912474e-06,
+      "loss": 2.4495,
+      "step": 5342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1740958545241586e-06,
+      "loss": 2.5192,
+      "step": 5343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.172226993845021e-06,
+      "loss": 2.5789,
+      "step": 5344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.170358427855148e-06,
+      "loss": 2.5553,
+      "step": 5345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1684901568558058e-06,
+      "loss": 2.5553,
+      "step": 5346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1666221811482123e-06,
+      "loss": 2.5861,
+      "step": 5347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.16475450103354e-06,
+      "loss": 2.3913,
+      "step": 5348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.162887116812913e-06,
+      "loss": 2.6615,
+      "step": 5349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.161020028787405e-06,
+      "loss": 2.5928,
+      "step": 5350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1591532372580438e-06,
+      "loss": 2.5309,
+      "step": 5351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1572867425258107e-06,
+      "loss": 2.456,
+      "step": 5352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.155420544891638e-06,
+      "loss": 2.5193,
+      "step": 5353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1535546446564107e-06,
+      "loss": 2.5478,
+      "step": 5354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.151689042120965e-06,
+      "loss": 2.5612,
+      "step": 5355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.149823737586089e-06,
+      "loss": 2.475,
+      "step": 5356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.147958731352523e-06,
+      "loss": 2.5624,
+      "step": 5357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.146094023720958e-06,
+      "loss": 2.5248,
+      "step": 5358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1442296149920416e-06,
+      "loss": 2.5248,
+      "step": 5359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1423655054663665e-06,
+      "loss": 2.6186,
+      "step": 5360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1405016954444824e-06,
+      "loss": 2.2934,
+      "step": 5361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1386381852268878e-06,
+      "loss": 2.5569,
+      "step": 5362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1367749751140324e-06,
+      "loss": 2.4768,
+      "step": 5363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1349120654063224e-06,
+      "loss": 2.5852,
+      "step": 5364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.13304945640411e-06,
+      "loss": 2.486,
+      "step": 5365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1311871484077017e-06,
+      "loss": 2.6075,
+      "step": 5366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.129325141717354e-06,
+      "loss": 2.5423,
+      "step": 5367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1274634366332775e-06,
+      "loss": 2.6243,
+      "step": 5368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.12560203345563e-06,
+      "loss": 2.4344,
+      "step": 5369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1237409324845224e-06,
+      "loss": 2.5058,
+      "step": 5370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1218801340200207e-06,
+      "loss": 2.5332,
+      "step": 5371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1200196383621363e-06,
+      "loss": 2.6755,
+      "step": 5372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.118159445810835e-06,
+      "loss": 2.5484,
+      "step": 5373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1162995566660307e-06,
+      "loss": 2.4251,
+      "step": 5374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1144399712275952e-06,
+      "loss": 2.4665,
+      "step": 5375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1125806897953447e-06,
+      "loss": 2.3831,
+      "step": 5376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.110721712669047e-06,
+      "loss": 2.5155,
+      "step": 5377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1088630401484245e-06,
+      "loss": 2.5082,
+      "step": 5378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.107004672533148e-06,
+      "loss": 2.5048,
+      "step": 5379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.105146610122839e-06,
+      "loss": 2.495,
+      "step": 5380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1032888532170693e-06,
+      "loss": 2.4757,
+      "step": 5381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.101431402115364e-06,
+      "loss": 2.3197,
+      "step": 5382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0995742571171972e-06,
+      "loss": 2.5489,
+      "step": 5383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0977174185219937e-06,
+      "loss": 2.5335,
+      "step": 5384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.095860886629126e-06,
+      "loss": 2.4862,
+      "step": 5385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0940046617379247e-06,
+      "loss": 2.4774,
+      "step": 5386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.092148744147665e-06,
+      "loss": 2.5163,
+      "step": 5387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.090293134157572e-06,
+      "loss": 2.493,
+      "step": 5388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.088437832066825e-06,
+      "loss": 2.4894,
+      "step": 5389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0865828381745515e-06,
+      "loss": 2.3907,
+      "step": 5390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.084728152779829e-06,
+      "loss": 2.4862,
+      "step": 5391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0828737761816864e-06,
+      "loss": 2.5417,
+      "step": 5392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0810197086791e-06,
+      "loss": 2.5239,
+      "step": 5393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0791659505710015e-06,
+      "loss": 2.4243,
+      "step": 5394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0773125021562683e-06,
+      "loss": 2.577,
+      "step": 5395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0754593637337276e-06,
+      "loss": 2.4012,
+      "step": 5396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0736065356021623e-06,
+      "loss": 2.4704,
+      "step": 5397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.071754018060299e-06,
+      "loss": 2.478,
+      "step": 5398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0699018114068163e-06,
+      "loss": 2.5369,
+      "step": 5399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.068049915940342e-06,
+      "loss": 2.5155,
+      "step": 5400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0661983319594566e-06,
+      "loss": 2.5737,
+      "step": 5401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0643470597626885e-06,
+      "loss": 2.5264,
+      "step": 5402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.062496099648514e-06,
+      "loss": 2.5013,
+      "step": 5403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0606454519153608e-06,
+      "loss": 2.6456,
+      "step": 5404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.058795116861608e-06,
+      "loss": 2.5562,
+      "step": 5405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.056945094785581e-06,
+      "loss": 2.3531,
+      "step": 5406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.055095385985555e-06,
+      "loss": 2.5611,
+      "step": 5407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.05324599075976e-06,
+      "loss": 2.4521,
+      "step": 5408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0513969094063685e-06,
+      "loss": 2.5203,
+      "step": 5409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.049548142223506e-06,
+      "loss": 2.4905,
+      "step": 5410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0476996895092448e-06,
+      "loss": 2.3758,
+      "step": 5411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0458515515616117e-06,
+      "loss": 2.5921,
+      "step": 5412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.044003728678577e-06,
+      "loss": 2.3541,
+      "step": 5413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.042156221158064e-06,
+      "loss": 2.6618,
+      "step": 5414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0403090292979425e-06,
+      "loss": 2.4225,
+      "step": 5415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0384621533960324e-06,
+      "loss": 2.3838,
+      "step": 5416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.036615593750104e-06,
+      "loss": 2.6415,
+      "step": 5417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0347693506578737e-06,
+      "loss": 2.7798,
+      "step": 5418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0329234244170113e-06,
+      "loss": 2.579,
+      "step": 5419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0310778153251325e-06,
+      "loss": 2.5691,
+      "step": 5420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0292325236798008e-06,
+      "loss": 2.5401,
+      "step": 5421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.027387549778531e-06,
+      "loss": 2.6773,
+      "step": 5422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0255428939187836e-06,
+      "loss": 2.4552,
+      "step": 5423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0236985563979727e-06,
+      "loss": 2.5276,
+      "step": 5424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0218545375134574e-06,
+      "loss": 2.47,
+      "step": 5425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.020010837562546e-06,
+      "loss": 2.6771,
+      "step": 5426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0181674568424944e-06,
+      "loss": 2.7199,
+      "step": 5427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0163243956505093e-06,
+      "loss": 2.5584,
+      "step": 5428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0144816542837444e-06,
+      "loss": 2.3753,
+      "step": 5429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0126392330393033e-06,
+      "loss": 2.3524,
+      "step": 5430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0107971322142377e-06,
+      "loss": 2.5185,
+      "step": 5431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0089553521055455e-06,
+      "loss": 2.5123,
+      "step": 5432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.007113893010174e-06,
+      "loss": 2.3808,
+      "step": 5433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.005272755225019e-06,
+      "loss": 2.5056,
+      "step": 5434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0034319390469263e-06,
+      "loss": 2.6498,
+      "step": 5435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.001591444772687e-06,
+      "loss": 2.687,
+      "step": 5436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.999751272699041e-06,
+      "loss": 2.4794,
+      "step": 5437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9979114231226776e-06,
+      "loss": 2.5409,
+      "step": 5438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.996071896340231e-06,
+      "loss": 2.5666,
+      "step": 5439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9942326926482877e-06,
+      "loss": 2.4712,
+      "step": 5440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.992393812343378e-06,
+      "loss": 2.3693,
+      "step": 5441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9905552557219853e-06,
+      "loss": 2.6952,
+      "step": 5442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.988717023080535e-06,
+      "loss": 2.5596,
+      "step": 5443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.986879114715403e-06,
+      "loss": 2.5059,
+      "step": 5444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.985041530922913e-06,
+      "loss": 2.5392,
+      "step": 5445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9832042719993354e-06,
+      "loss": 2.5685,
+      "step": 5446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9813673382408905e-06,
+      "loss": 2.5261,
+      "step": 5447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9795307299437425e-06,
+      "loss": 2.4393,
+      "step": 5448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.977694447404007e-06,
+      "loss": 2.5145,
+      "step": 5449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9758584909177434e-06,
+      "loss": 2.5438,
+      "step": 5450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9740228607809607e-06,
+      "loss": 2.4859,
+      "step": 5451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.972187557289616e-06,
+      "loss": 2.6954,
+      "step": 5452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9703525807396117e-06,
+      "loss": 2.4924,
+      "step": 5453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9685179314268004e-06,
+      "loss": 2.6191,
+      "step": 5454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9666836096469786e-06,
+      "loss": 2.6432,
+      "step": 5455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.964849615695892e-06,
+      "loss": 2.626,
+      "step": 5456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9630159498692303e-06,
+      "loss": 2.6534,
+      "step": 5457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9611826124626364e-06,
+      "loss": 2.4933,
+      "step": 5458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.959349603771695e-06,
+      "loss": 2.562,
+      "step": 5459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.95751692409194e-06,
+      "loss": 2.5579,
+      "step": 5460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9556845737188512e-06,
+      "loss": 2.6642,
+      "step": 5461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9538525529478557e-06,
+      "loss": 2.6059,
+      "step": 5462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9520208620743275e-06,
+      "loss": 2.5386,
+      "step": 5463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.950189501393588e-06,
+      "loss": 2.4603,
+      "step": 5464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9483584712009063e-06,
+      "loss": 2.5256,
+      "step": 5465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9465277717914952e-06,
+      "loss": 2.4229,
+      "step": 5466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9446974034605165e-06,
+      "loss": 2.4973,
+      "step": 5467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9428673665030772e-06,
+      "loss": 2.6098,
+      "step": 5468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.941037661214231e-06,
+      "loss": 2.5266,
+      "step": 5469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9392082878889804e-06,
+      "loss": 2.5528,
+      "step": 5470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9373792468222716e-06,
+      "loss": 2.6064,
+      "step": 5471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.935550538309e-06,
+      "loss": 2.5225,
+      "step": 5472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9337221626440026e-06,
+      "loss": 2.6076,
+      "step": 5473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9318941201220664e-06,
+      "loss": 2.5912,
+      "step": 5474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9300664110379275e-06,
+      "loss": 2.6216,
+      "step": 5475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9282390356862606e-06,
+      "loss": 2.6873,
+      "step": 5476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9264119943616942e-06,
+      "loss": 2.4719,
+      "step": 5477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.924585287358799e-06,
+      "loss": 2.2681,
+      "step": 5478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.92275891497209e-06,
+      "loss": 2.4833,
+      "step": 5479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9209328774960327e-06,
+      "loss": 2.647,
+      "step": 5480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9191071752250376e-06,
+      "loss": 2.5384,
+      "step": 5481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.917281808453457e-06,
+      "loss": 2.5378,
+      "step": 5482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.915456777475593e-06,
+      "loss": 2.3832,
+      "step": 5483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9136320825856967e-06,
+      "loss": 2.5508,
+      "step": 5484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.911807724077953e-06,
+      "loss": 2.6266,
+      "step": 5485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.90998370224651e-06,
+      "loss": 2.4925,
+      "step": 5486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.908160017385446e-06,
+      "loss": 2.4648,
+      "step": 5487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9063366697887947e-06,
+      "loss": 2.5367,
+      "step": 5488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.904513659750528e-06,
+      "loss": 2.7161,
+      "step": 5489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9026909875645693e-06,
+      "loss": 2.3945,
+      "step": 5490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9008686535247875e-06,
+      "loss": 2.4406,
+      "step": 5491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.899046657924992e-06,
+      "loss": 2.6448,
+      "step": 5492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.897225001058942e-06,
+      "loss": 2.587,
+      "step": 5493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8954036832203393e-06,
+      "loss": 2.6713,
+      "step": 5494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8935827047028333e-06,
+      "loss": 2.5569,
+      "step": 5495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8917620658000166e-06,
+      "loss": 2.3651,
+      "step": 5496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8899417668054307e-06,
+      "loss": 2.5694,
+      "step": 5497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8881218080125605e-06,
+      "loss": 2.5571,
+      "step": 5498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8863021897148306e-06,
+      "loss": 2.5118,
+      "step": 5499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.884482912205621e-06,
+      "loss": 2.5589,
+      "step": 5500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.882663975778247e-06,
+      "loss": 2.6359,
+      "step": 5501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.880845380725974e-06,
+      "loss": 2.4155,
+      "step": 5502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8790271273420147e-06,
+      "loss": 2.4688,
+      "step": 5503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8772092159195193e-06,
+      "loss": 2.4699,
+      "step": 5504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.875391646751591e-06,
+      "loss": 2.5513,
+      "step": 5505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8735744201312688e-06,
+      "loss": 2.5953,
+      "step": 5506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.871757536351546e-06,
+      "loss": 2.5137,
+      "step": 5507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8699409957053535e-06,
+      "loss": 2.5473,
+      "step": 5508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.868124798485572e-06,
+      "loss": 2.5241,
+      "step": 5509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8663089449850246e-06,
+      "loss": 2.576,
+      "step": 5510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.864493435496476e-06,
+      "loss": 2.7372,
+      "step": 5511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8626782703126423e-06,
+      "loss": 2.5764,
+      "step": 5512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8608634497261743e-06,
+      "loss": 2.6119,
+      "step": 5513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8590489740296767e-06,
+      "loss": 2.4251,
+      "step": 5514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.857234843515696e-06,
+      "loss": 2.6492,
+      "step": 5515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8554210584767188e-06,
+      "loss": 2.506,
+      "step": 5516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.853607619205182e-06,
+      "loss": 2.3984,
+      "step": 5517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8517945259934586e-06,
+      "loss": 2.3663,
+      "step": 5518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.849981779133879e-06,
+      "loss": 2.5027,
+      "step": 5519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8481693789187036e-06,
+      "loss": 2.3971,
+      "step": 5520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8463573256401456e-06,
+      "loss": 2.5667,
+      "step": 5521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.844545619590361e-06,
+      "loss": 2.2879,
+      "step": 5522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8427342610614446e-06,
+      "loss": 2.6009,
+      "step": 5523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.840923250345442e-06,
+      "loss": 2.7398,
+      "step": 5524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8391125877343412e-06,
+      "loss": 2.4725,
+      "step": 5525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8373022735200694e-06,
+      "loss": 2.521,
+      "step": 5526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.835492307994504e-06,
+      "loss": 2.4121,
+      "step": 5527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8336826914494607e-06,
+      "loss": 2.5251,
+      "step": 5528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.831873424176702e-06,
+      "loss": 2.4974,
+      "step": 5529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8300645064679345e-06,
+      "loss": 2.4901,
+      "step": 5530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.828255938614807e-06,
+      "loss": 2.6349,
+      "step": 5531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8264477209089147e-06,
+      "loss": 2.3902,
+      "step": 5532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8246398536417897e-06,
+      "loss": 2.4349,
+      "step": 5533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.822832337104915e-06,
+      "loss": 2.5474,
+      "step": 5534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8210251715897112e-06,
+      "loss": 2.5008,
+      "step": 5535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.819218357387547e-06,
+      "loss": 2.4588,
+      "step": 5536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8174118947897338e-06,
+      "loss": 2.3905,
+      "step": 5537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.815605784087522e-06,
+      "loss": 2.421,
+      "step": 5538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.813800025572111e-06,
+      "loss": 2.6113,
+      "step": 5539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8119946195346375e-06,
+      "loss": 2.56,
+      "step": 5540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.810189566266187e-06,
+      "loss": 2.4991,
+      "step": 5541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8083848660577846e-06,
+      "loss": 2.5564,
+      "step": 5542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8065805192004005e-06,
+      "loss": 2.4544,
+      "step": 5543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.804776525984948e-06,
+      "loss": 2.4096,
+      "step": 5544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8029728867022795e-06,
+      "loss": 2.4401,
+      "step": 5545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8011696016431967e-06,
+      "loss": 2.6136,
+      "step": 5546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7993666710984356e-06,
+      "loss": 2.4375,
+      "step": 5547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7975640953586846e-06,
+      "loss": 2.6309,
+      "step": 5548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.795761874714571e-06,
+      "loss": 2.3641,
+      "step": 5549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.79396000945666e-06,
+      "loss": 2.5756,
+      "step": 5550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.792158499875468e-06,
+      "loss": 2.4682,
+      "step": 5551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.790357346261445e-06,
+      "loss": 2.5471,
+      "step": 5552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7885565489049948e-06,
+      "loss": 2.5293,
+      "step": 5553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7867561080964522e-06,
+      "loss": 2.5337,
+      "step": 5554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7849560241261013e-06,
+      "loss": 2.4718,
+      "step": 5555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.78315629728417e-06,
+      "loss": 2.5642,
+      "step": 5556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.781356927860821e-06,
+      "loss": 2.5596,
+      "step": 5557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7795579161461685e-06,
+      "loss": 2.6784,
+      "step": 5558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7777592624302594e-06,
+      "loss": 2.5047,
+      "step": 5559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7759609670030922e-06,
+      "loss": 2.7324,
+      "step": 5560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7741630301546043e-06,
+      "loss": 2.5808,
+      "step": 5561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.77236545217467e-06,
+      "loss": 2.6279,
+      "step": 5562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7705682333531137e-06,
+      "loss": 2.6312,
+      "step": 5563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7687713739796972e-06,
+      "loss": 2.4349,
+      "step": 5564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7669748743441284e-06,
+      "loss": 2.4243,
+      "step": 5565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7651787347360504e-06,
+      "loss": 2.5013,
+      "step": 5566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.763382955445053e-06,
+      "loss": 2.4225,
+      "step": 5567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7615875367606704e-06,
+      "loss": 2.4718,
+      "step": 5568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7597924789723716e-06,
+      "loss": 2.6053,
+      "step": 5569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7579977823695727e-06,
+      "loss": 2.6068,
+      "step": 5570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.756203447241632e-06,
+      "loss": 2.7228,
+      "step": 5571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7544094738778436e-06,
+      "loss": 2.5415,
+      "step": 5572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7526158625674514e-06,
+      "loss": 2.4305,
+      "step": 5573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7508226135996313e-06,
+      "loss": 2.441,
+      "step": 5574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7490297272635134e-06,
+      "loss": 2.4284,
+      "step": 5575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7472372038481575e-06,
+      "loss": 2.5711,
+      "step": 5576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7454450436425707e-06,
+      "loss": 2.6394,
+      "step": 5577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7436532469357017e-06,
+      "loss": 2.6004,
+      "step": 5578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7418618140164372e-06,
+      "loss": 2.7372,
+      "step": 5579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7400707451736103e-06,
+      "loss": 2.5114,
+      "step": 5580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7382800406959885e-06,
+      "loss": 2.6886,
+      "step": 5581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7364897008722867e-06,
+      "loss": 2.479,
+      "step": 5582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.734699725991161e-06,
+      "loss": 2.5019,
+      "step": 5583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.732910116341203e-06,
+      "loss": 2.4084,
+      "step": 5584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.73112087221095e-06,
+      "loss": 2.5474,
+      "step": 5585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7293319938888795e-06,
+      "loss": 2.4668,
+      "step": 5586
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.727543481663413e-06,
+      "loss": 2.5269,
+      "step": 5587
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.725755335822903e-06,
+      "loss": 2.5351,
+      "step": 5588
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.723967556655655e-06,
+      "loss": 2.6037,
+      "step": 5589
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7221801444499104e-06,
+      "loss": 2.687,
+      "step": 5590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.720393099493847e-06,
+      "loss": 2.5428,
+      "step": 5591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7186064220755925e-06,
+      "loss": 2.5088,
+      "step": 5592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7168201124832057e-06,
+      "loss": 2.4969,
+      "step": 5593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7150341710046924e-06,
+      "loss": 2.7266,
+      "step": 5594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7132485979280005e-06,
+      "loss": 2.5336,
+      "step": 5595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7114633935410083e-06,
+      "loss": 2.4457,
+      "step": 5596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.709678558131551e-06,
+      "loss": 2.4767,
+      "step": 5597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7078940919873886e-06,
+      "loss": 2.4201,
+      "step": 5598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7061099953962316e-06,
+      "loss": 2.5372,
+      "step": 5599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.704326268645724e-06,
+      "loss": 2.5404,
+      "step": 5600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.702542912023455e-06,
+      "loss": 2.4303,
+      "step": 5601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7007599258169546e-06,
+      "loss": 2.4435,
+      "step": 5602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.698977310313687e-06,
+      "loss": 2.505,
+      "step": 5603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6971950658010666e-06,
+      "loss": 2.4273,
+      "step": 5604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.695413192566436e-06,
+      "loss": 2.387,
+      "step": 5605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6936316908970874e-06,
+      "loss": 2.5283,
+      "step": 5606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6918505610802493e-06,
+      "loss": 2.4471,
+      "step": 5607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6900698034030904e-06,
+      "loss": 2.4907,
+      "step": 5608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.688289418152722e-06,
+      "loss": 2.4815,
+      "step": 5609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.68650940561619e-06,
+      "loss": 2.4569,
+      "step": 5610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.684729766080486e-06,
+      "loss": 2.4151,
+      "step": 5611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6829504998325352e-06,
+      "loss": 2.5173,
+      "step": 5612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6811716071592086e-06,
+      "loss": 2.3114,
+      "step": 5613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.679393088347316e-06,
+      "loss": 2.585,
+      "step": 5614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6776149436836023e-06,
+      "loss": 2.4138,
+      "step": 5615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6758371734547584e-06,
+      "loss": 2.4584,
+      "step": 5616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.674059777947408e-06,
+      "loss": 2.5041,
+      "step": 5617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.67228275744812e-06,
+      "loss": 2.4311,
+      "step": 5618
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.670506112243401e-06,
+      "loss": 2.5432,
+      "step": 5619
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6687298426196974e-06,
+      "loss": 2.6252,
+      "step": 5620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6669539488633954e-06,
+      "loss": 2.4485,
+      "step": 5621
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6651784312608177e-06,
+      "loss": 2.6797,
+      "step": 5622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6634032900982288e-06,
+      "loss": 2.3932,
+      "step": 5623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.661628525661835e-06,
+      "loss": 2.4724,
+      "step": 5624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6598541382377744e-06,
+      "loss": 2.3866,
+      "step": 5625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.658080128112134e-06,
+      "loss": 2.604,
+      "step": 5626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6563064955709305e-06,
+      "loss": 2.5153,
+      "step": 5627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6545332409001267e-06,
+      "loss": 2.5127,
+      "step": 5628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6527603643856203e-06,
+      "loss": 2.5399,
+      "step": 5629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6509878663132514e-06,
+      "loss": 2.355,
+      "step": 5630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.649215746968799e-06,
+      "loss": 2.4667,
+      "step": 5631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6474440066379746e-06,
+      "loss": 2.4941,
+      "step": 5632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6456726456064384e-06,
+      "loss": 2.531,
+      "step": 5633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.64390166415978e-06,
+      "loss": 2.4794,
+      "step": 5634
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.642131062583534e-06,
+      "loss": 2.6076,
+      "step": 5635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6403608411631744e-06,
+      "loss": 2.4948,
+      "step": 5636
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.638591000184107e-06,
+      "loss": 2.4739,
+      "step": 5637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6368215399316856e-06,
+      "loss": 2.5466,
+      "step": 5638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6350524606911928e-06,
+      "loss": 2.5255,
+      "step": 5639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.633283762747857e-06,
+      "loss": 2.6151,
+      "step": 5640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.631515446386844e-06,
+      "loss": 2.4614,
+      "step": 5641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.629747511893255e-06,
+      "loss": 2.4149,
+      "step": 5642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6279799595521354e-06,
+      "loss": 2.5482,
+      "step": 5643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.62621278964846e-06,
+      "loss": 2.5784,
+      "step": 5644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.624446002467152e-06,
+      "loss": 2.323,
+      "step": 5645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.622679598293063e-06,
+      "loss": 2.591,
+      "step": 5646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6209135774109907e-06,
+      "loss": 2.4178,
+      "step": 5647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.61914794010567e-06,
+      "loss": 2.3658,
+      "step": 5648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.617382686661768e-06,
+      "loss": 2.7183,
+      "step": 5649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6156178173638978e-06,
+      "loss": 2.4108,
+      "step": 5650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.613853332496603e-06,
+      "loss": 2.4953,
+      "step": 5651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.612089232344371e-06,
+      "loss": 2.6214,
+      "step": 5652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6103255171916246e-06,
+      "loss": 2.5185,
+      "step": 5653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6085621873227264e-06,
+      "loss": 2.4915,
+      "step": 5654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6067992430219768e-06,
+      "loss": 2.5804,
+      "step": 5655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6050366845736087e-06,
+      "loss": 2.4007,
+      "step": 5656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.603274512261801e-06,
+      "loss": 2.5762,
+      "step": 5657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.601512726370662e-06,
+      "loss": 2.4189,
+      "step": 5658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.599751327184245e-06,
+      "loss": 2.5183,
+      "step": 5659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5979903149865386e-06,
+      "loss": 2.4393,
+      "step": 5660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.596229690061466e-06,
+      "loss": 2.4669,
+      "step": 5661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.594469452692892e-06,
+      "loss": 2.5278,
+      "step": 5662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5927096031646128e-06,
+      "loss": 2.5713,
+      "step": 5663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5909501417603744e-06,
+      "loss": 2.5852,
+      "step": 5664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5891910687638453e-06,
+      "loss": 2.3829,
+      "step": 5665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5874323844586414e-06,
+      "loss": 2.5066,
+      "step": 5666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5856740891283145e-06,
+      "loss": 2.4272,
+      "step": 5667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5839161830563475e-06,
+      "loss": 2.4649,
+      "step": 5668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.58215866652617e-06,
+      "loss": 2.4383,
+      "step": 5669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.58040153982114e-06,
+      "loss": 2.4128,
+      "step": 5670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.578644803224558e-06,
+      "loss": 2.5301,
+      "step": 5671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5768884570196615e-06,
+      "loss": 2.3275,
+      "step": 5672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5751325014896216e-06,
+      "loss": 2.4905,
+      "step": 5673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5733769369175493e-06,
+      "loss": 2.4747,
+      "step": 5674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5716217635864915e-06,
+      "loss": 2.5568,
+      "step": 5675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.569866981779433e-06,
+      "loss": 2.3576,
+      "step": 5676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.568112591779297e-06,
+      "loss": 2.4917,
+      "step": 5677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.566358593868936e-06,
+      "loss": 2.525,
+      "step": 5678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5646049883311504e-06,
+      "loss": 2.4985,
+      "step": 5679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.562851775448666e-06,
+      "loss": 2.3664,
+      "step": 5680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5610989555041536e-06,
+      "loss": 2.4181,
+      "step": 5681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.55934652878022e-06,
+      "loss": 2.6926,
+      "step": 5682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5575944955594025e-06,
+      "loss": 2.6438,
+      "step": 5683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.555842856124182e-06,
+      "loss": 2.5326,
+      "step": 5684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5540916107569693e-06,
+      "loss": 2.4401,
+      "step": 5685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5523407597401194e-06,
+      "loss": 2.5357,
+      "step": 5686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5505903033559163e-06,
+      "loss": 2.5441,
+      "step": 5687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5488402418865854e-06,
+      "loss": 2.4151,
+      "step": 5688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.547090575614288e-06,
+      "loss": 2.3912,
+      "step": 5689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5453413048211175e-06,
+      "loss": 2.6546,
+      "step": 5690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.543592429789108e-06,
+      "loss": 2.4842,
+      "step": 5691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.541843950800226e-06,
+      "loss": 2.7177,
+      "step": 5692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.540095868136378e-06,
+      "loss": 2.2205,
+      "step": 5693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5383481820794064e-06,
+      "loss": 2.4336,
+      "step": 5694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5366008929110854e-06,
+      "loss": 2.6279,
+      "step": 5695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5348540009131283e-06,
+      "loss": 2.6147,
+      "step": 5696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5331075063671852e-06,
+      "loss": 2.5281,
+      "step": 5697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5313614095548423e-06,
+      "loss": 2.5244,
+      "step": 5698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5296157107576162e-06,
+      "loss": 2.5064,
+      "step": 5699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.527870410256966e-06,
+      "loss": 2.5034,
+      "step": 5700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5261255083342863e-06,
+      "loss": 2.4018,
+      "step": 5701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5243810052709006e-06,
+      "loss": 2.6909,
+      "step": 5702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5226369013480773e-06,
+      "loss": 2.5642,
+      "step": 5703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5208931968470107e-06,
+      "loss": 2.6771,
+      "step": 5704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.519149892048839e-06,
+      "loss": 2.3836,
+      "step": 5705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.517406987234634e-06,
+      "loss": 2.709,
+      "step": 5706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.515664482685396e-06,
+      "loss": 2.5093,
+      "step": 5707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.513922378682075e-06,
+      "loss": 2.5555,
+      "step": 5708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.512180675505541e-06,
+      "loss": 2.5321,
+      "step": 5709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5104393734366113e-06,
+      "loss": 2.7031,
+      "step": 5710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5086984727560288e-06,
+      "loss": 2.4588,
+      "step": 5711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5069579737444795e-06,
+      "loss": 2.5738,
+      "step": 5712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5052178766825817e-06,
+      "loss": 2.3612,
+      "step": 5713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5034781818508867e-06,
+      "loss": 2.5201,
+      "step": 5714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5017388895298865e-06,
+      "loss": 2.5114,
+      "step": 5715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 2.6406,
+      "step": 5716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4982615135415907e-06,
+      "loss": 2.2936,
+      "step": 5717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4965234304349483e-06,
+      "loss": 2.5381,
+      "step": 5718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.494785750960304e-06,
+      "loss": 2.6164,
+      "step": 5719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4930484753978224e-06,
+      "loss": 2.5402,
+      "step": 5720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.491311604027599e-06,
+      "loss": 2.4872,
+      "step": 5721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4895751371296695e-06,
+      "loss": 2.5674,
+      "step": 5722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4878390749839992e-06,
+      "loss": 2.4377,
+      "step": 5723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.486103417870493e-06,
+      "loss": 2.4873,
+      "step": 5724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.48436816606899e-06,
+      "loss": 2.4355,
+      "step": 5725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.482633319859259e-06,
+      "loss": 2.5612,
+      "step": 5726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4808988795210084e-06,
+      "loss": 2.4526,
+      "step": 5727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.479164845333881e-06,
+      "loss": 2.7242,
+      "step": 5728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.477431217577448e-06,
+      "loss": 2.4699,
+      "step": 5729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4756979965312262e-06,
+      "loss": 2.4682,
+      "step": 5730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4739651824746557e-06,
+      "loss": 2.6249,
+      "step": 5731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.472232775687119e-06,
+      "loss": 2.5425,
+      "step": 5732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4705007764479256e-06,
+      "loss": 2.7044,
+      "step": 5733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.468769185036326e-06,
+      "loss": 2.4582,
+      "step": 5734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.467038001731503e-06,
+      "loss": 2.5087,
+      "step": 5735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.46530722681257e-06,
+      "loss": 2.4683,
+      "step": 5736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.463576860558581e-06,
+      "loss": 2.4651,
+      "step": 5737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4618469032485164e-06,
+      "loss": 2.4172,
+      "step": 5738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4601173551612966e-06,
+      "loss": 2.5539,
+      "step": 5739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4583882165757766e-06,
+      "loss": 2.4726,
+      "step": 5740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.456659487770737e-06,
+      "loss": 2.3692,
+      "step": 5741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.454931169024906e-06,
+      "loss": 2.3792,
+      "step": 5742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.453203260616932e-06,
+      "loss": 2.5887,
+      "step": 5743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4514757628254075e-06,
+      "loss": 2.3372,
+      "step": 5744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.44974867592885e-06,
+      "loss": 2.2937,
+      "step": 5745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4480220002057188e-06,
+      "loss": 2.5421,
+      "step": 5746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.446295735934403e-06,
+      "loss": 2.6029,
+      "step": 5747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4445698833932236e-06,
+      "loss": 2.517,
+      "step": 5748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.442844442860441e-06,
+      "loss": 2.5806,
+      "step": 5749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.441119414614241e-06,
+      "loss": 2.5828,
+      "step": 5750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.439394798932749e-06,
+      "loss": 2.3056,
+      "step": 5751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4376705960940245e-06,
+      "loss": 2.4688,
+      "step": 5752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.435946806376057e-06,
+      "loss": 2.4399,
+      "step": 5753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.434223430056771e-06,
+      "loss": 2.5264,
+      "step": 5754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4325004674140235e-06,
+      "loss": 2.5655,
+      "step": 5755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4307779187256064e-06,
+      "loss": 2.5325,
+      "step": 5756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4290557842692416e-06,
+      "loss": 2.6505,
+      "step": 5757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4273340643225874e-06,
+      "loss": 2.4826,
+      "step": 5758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4256127591632363e-06,
+      "loss": 2.6069,
+      "step": 5759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4238918690687086e-06,
+      "loss": 2.5471,
+      "step": 5760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4221713943164643e-06,
+      "loss": 2.51,
+      "step": 5761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4204513351838897e-06,
+      "loss": 2.3914,
+      "step": 5762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.418731691948309e-06,
+      "loss": 2.3416,
+      "step": 5763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.417012464886978e-06,
+      "loss": 2.6103,
+      "step": 5764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.415293654277086e-06,
+      "loss": 2.5468,
+      "step": 5765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.413575260395755e-06,
+      "loss": 2.6844,
+      "step": 5766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.411857283520036e-06,
+      "loss": 2.6653,
+      "step": 5767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4101397239269202e-06,
+      "loss": 2.5732,
+      "step": 5768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.408422581893323e-06,
+      "loss": 2.5474,
+      "step": 5769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.406705857696099e-06,
+      "loss": 2.507,
+      "step": 5770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.404989551612035e-06,
+      "loss": 2.7097,
+      "step": 5771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4032736639178443e-06,
+      "loss": 2.7728,
+      "step": 5772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.401558194890182e-06,
+      "loss": 2.6011,
+      "step": 5773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3998431448056237e-06,
+      "loss": 2.2879,
+      "step": 5774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3981285139406927e-06,
+      "loss": 2.4575,
+      "step": 5775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3964143025718316e-06,
+      "loss": 2.4089,
+      "step": 5776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.394700510975421e-06,
+      "loss": 2.4577,
+      "step": 5777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.392987139427776e-06,
+      "loss": 2.5017,
+      "step": 5778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.391274188205136e-06,
+      "loss": 2.524,
+      "step": 5779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.389561657583681e-06,
+      "loss": 2.5933,
+      "step": 5780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3878495478395207e-06,
+      "loss": 2.4295,
+      "step": 5781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3861378592486932e-06,
+      "loss": 2.3969,
+      "step": 5782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3844265920871753e-06,
+      "loss": 2.6997,
+      "step": 5783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3827157466308674e-06,
+      "loss": 2.4229,
+      "step": 5784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.38100532315561e-06,
+      "loss": 2.5702,
+      "step": 5785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3792953219371713e-06,
+      "loss": 2.4664,
+      "step": 5786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.377585743251253e-06,
+      "loss": 2.6392,
+      "step": 5787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3758765873734897e-06,
+      "loss": 2.5587,
+      "step": 5788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.374167854579442e-06,
+      "loss": 2.554,
+      "step": 5789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3724595451446104e-06,
+      "loss": 2.6561,
+      "step": 5790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.37075165934442e-06,
+      "loss": 2.5094,
+      "step": 5791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3690441974542322e-06,
+      "loss": 2.5966,
+      "step": 5792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3673371597493412e-06,
+      "loss": 2.4599,
+      "step": 5793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.365630546504966e-06,
+      "loss": 2.5728,
+      "step": 5794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3639243579962646e-06,
+      "loss": 2.6712,
+      "step": 5795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3622185944983187e-06,
+      "loss": 2.7006,
+      "step": 5796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.360513256286153e-06,
+      "loss": 2.3518,
+      "step": 5797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.358808343634712e-06,
+      "loss": 2.5924,
+      "step": 5798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3571038568188783e-06,
+      "loss": 2.5282,
+      "step": 5799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.355399796113464e-06,
+      "loss": 2.4929,
+      "step": 5800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3536961617932103e-06,
+      "loss": 2.2665,
+      "step": 5801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.351992954132795e-06,
+      "loss": 2.5944,
+      "step": 5802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3502901734068206e-06,
+      "loss": 2.5106,
+      "step": 5803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3485878198898253e-06,
+      "loss": 2.6461,
+      "step": 5804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.346885893856279e-06,
+      "loss": 2.4756,
+      "step": 5805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3451843955805774e-06,
+      "loss": 2.6545,
+      "step": 5806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3434833253370527e-06,
+      "loss": 2.5651,
+      "step": 5807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3417826833999657e-06,
+      "loss": 2.5341,
+      "step": 5808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3400824700435104e-06,
+      "loss": 2.3091,
+      "step": 5809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3383826855418064e-06,
+      "loss": 2.5746,
+      "step": 5810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.336683330168909e-06,
+      "loss": 2.6376,
+      "step": 5811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3349844041988044e-06,
+      "loss": 2.5146,
+      "step": 5812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3332859079054054e-06,
+      "loss": 2.5075,
+      "step": 5813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3315878415625613e-06,
+      "loss": 2.5967,
+      "step": 5814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3298902054440454e-06,
+      "loss": 2.5576,
+      "step": 5815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3281929998235664e-06,
+      "loss": 2.4374,
+      "step": 5816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.326496224974765e-06,
+      "loss": 2.469,
+      "step": 5817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3247998811712036e-06,
+      "loss": 2.6438,
+      "step": 5818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.32310396868639e-06,
+      "loss": 2.5341,
+      "step": 5819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3214084877937464e-06,
+      "loss": 2.5484,
+      "step": 5820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3197134387666375e-06,
+      "loss": 2.4038,
+      "step": 5821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.31801882187835e-06,
+      "loss": 2.4635,
+      "step": 5822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3163246374021065e-06,
+      "loss": 2.595,
+      "step": 5823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3146308856110593e-06,
+      "loss": 2.5702,
+      "step": 5824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3129375667782865e-06,
+      "loss": 2.4857,
+      "step": 5825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.311244681176803e-06,
+      "loss": 2.5409,
+      "step": 5826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3095522290795463e-06,
+      "loss": 2.4583,
+      "step": 5827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.30786021075939e-06,
+      "loss": 2.3971,
+      "step": 5828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3061686264891365e-06,
+      "loss": 2.4815,
+      "step": 5829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3044774765415172e-06,
+      "loss": 2.4741,
+      "step": 5830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3027867611891953e-06,
+      "loss": 2.4398,
+      "step": 5831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.301096480704759e-06,
+      "loss": 2.4391,
+      "step": 5832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.299406635360732e-06,
+      "loss": 2.6364,
+      "step": 5833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.297717225429567e-06,
+      "loss": 2.6577,
+      "step": 5834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2960282511836416e-06,
+      "loss": 2.531,
+      "step": 5835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.294339712895271e-06,
+      "loss": 2.3623,
+      "step": 5836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.292651610836691e-06,
+      "loss": 2.4875,
+      "step": 5837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2909639452800754e-06,
+      "loss": 2.3647,
+      "step": 5838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2892767164975244e-06,
+      "loss": 2.5737,
+      "step": 5839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.287589924761065e-06,
+      "loss": 2.5266,
+      "step": 5840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.285903570342657e-06,
+      "loss": 2.5632,
+      "step": 5841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2842176535141897e-06,
+      "loss": 2.5344,
+      "step": 5842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2825321745474833e-06,
+      "loss": 2.3309,
+      "step": 5843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.28084713371428e-06,
+      "loss": 2.3182,
+      "step": 5844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.27916253128626e-06,
+      "loss": 2.4997,
+      "step": 5845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2774783675350303e-06,
+      "loss": 2.4612,
+      "step": 5846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2757946427321233e-06,
+      "loss": 2.5664,
+      "step": 5847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2741113571490066e-06,
+      "loss": 2.6133,
+      "step": 5848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2724285110570704e-06,
+      "loss": 2.4452,
+      "step": 5849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2707461047276395e-06,
+      "loss": 2.5697,
+      "step": 5850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.269064138431968e-06,
+      "loss": 2.465,
+      "step": 5851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2673826124412314e-06,
+      "loss": 2.5376,
+      "step": 5852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2657015270265463e-06,
+      "loss": 2.4654,
+      "step": 5853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.264020882458947e-06,
+      "loss": 2.4883,
+      "step": 5854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2623406790094057e-06,
+      "loss": 2.3018,
+      "step": 5855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.260660916948814e-06,
+      "loss": 2.5183,
+      "step": 5856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2589815965480007e-06,
+      "loss": 2.5283,
+      "step": 5857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.257302718077721e-06,
+      "loss": 2.5654,
+      "step": 5858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2556242818086555e-06,
+      "loss": 2.3642,
+      "step": 5859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.253946288011419e-06,
+      "loss": 2.4153,
+      "step": 5860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2522687369565498e-06,
+      "loss": 2.4917,
+      "step": 5861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2505916289145176e-06,
+      "loss": 2.2995,
+      "step": 5862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2489149641557196e-06,
+      "loss": 2.5851,
+      "step": 5863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.247238742950483e-06,
+      "loss": 2.5418,
+      "step": 5864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2455629655690655e-06,
+      "loss": 2.4183,
+      "step": 5865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2438876322816442e-06,
+      "loss": 2.53,
+      "step": 5866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.242212743358337e-06,
+      "loss": 2.3489,
+      "step": 5867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.240538299069178e-06,
+      "loss": 2.4744,
+      "step": 5868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.238864299684139e-06,
+      "loss": 2.3643,
+      "step": 5869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2371907454731168e-06,
+      "loss": 2.5134,
+      "step": 5870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.235517636705934e-06,
+      "loss": 2.4054,
+      "step": 5871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.233844973652346e-06,
+      "loss": 2.4568,
+      "step": 5872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2321727565820313e-06,
+      "loss": 2.6224,
+      "step": 5873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2305009857646003e-06,
+      "loss": 2.6461,
+      "step": 5874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2288296614695905e-06,
+      "loss": 2.4097,
+      "step": 5875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2271587839664673e-06,
+      "loss": 2.3467,
+      "step": 5876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.225488353524625e-06,
+      "loss": 2.5661,
+      "step": 5877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.223818370413382e-06,
+      "loss": 2.4322,
+      "step": 5878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2221488349019903e-06,
+      "loss": 2.5194,
+      "step": 5879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2204797472596233e-06,
+      "loss": 2.5257,
+      "step": 5880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.218811107755388e-06,
+      "loss": 2.5032,
+      "step": 5881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.217142916658318e-06,
+      "loss": 2.5394,
+      "step": 5882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2154751742373706e-06,
+      "loss": 2.5377,
+      "step": 5883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.213807880761434e-06,
+      "loss": 2.5006,
+      "step": 5884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2121410364993244e-06,
+      "loss": 2.5967,
+      "step": 5885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.210474641719785e-06,
+      "loss": 2.5582,
+      "step": 5886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2088086966914873e-06,
+      "loss": 2.5298,
+      "step": 5887
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2071432016830257e-06,
+      "loss": 2.4842,
+      "step": 5888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2054781569629306e-06,
+      "loss": 2.4738,
+      "step": 5889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2038135627996498e-06,
+      "loss": 2.4383,
+      "step": 5890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2021494194615655e-06,
+      "loss": 2.5851,
+      "step": 5891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2004857272169878e-06,
+      "loss": 2.616,
+      "step": 5892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.198822486334147e-06,
+      "loss": 2.2778,
+      "step": 5893
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.197159697081208e-06,
+      "loss": 2.5006,
+      "step": 5894
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.195497359726258e-06,
+      "loss": 2.58,
+      "step": 5895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1938354745373147e-06,
+      "loss": 2.6181,
+      "step": 5896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1921740417823206e-06,
+      "loss": 2.4936,
+      "step": 5897
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1905130617291474e-06,
+      "loss": 2.5788,
+      "step": 5898
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1888525346455936e-06,
+      "loss": 2.4094,
+      "step": 5899
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.18719246079938e-06,
+      "loss": 2.5961,
+      "step": 5900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1855328404581616e-06,
+      "loss": 2.4909,
+      "step": 5901
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1838736738895134e-06,
+      "loss": 2.5542,
+      "step": 5902
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1822149613609415e-06,
+      "loss": 2.4897,
+      "step": 5903
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1805567031398812e-06,
+      "loss": 2.5184,
+      "step": 5904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1788988994936855e-06,
+      "loss": 2.4122,
+      "step": 5905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1772415506896447e-06,
+      "loss": 2.5108,
+      "step": 5906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.175584656994965e-06,
+      "loss": 2.3596,
+      "step": 5907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.173928218676792e-06,
+      "loss": 2.417,
+      "step": 5908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.172272236002187e-06,
+      "loss": 2.3193,
+      "step": 5909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1706167092381416e-06,
+      "loss": 2.3345,
+      "step": 5910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.168961638651576e-06,
+      "loss": 2.3646,
+      "step": 5911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1673070245093324e-06,
+      "loss": 2.4597,
+      "step": 5912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.165652867078185e-06,
+      "loss": 2.5869,
+      "step": 5913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.163999166624828e-06,
+      "loss": 2.7108,
+      "step": 5914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1623459234158863e-06,
+      "loss": 2.4202,
+      "step": 5915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.160693137717912e-06,
+      "loss": 2.3703,
+      "step": 5916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.159040809797378e-06,
+      "loss": 2.358,
+      "step": 5917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.157388939920689e-06,
+      "loss": 2.5393,
+      "step": 5918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1557375283541726e-06,
+      "loss": 2.5896,
+      "step": 5919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.154086575364086e-06,
+      "loss": 2.6151,
+      "step": 5920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.152436081216607e-06,
+      "loss": 2.5488,
+      "step": 5921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1507860461778424e-06,
+      "loss": 2.4761,
+      "step": 5922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.149136470513829e-06,
+      "loss": 2.4714,
+      "step": 5923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1474873544905204e-06,
+      "loss": 2.5399,
+      "step": 5924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.145838698373806e-06,
+      "loss": 2.4887,
+      "step": 5925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1441905024294917e-06,
+      "loss": 2.502,
+      "step": 5926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.142542766923317e-06,
+      "loss": 2.4718,
+      "step": 5927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1408954921209435e-06,
+      "loss": 2.4556,
+      "step": 5928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1392486782879573e-06,
+      "loss": 2.5141,
+      "step": 5929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.137602325689873e-06,
+      "loss": 2.4717,
+      "step": 5930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1359564345921304e-06,
+      "loss": 2.5349,
+      "step": 5931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.134311005260093e-06,
+      "loss": 2.4174,
+      "step": 5932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1326660379590538e-06,
+      "loss": 2.5451,
+      "step": 5933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1310215329542243e-06,
+      "loss": 2.5673,
+      "step": 5934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1293774905107496e-06,
+      "loss": 2.439,
+      "step": 5935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.127733910893693e-06,
+      "loss": 2.4341,
+      "step": 5936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1260907943680482e-06,
+      "loss": 2.522,
+      "step": 5937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.124448141198734e-06,
+      "loss": 2.3849,
+      "step": 5938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1228059516505894e-06,
+      "loss": 2.3982,
+      "step": 5939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.121164225988387e-06,
+      "loss": 2.4897,
+      "step": 5940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1195229644768136e-06,
+      "loss": 2.4818,
+      "step": 5941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1178821673804943e-06,
+      "loss": 2.434,
+      "step": 5942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.116241834963968e-06,
+      "loss": 2.5565,
+      "step": 5943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1146019674917044e-06,
+      "loss": 2.6814,
+      "step": 5944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1129625652280987e-06,
+      "loss": 2.5577,
+      "step": 5945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1113236284374673e-06,
+      "loss": 2.5071,
+      "step": 5946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.109685157384055e-06,
+      "loss": 2.54,
+      "step": 5947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.108047152332028e-06,
+      "loss": 2.4774,
+      "step": 5948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.106409613545481e-06,
+      "loss": 2.4907,
+      "step": 5949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.104772541288434e-06,
+      "loss": 2.6219,
+      "step": 5950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.103135935824826e-06,
+      "loss": 2.6793,
+      "step": 5951
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.101499797418527e-06,
+      "loss": 2.5391,
+      "step": 5952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0998641263333287e-06,
+      "loss": 2.3052,
+      "step": 5953
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0982289228329495e-06,
+      "loss": 2.3836,
+      "step": 5954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.096594187181028e-06,
+      "loss": 2.5218,
+      "step": 5955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0949599196411326e-06,
+      "loss": 2.4734,
+      "step": 5956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.093326120476754e-06,
+      "loss": 2.4891,
+      "step": 5957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0916927899513055e-06,
+      "loss": 2.3335,
+      "step": 5958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0900599283281296e-06,
+      "loss": 2.3977,
+      "step": 5959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0884275358704863e-06,
+      "loss": 2.4655,
+      "step": 5960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.086795612841567e-06,
+      "loss": 2.465,
+      "step": 5961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.085164159504484e-06,
+      "loss": 2.6598,
+      "step": 5962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.083533176122271e-06,
+      "loss": 2.3651,
+      "step": 5963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.081902662957895e-06,
+      "loss": 2.4563,
+      "step": 5964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0802726202742375e-06,
+      "loss": 2.5568,
+      "step": 5965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.078643048334109e-06,
+      "loss": 2.5168,
+      "step": 5966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0770139474002413e-06,
+      "loss": 2.6272,
+      "step": 5967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0753853177352945e-06,
+      "loss": 2.5647,
+      "step": 5968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.07375715960185e-06,
+      "loss": 2.5741,
+      "step": 5969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0721294732624113e-06,
+      "loss": 2.4754,
+      "step": 5970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.070502258979411e-06,
+      "loss": 2.5973,
+      "step": 5971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0688755170152e-06,
+      "loss": 2.5362,
+      "step": 5972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0672492476320555e-06,
+      "loss": 2.7385,
+      "step": 5973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.06562345109218e-06,
+      "loss": 2.445,
+      "step": 5974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.063998127657698e-06,
+      "loss": 2.5785,
+      "step": 5975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.06237327759066e-06,
+      "loss": 2.4941,
+      "step": 5976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0607489011530346e-06,
+      "loss": 2.5725,
+      "step": 5977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0591249986067214e-06,
+      "loss": 2.5758,
+      "step": 5978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0575015702135363e-06,
+      "loss": 2.4613,
+      "step": 5979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0558786162352245e-06,
+      "loss": 2.731,
+      "step": 5980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.054256136933454e-06,
+      "loss": 2.5043,
+      "step": 5981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0526341325698107e-06,
+      "loss": 2.4693,
+      "step": 5982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.051012603405811e-06,
+      "loss": 2.4341,
+      "step": 5983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0493915497028933e-06,
+      "loss": 2.5523,
+      "step": 5984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0477709717224115e-06,
+      "loss": 2.5866,
+      "step": 5985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0461508697256573e-06,
+      "loss": 2.567,
+      "step": 5986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0445312439738313e-06,
+      "loss": 2.4185,
+      "step": 5987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.042912094728068e-06,
+      "loss": 2.589,
+      "step": 5988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0412934222494155e-06,
+      "loss": 2.5847,
+      "step": 5989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.039675226798854e-06,
+      "loss": 2.7594,
+      "step": 5990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0380575086372826e-06,
+      "loss": 2.4887,
+      "step": 5991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.036440268025522e-06,
+      "loss": 2.3989,
+      "step": 5992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0348235052243193e-06,
+      "loss": 2.7113,
+      "step": 5993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0332072204943405e-06,
+      "loss": 2.5177,
+      "step": 5994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0315914140961786e-06,
+      "loss": 2.5275,
+      "step": 5995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.029976086290347e-06,
+      "loss": 2.5828,
+      "step": 5996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.028361237337284e-06,
+      "loss": 2.6801,
+      "step": 5997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.02674686749735e-06,
+      "loss": 2.5148,
+      "step": 5998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.025132977030825e-06,
+      "loss": 2.625,
+      "step": 5999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0235195661979175e-06,
+      "loss": 2.5928,
+      "step": 6000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0219066352587513e-06,
+      "loss": 2.4292,
+      "step": 6001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0202941844733792e-06,
+      "loss": 2.5492,
+      "step": 6002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0186822141017764e-06,
+      "loss": 2.4903,
+      "step": 6003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.017070724403835e-06,
+      "loss": 2.631,
+      "step": 6004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.015459715639376e-06,
+      "loss": 2.6201,
+      "step": 6005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.013849188068138e-06,
+      "loss": 2.5436,
+      "step": 6006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0122391419497844e-06,
+      "loss": 2.4297,
+      "step": 6007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0106295775439018e-06,
+      "loss": 2.5309,
+      "step": 6008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0090204951099966e-06,
+      "loss": 2.2816,
+      "step": 6009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0074118949075024e-06,
+      "loss": 2.5731,
+      "step": 6010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0058037771957666e-06,
+      "loss": 2.4621,
+      "step": 6011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.004196142234068e-06,
+      "loss": 2.5809,
+      "step": 6012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0025889902815993e-06,
+      "loss": 2.3612,
+      "step": 6013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0009823215974815e-06,
+      "loss": 2.5492,
+      "step": 6014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9993761364407575e-06,
+      "loss": 2.5757,
+      "step": 6015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9977704350703865e-06,
+      "loss": 2.5392,
+      "step": 6016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.996165217745257e-06,
+      "loss": 2.6633,
+      "step": 6017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9945604847241697e-06,
+      "loss": 2.5084,
+      "step": 6018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9929562362658623e-06,
+      "loss": 2.585,
+      "step": 6019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9913524726289784e-06,
+      "loss": 2.6002,
+      "step": 6020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9897491940720936e-06,
+      "loss": 2.531,
+      "step": 6021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988146400853704e-06,
+      "loss": 2.3933,
+      "step": 6022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.986544093232222e-06,
+      "loss": 2.582,
+      "step": 6023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9849422714659875e-06,
+      "loss": 2.5339,
+      "step": 6024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9833409358132584e-06,
+      "loss": 2.5074,
+      "step": 6025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9817400865322166e-06,
+      "loss": 2.48,
+      "step": 6026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9801397238809665e-06,
+      "loss": 2.3314,
+      "step": 6027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9785398481175295e-06,
+      "loss": 2.5114,
+      "step": 6028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976940459499854e-06,
+      "loss": 2.3695,
+      "step": 6029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9753415582858026e-06,
+      "loss": 2.3009,
+      "step": 6030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9737431447331706e-06,
+      "loss": 2.6007,
+      "step": 6031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.972145219099662e-06,
+      "loss": 2.4165,
+      "step": 6032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9705477816429107e-06,
+      "loss": 2.5151,
+      "step": 6033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9689508326204716e-06,
+      "loss": 2.5491,
+      "step": 6034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9673543722898137e-06,
+      "loss": 2.4833,
+      "step": 6035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.965758400908334e-06,
+      "loss": 2.5457,
+      "step": 6036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9641629187333515e-06,
+      "loss": 2.6067,
+      "step": 6037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.962567926022099e-06,
+      "loss": 2.5089,
+      "step": 6038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.960973423031739e-06,
+      "loss": 2.4966,
+      "step": 6039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9593794100193474e-06,
+      "loss": 2.3741,
+      "step": 6040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.957785887241926e-06,
+      "loss": 2.3839,
+      "step": 6041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.956192854956397e-06,
+      "loss": 2.4711,
+      "step": 6042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.954600313419602e-06,
+      "loss": 2.4952,
+      "step": 6043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9530082628883058e-06,
+      "loss": 2.376,
+      "step": 6044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.95141670361919e-06,
+      "loss": 2.4977,
+      "step": 6045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.949825635868862e-06,
+      "loss": 2.564,
+      "step": 6046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9482350598938447e-06,
+      "loss": 2.4423,
+      "step": 6047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9466449759505856e-06,
+      "loss": 2.5614,
+      "step": 6048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9450553842954533e-06,
+      "loss": 2.6503,
+      "step": 6049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9434662851847325e-06,
+      "loss": 2.4795,
+      "step": 6050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9418776788746345e-06,
+      "loss": 2.5971,
+      "step": 6051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9402895656212834e-06,
+      "loss": 2.5378,
+      "step": 6052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.938701945680735e-06,
+      "loss": 2.5061,
+      "step": 6053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.937114819308953e-06,
+      "loss": 2.654,
+      "step": 6054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9355281867618306e-06,
+      "loss": 2.5217,
+      "step": 6055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9339420482951794e-06,
+      "loss": 2.4749,
+      "step": 6056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.932356404164727e-06,
+      "loss": 2.3838,
+      "step": 6057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930771254626128e-06,
+      "loss": 2.5949,
+      "step": 6058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.92918659993495e-06,
+      "loss": 2.5186,
+      "step": 6059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.927602440346687e-06,
+      "loss": 2.389,
+      "step": 6060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9260187761167527e-06,
+      "loss": 2.6009,
+      "step": 6061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9244356075004743e-06,
+      "loss": 2.4803,
+      "step": 6062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9228529347531067e-06,
+      "loss": 2.5759,
+      "step": 6063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.921270758129822e-06,
+      "loss": 2.4875,
+      "step": 6064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9196890778857137e-06,
+      "loss": 2.4662,
+      "step": 6065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.91810789427579e-06,
+      "loss": 2.5986,
+      "step": 6066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9165272075549856e-06,
+      "loss": 2.4909,
+      "step": 6067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.914947017978153e-06,
+      "loss": 2.6352,
+      "step": 6068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9133673258000617e-06,
+      "loss": 2.5849,
+      "step": 6069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.911788131275406e-06,
+      "loss": 2.5037,
+      "step": 6070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9102094346587935e-06,
+      "loss": 2.5859,
+      "step": 6071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9086312362047576e-06,
+      "loss": 2.6319,
+      "step": 6072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9070535361677506e-06,
+      "loss": 2.5948,
+      "step": 6073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9054763348021372e-06,
+      "loss": 2.6972,
+      "step": 6074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.903899632362215e-06,
+      "loss": 2.5334,
+      "step": 6075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9023234291021875e-06,
+      "loss": 2.3813,
+      "step": 6076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9007477252761875e-06,
+      "loss": 2.4932,
+      "step": 6077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.89917252113826e-06,
+      "loss": 2.5408,
+      "step": 6078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8975978169423754e-06,
+      "loss": 2.5393,
+      "step": 6079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.896023612942422e-06,
+      "loss": 2.6058,
+      "step": 6080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.894449909392203e-06,
+      "loss": 2.4534,
+      "step": 6081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8928767065454478e-06,
+      "loss": 2.6178,
+      "step": 6082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.891304004655799e-06,
+      "loss": 2.5869,
+      "step": 6083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.889731803976822e-06,
+      "loss": 2.6212,
+      "step": 6084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8881601047620013e-06,
+      "loss": 2.4643,
+      "step": 6085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8865889072647386e-06,
+      "loss": 2.5527,
+      "step": 6086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8850182117383587e-06,
+      "loss": 2.434,
+      "step": 6087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8834480184360987e-06,
+      "loss": 2.4418,
+      "step": 6088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8818783276111203e-06,
+      "loss": 2.6916,
+      "step": 6089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8803091395165046e-06,
+      "loss": 2.3544,
+      "step": 6090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8787404544052457e-06,
+      "loss": 2.5289,
+      "step": 6091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8771722725302644e-06,
+      "loss": 2.59,
+      "step": 6092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8756045941443923e-06,
+      "loss": 2.4017,
+      "step": 6093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8740374195003863e-06,
+      "loss": 2.4897,
+      "step": 6094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8724707488509213e-06,
+      "loss": 2.4571,
+      "step": 6095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8709045824485845e-06,
+      "loss": 2.6449,
+      "step": 6096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8693389205458928e-06,
+      "loss": 2.4249,
+      "step": 6097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8677737633952715e-06,
+      "loss": 2.4835,
+      "step": 6098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8662091112490717e-06,
+      "loss": 2.343,
+      "step": 6099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8646449643595565e-06,
+      "loss": 2.5574,
+      "step": 6100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.863081322978913e-06,
+      "loss": 2.5299,
+      "step": 6101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8615181873592465e-06,
+      "loss": 2.478,
+      "step": 6102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8599555577525752e-06,
+      "loss": 2.6043,
+      "step": 6103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8583934344108446e-06,
+      "loss": 2.3894,
+      "step": 6104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8568318175859084e-06,
+      "loss": 2.4664,
+      "step": 6105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8552707075295468e-06,
+      "loss": 2.364,
+      "step": 6106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8537101044934553e-06,
+      "loss": 2.5101,
+      "step": 6107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8521500087292466e-06,
+      "loss": 2.5445,
+      "step": 6108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8505904204884563e-06,
+      "loss": 2.6182,
+      "step": 6109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8490313400225296e-06,
+      "loss": 2.5294,
+      "step": 6110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8474727675828386e-06,
+      "loss": 2.532,
+      "step": 6111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8459147034206671e-06,
+      "loss": 2.5182,
+      "step": 6112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8443571477872207e-06,
+      "loss": 2.5269,
+      "step": 6113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8428001009336233e-06,
+      "loss": 2.5126,
+      "step": 6114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8412435631109126e-06,
+      "loss": 2.3535,
+      "step": 6115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8396875345700498e-06,
+      "loss": 2.6907,
+      "step": 6116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8381320155619088e-06,
+      "loss": 2.4706,
+      "step": 6117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8365770063372844e-06,
+      "loss": 2.4135,
+      "step": 6118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8350225071468887e-06,
+      "loss": 2.5322,
+      "step": 6119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.833468518241352e-06,
+      "loss": 2.4684,
+      "step": 6120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8319150398712227e-06,
+      "loss": 2.4512,
+      "step": 6121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.830362072286963e-06,
+      "loss": 2.5191,
+      "step": 6122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.828809615738959e-06,
+      "loss": 2.6897,
+      "step": 6123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8272576704775074e-06,
+      "loss": 2.5788,
+      "step": 6124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8257062367528284e-06,
+      "loss": 2.4151,
+      "step": 6125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8241553148150587e-06,
+      "loss": 2.5673,
+      "step": 6126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8226049049142481e-06,
+      "loss": 2.5084,
+      "step": 6127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8210550073003701e-06,
+      "loss": 2.4797,
+      "step": 6128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8195056222233098e-06,
+      "loss": 2.4562,
+      "step": 6129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.817956749932873e-06,
+      "loss": 2.459,
+      "step": 6130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8164083906787832e-06,
+      "loss": 2.7372,
+      "step": 6131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.81486054471068e-06,
+      "loss": 2.4502,
+      "step": 6132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.813313212278121e-06,
+      "loss": 2.6849,
+      "step": 6133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8117663936305785e-06,
+      "loss": 2.4881,
+      "step": 6134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8102200890174465e-06,
+      "loss": 2.6765,
+      "step": 6135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8086742986880308e-06,
+      "loss": 2.4205,
+      "step": 6136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8071290228915577e-06,
+      "loss": 2.3796,
+      "step": 6137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8055842618771724e-06,
+      "loss": 2.453,
+      "step": 6138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8040400158939302e-06,
+      "loss": 2.4089,
+      "step": 6139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8024962851908106e-06,
+      "loss": 2.5198,
+      "step": 6140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.800953070016706e-06,
+      "loss": 2.4407,
+      "step": 6141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.799410370620428e-06,
+      "loss": 2.6879,
+      "step": 6142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7978681872507043e-06,
+      "loss": 2.5456,
+      "step": 6143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7963265201561763e-06,
+      "loss": 2.5073,
+      "step": 6144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7947853695854083e-06,
+      "loss": 2.5424,
+      "step": 6145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7932447357868737e-06,
+      "loss": 2.6657,
+      "step": 6146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7917046190089693e-06,
+      "loss": 2.6421,
+      "step": 6147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.790165019500007e-06,
+      "loss": 2.4667,
+      "step": 6148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.788625937508211e-06,
+      "loss": 2.5344,
+      "step": 6149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.787087373281729e-06,
+      "loss": 2.4343,
+      "step": 6150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7855493270686181e-06,
+      "loss": 2.4672,
+      "step": 6151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7840117991168566e-06,
+      "loss": 2.455,
+      "step": 6152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7824747896743384e-06,
+      "loss": 2.4995,
+      "step": 6153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.780938298988873e-06,
+      "loss": 2.5538,
+      "step": 6154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.779402327308189e-06,
+      "loss": 2.5026,
+      "step": 6155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7778668748799244e-06,
+      "loss": 2.4977,
+      "step": 6156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7763319419516428e-06,
+      "loss": 2.5427,
+      "step": 6157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7747975287708147e-06,
+      "loss": 2.5118,
+      "step": 6158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7732636355848336e-06,
+      "loss": 2.7009,
+      "step": 6159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7717302626410087e-06,
+      "loss": 2.497,
+      "step": 6160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7701974101865598e-06,
+      "loss": 2.5041,
+      "step": 6161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.768665078468631e-06,
+      "loss": 2.6949,
+      "step": 6162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7671332677342711e-06,
+      "loss": 2.3949,
+      "step": 6163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7656019782304602e-06,
+      "loss": 2.6839,
+      "step": 6164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7640712102040802e-06,
+      "loss": 2.5522,
+      "step": 6165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7625409639019358e-06,
+      "loss": 2.4793,
+      "step": 6166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7610112395707496e-06,
+      "loss": 2.4227,
+      "step": 6167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.759482037457152e-06,
+      "loss": 2.5532,
+      "step": 6168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7579533578076985e-06,
+      "loss": 2.5204,
+      "step": 6169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7564252008688514e-06,
+      "loss": 2.5635,
+      "step": 6170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7548975668869966e-06,
+      "loss": 2.4454,
+      "step": 6171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7533704561084331e-06,
+      "loss": 2.5629,
+      "step": 6172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7518438687793715e-06,
+      "loss": 2.4768,
+      "step": 6173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.750317805145943e-06,
+      "loss": 2.546,
+      "step": 6174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7487922654541927e-06,
+      "loss": 2.5994,
+      "step": 6175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7472672499500837e-06,
+      "loss": 2.64,
+      "step": 6176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7457427588794874e-06,
+      "loss": 2.4444,
+      "step": 6177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7442187924881975e-06,
+      "loss": 2.4877,
+      "step": 6178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.742695351021923e-06,
+      "loss": 2.5095,
+      "step": 6179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7411724347262826e-06,
+      "loss": 2.4172,
+      "step": 6180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7396500438468172e-06,
+      "loss": 2.344,
+      "step": 6181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7381281786289767e-06,
+      "loss": 2.3966,
+      "step": 6182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7366068393181307e-06,
+      "loss": 2.4993,
+      "step": 6183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7350860261595647e-06,
+      "loss": 2.5481,
+      "step": 6184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7335657393984718e-06,
+      "loss": 2.5684,
+      "step": 6185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.732045979279972e-06,
+      "loss": 2.5723,
+      "step": 6186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7305267460490898e-06,
+      "loss": 2.4548,
+      "step": 6187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.729008039950772e-06,
+      "loss": 2.3987,
+      "step": 6188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7274898612298745e-06,
+      "loss": 2.6342,
+      "step": 6189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.725972210131172e-06,
+      "loss": 2.5457,
+      "step": 6190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.724455086899356e-06,
+      "loss": 2.6236,
+      "step": 6191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.722938491779026e-06,
+      "loss": 2.5,
+      "step": 6192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.721422425014702e-06,
+      "loss": 2.4481,
+      "step": 6193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.71990688685082e-06,
+      "loss": 2.6013,
+      "step": 6194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7183918775317237e-06,
+      "loss": 2.615,
+      "step": 6195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7168773973016779e-06,
+      "loss": 2.553,
+      "step": 6196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7153634464048602e-06,
+      "loss": 2.4336,
+      "step": 6197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7138500250853645e-06,
+      "loss": 2.542,
+      "step": 6198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7123371335871946e-06,
+      "loss": 2.6031,
+      "step": 6199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7108247721542725e-06,
+      "loss": 2.6543,
+      "step": 6200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7093129410304366e-06,
+      "loss": 2.5589,
+      "step": 6201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7078016404594338e-06,
+      "loss": 2.3019,
+      "step": 6202
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7062908706849324e-06,
+      "loss": 2.4995,
+      "step": 6203
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7047806319505079e-06,
+      "loss": 2.5481,
+      "step": 6204
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7032709244996559e-06,
+      "loss": 2.724,
+      "step": 6205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.701761748575786e-06,
+      "loss": 2.6534,
+      "step": 6206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7002531044222158e-06,
+      "loss": 2.5079,
+      "step": 6207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6987449922821887e-06,
+      "loss": 2.3036,
+      "step": 6208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6972374123988494e-06,
+      "loss": 2.5034,
+      "step": 6209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6957303650152674e-06,
+      "loss": 2.5834,
+      "step": 6210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6942238503744173e-06,
+      "loss": 2.5031,
+      "step": 6211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6927178687191953e-06,
+      "loss": 2.4492,
+      "step": 6212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6912124202924101e-06,
+      "loss": 2.3933,
+      "step": 6213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.689707505336779e-06,
+      "loss": 2.3531,
+      "step": 6214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6882031240949408e-06,
+      "loss": 2.5352,
+      "step": 6215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.686699276809442e-06,
+      "loss": 2.6495,
+      "step": 6216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6851959637227472e-06,
+      "loss": 2.5844,
+      "step": 6217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6836931850772331e-06,
+      "loss": 2.3818,
+      "step": 6218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.682190941115191e-06,
+      "loss": 2.4075,
+      "step": 6219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.680689232078827e-06,
+      "loss": 2.4176,
+      "step": 6220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.679188058210256e-06,
+      "loss": 2.4942,
+      "step": 6221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6776874197515141e-06,
+      "loss": 2.5682,
+      "step": 6222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6761873169445437e-06,
+      "loss": 2.5357,
+      "step": 6223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6746877500312054e-06,
+      "loss": 2.6225,
+      "step": 6224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6731887192532743e-06,
+      "loss": 2.477,
+      "step": 6225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6716902248524331e-06,
+      "loss": 2.5823,
+      "step": 6226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6701922670702863e-06,
+      "loss": 2.4594,
+      "step": 6227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6686948461483432e-06,
+      "loss": 2.3283,
+      "step": 6228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6671979623280331e-06,
+      "loss": 2.459,
+      "step": 6229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6657016158506966e-06,
+      "loss": 2.3691,
+      "step": 6230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6642058069575872e-06,
+      "loss": 2.5298,
+      "step": 6231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6627105358898737e-06,
+      "loss": 2.4684,
+      "step": 6232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6612158028886333e-06,
+      "loss": 2.4497,
+      "step": 6233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.659721608194863e-06,
+      "loss": 2.594,
+      "step": 6234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6582279520494665e-06,
+      "loss": 2.6507,
+      "step": 6235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.656734834693266e-06,
+      "loss": 2.2544,
+      "step": 6236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.655242256366995e-06,
+      "loss": 2.557,
+      "step": 6237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6537502173112974e-06,
+      "loss": 2.334,
+      "step": 6238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6522587177667343e-06,
+      "loss": 2.6791,
+      "step": 6239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6507677579737795e-06,
+      "loss": 2.5374,
+      "step": 6240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.649277338172814e-06,
+      "loss": 2.459,
+      "step": 6241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6477874586041415e-06,
+      "loss": 2.4842,
+      "step": 6242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6462981195079685e-06,
+      "loss": 2.3779,
+      "step": 6243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6448093211244232e-06,
+      "loss": 2.3774,
+      "step": 6244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6433210636935388e-06,
+      "loss": 2.7192,
+      "step": 6245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.641833347455266e-06,
+      "loss": 2.5152,
+      "step": 6246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6403461726494697e-06,
+      "loss": 2.6255,
+      "step": 6247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6388595395159207e-06,
+      "loss": 2.4622,
+      "step": 6248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.637373448294311e-06,
+      "loss": 2.5294,
+      "step": 6249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6358878992242377e-06,
+      "loss": 2.5627,
+      "step": 6250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6344028925452144e-06,
+      "loss": 2.4048,
+      "step": 6251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6329184284966675e-06,
+      "loss": 2.637,
+      "step": 6252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6314345073179356e-06,
+      "loss": 2.5374,
+      "step": 6253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6299511292482689e-06,
+      "loss": 2.5172,
+      "step": 6254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6284682945268292e-06,
+      "loss": 2.5292,
+      "step": 6255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6269860033926938e-06,
+      "loss": 2.4974,
+      "step": 6256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6255042560848478e-06,
+      "loss": 2.5602,
+      "step": 6257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6240230528421924e-06,
+      "loss": 2.5591,
+      "step": 6258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6225423939035418e-06,
+      "loss": 2.4685,
+      "step": 6259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.621062279507617e-06,
+      "loss": 2.5955,
+      "step": 6260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6195827098930588e-06,
+      "loss": 2.6571,
+      "step": 6261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6181036852984118e-06,
+      "loss": 2.4877,
+      "step": 6262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6166252059621396e-06,
+      "loss": 2.3767,
+      "step": 6263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6151472721226148e-06,
+      "loss": 2.376,
+      "step": 6264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6136698840181226e-06,
+      "loss": 2.4114,
+      "step": 6265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.612193041886862e-06,
+      "loss": 2.3644,
+      "step": 6266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6107167459669393e-06,
+      "loss": 2.4428,
+      "step": 6267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6092409964963779e-06,
+      "loss": 2.3334,
+      "step": 6268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6077657937131086e-06,
+      "loss": 2.5282,
+      "step": 6269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6062911378549768e-06,
+      "loss": 2.5719,
+      "step": 6270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.604817029159742e-06,
+      "loss": 2.5855,
+      "step": 6271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6033434678650684e-06,
+      "loss": 2.579,
+      "step": 6272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.60187045420854e-06,
+      "loss": 2.4394,
+      "step": 6273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.600397988427644e-06,
+      "loss": 2.3723,
+      "step": 6274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5989260707597898e-06,
+      "loss": 2.556,
+      "step": 6275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.597454701442288e-06,
+      "loss": 2.556,
+      "step": 6276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5959838807123672e-06,
+      "loss": 2.5655,
+      "step": 6277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5945136088071677e-06,
+      "loss": 2.5421,
+      "step": 6278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.593043885963736e-06,
+      "loss": 2.4283,
+      "step": 6279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5915747124190356e-06,
+      "loss": 2.4524,
+      "step": 6280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5901060884099374e-06,
+      "loss": 2.6047,
+      "step": 6281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5886380141732267e-06,
+      "loss": 2.4911,
+      "step": 6282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.587170489945601e-06,
+      "loss": 2.5257,
+      "step": 6283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5857035159636625e-06,
+      "loss": 2.5221,
+      "step": 6284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5842370924639327e-06,
+      "loss": 2.4198,
+      "step": 6285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5827712196828405e-06,
+      "loss": 2.3544,
+      "step": 6286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5813058978567276e-06,
+      "loss": 2.535,
+      "step": 6287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5798411272218427e-06,
+      "loss": 2.5314,
+      "step": 6288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5783769080143514e-06,
+      "loss": 2.5093,
+      "step": 6289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5769132404703274e-06,
+      "loss": 2.4158,
+      "step": 6290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5754501248257537e-06,
+      "loss": 2.5725,
+      "step": 6291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5739875613165283e-06,
+      "loss": 2.3843,
+      "step": 6292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5725255501784592e-06,
+      "loss": 2.4339,
+      "step": 6293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5710640916472614e-06,
+      "loss": 2.2395,
+      "step": 6294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5696031859585664e-06,
+      "loss": 2.4804,
+      "step": 6295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5681428333479104e-06,
+      "loss": 2.517,
+      "step": 6296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5666830340507494e-06,
+      "loss": 2.5937,
+      "step": 6297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5652237883024407e-06,
+      "loss": 2.5128,
+      "step": 6298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5637650963382573e-06,
+      "loss": 2.5078,
+      "step": 6299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5623069583933836e-06,
+      "loss": 2.5047,
+      "step": 6300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5608493747029108e-06,
+      "loss": 2.6039,
+      "step": 6301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5593923455018462e-06,
+      "loss": 2.557,
+      "step": 6302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5579358710251003e-06,
+      "loss": 2.5642,
+      "step": 6303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5564799515075007e-06,
+      "loss": 2.6319,
+      "step": 6304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5550245871837848e-06,
+      "loss": 2.5084,
+      "step": 6305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.553569778288596e-06,
+      "loss": 2.5955,
+      "step": 6306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5521155250564923e-06,
+      "loss": 2.6302,
+      "step": 6307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.550661827721941e-06,
+      "loss": 2.6135,
+      "step": 6308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5492086865193217e-06,
+      "loss": 2.5802,
+      "step": 6309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5477561016829185e-06,
+      "loss": 2.5337,
+      "step": 6310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5463040734469314e-06,
+      "loss": 2.4724,
+      "step": 6311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5448526020454702e-06,
+      "loss": 2.6671,
+      "step": 6312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5434016877125508e-06,
+      "loss": 2.578,
+      "step": 6313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5419513306821055e-06,
+      "loss": 2.5598,
+      "step": 6314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5405015311879696e-06,
+      "loss": 2.5154,
+      "step": 6315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5390522894638937e-06,
+      "loss": 2.5111,
+      "step": 6316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5376036057435394e-06,
+      "loss": 2.5135,
+      "step": 6317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.53615548026047e-06,
+      "loss": 2.596,
+      "step": 6318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5347079132481723e-06,
+      "loss": 2.4466,
+      "step": 6319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5332609049400298e-06,
+      "loss": 2.6255,
+      "step": 6320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5318144555693454e-06,
+      "loss": 2.576,
+      "step": 6321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5303685653693246e-06,
+      "loss": 2.5353,
+      "step": 6322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5289232345730882e-06,
+      "loss": 2.4884,
+      "step": 6323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5274784634136658e-06,
+      "loss": 2.3854,
+      "step": 6324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.526034252123993e-06,
+      "loss": 2.5539,
+      "step": 6325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5245906009369211e-06,
+      "loss": 2.3929,
+      "step": 6326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5231475100852044e-06,
+      "loss": 2.5888,
+      "step": 6327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5217049798015127e-06,
+      "loss": 2.3624,
+      "step": 6328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.520263010318424e-06,
+      "loss": 2.5663,
+      "step": 6329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.518821601868421e-06,
+      "loss": 2.5721,
+      "step": 6330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.517380754683906e-06,
+      "loss": 2.6077,
+      "step": 6331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5159404689971797e-06,
+      "loss": 2.4003,
+      "step": 6332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5145007450404608e-06,
+      "loss": 2.474,
+      "step": 6333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5130615830458705e-06,
+      "loss": 2.4282,
+      "step": 6334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5116229832454448e-06,
+      "loss": 2.4419,
+      "step": 6335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5101849458711288e-06,
+      "loss": 2.4102,
+      "step": 6336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5087474711547711e-06,
+      "loss": 2.5742,
+      "step": 6337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5073105593281385e-06,
+      "loss": 2.5421,
+      "step": 6338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.505874210622898e-06,
+      "loss": 2.336,
+      "step": 6339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5044384252706312e-06,
+      "loss": 2.5405,
+      "step": 6340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5030032035028292e-06,
+      "loss": 2.4859,
+      "step": 6341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5015685455508894e-06,
+      "loss": 2.3305,
+      "step": 6342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.500134451646122e-06,
+      "loss": 2.5273,
+      "step": 6343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4987009220197406e-06,
+      "loss": 2.4403,
+      "step": 6344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.497267956902873e-06,
+      "loss": 2.5051,
+      "step": 6345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4958355565265559e-06,
+      "loss": 2.5182,
+      "step": 6346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4944037211217289e-06,
+      "loss": 2.531,
+      "step": 6347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.492972450919249e-06,
+      "loss": 2.5816,
+      "step": 6348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4915417461498743e-06,
+      "loss": 2.3879,
+      "step": 6349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4901116070442772e-06,
+      "loss": 2.4461,
+      "step": 6350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4886820338330382e-06,
+      "loss": 2.459,
+      "step": 6351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4872530267466411e-06,
+      "loss": 2.5567,
+      "step": 6352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4858245860154885e-06,
+      "loss": 2.4055,
+      "step": 6353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4843967118698816e-06,
+      "loss": 2.5041,
+      "step": 6354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4829694045400373e-06,
+      "loss": 2.5423,
+      "step": 6355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4815426642560753e-06,
+      "loss": 2.3766,
+      "step": 6356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4801164912480286e-06,
+      "loss": 2.53,
+      "step": 6357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4786908857458388e-06,
+      "loss": 2.4645,
+      "step": 6358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4772658479793506e-06,
+      "loss": 2.2306,
+      "step": 6359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4758413781783249e-06,
+      "loss": 2.5422,
+      "step": 6360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4744174765724228e-06,
+      "loss": 2.6254,
+      "step": 6361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.47299414339122e-06,
+      "loss": 2.488,
+      "step": 6362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4715713788641982e-06,
+      "loss": 2.5071,
+      "step": 6363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4701491832207481e-06,
+      "loss": 2.297,
+      "step": 6364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4687275566901705e-06,
+      "loss": 2.5722,
+      "step": 6365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4673064995016672e-06,
+      "loss": 2.6829,
+      "step": 6366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4658860118843583e-06,
+      "loss": 2.6527,
+      "step": 6367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 2.6203,
+      "step": 6368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4630467462793142e-06,
+      "loss": 2.4755,
+      "step": 6369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4616279687493528e-06,
+      "loss": 2.599,
+      "step": 6370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4602097617061235e-06,
+      "loss": 2.4989,
+      "step": 6371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.458792125378285e-06,
+      "loss": 2.5067,
+      "step": 6372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4573750599943975e-06,
+      "loss": 2.5169,
+      "step": 6373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.455958565782934e-06,
+      "loss": 2.2895,
+      "step": 6374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4545426429722737e-06,
+      "loss": 2.4549,
+      "step": 6375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.453127291790704e-06,
+      "loss": 2.2885,
+      "step": 6376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4517125124664222e-06,
+      "loss": 2.4892,
+      "step": 6377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.450298305227527e-06,
+      "loss": 2.6348,
+      "step": 6378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4488846703020326e-06,
+      "loss": 2.5849,
+      "step": 6379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4474716079178541e-06,
+      "loss": 2.4127,
+      "step": 6380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4460591183028195e-06,
+      "loss": 2.3603,
+      "step": 6381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.444647201684664e-06,
+      "loss": 2.4361,
+      "step": 6382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.443235858291025e-06,
+      "loss": 2.5296,
+      "step": 6383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4418250883494556e-06,
+      "loss": 2.3567,
+      "step": 6384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4404148920874073e-06,
+      "loss": 2.4745,
+      "step": 6385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4390052697322499e-06,
+      "loss": 2.4612,
+      "step": 6386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4375962215112504e-06,
+      "loss": 2.5648,
+      "step": 6387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.436187747651589e-06,
+      "loss": 2.5367,
+      "step": 6388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4347798483803538e-06,
+      "loss": 2.4555,
+      "step": 6389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4333725239245355e-06,
+      "loss": 2.5323,
+      "step": 6390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4319657745110372e-06,
+      "loss": 2.4196,
+      "step": 6391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.430559600366665e-06,
+      "loss": 2.5982,
+      "step": 6392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.429154001718136e-06,
+      "loss": 2.5836,
+      "step": 6393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4277489787920733e-06,
+      "loss": 2.5185,
+      "step": 6394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4263445318150048e-06,
+      "loss": 2.567,
+      "step": 6395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4249406610133686e-06,
+      "loss": 2.3482,
+      "step": 6396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4235373666135087e-06,
+      "loss": 2.2306,
+      "step": 6397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4221346488416765e-06,
+      "loss": 2.5274,
+      "step": 6398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.420732507924032e-06,
+      "loss": 2.3507,
+      "step": 6399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4193309440866366e-06,
+      "loss": 2.5357,
+      "step": 6400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4179299575554661e-06,
+      "loss": 2.6459,
+      "step": 6401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4165295485563962e-06,
+      "loss": 2.5064,
+      "step": 6402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4151297173152152e-06,
+      "loss": 2.5294,
+      "step": 6403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4137304640576161e-06,
+      "loss": 2.5064,
+      "step": 6404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4123317890091965e-06,
+      "loss": 2.6676,
+      "step": 6405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4109336923954652e-06,
+      "loss": 2.5421,
+      "step": 6406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4095361744418312e-06,
+      "loss": 2.5511,
+      "step": 6407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4081392353736206e-06,
+      "loss": 2.5696,
+      "step": 6408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.406742875416055e-06,
+      "loss": 2.635,
+      "step": 6409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4053470947942694e-06,
+      "loss": 2.6467,
+      "step": 6410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4039518937333047e-06,
+      "loss": 2.5732,
+      "step": 6411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4025572724581037e-06,
+      "loss": 2.5326,
+      "step": 6412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4011632311935237e-06,
+      "loss": 2.4711,
+      "step": 6413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.39976977016432e-06,
+      "loss": 2.6506,
+      "step": 6414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3983768895951594e-06,
+      "loss": 2.3152,
+      "step": 6415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3969845897106166e-06,
+      "loss": 2.4289,
+      "step": 6416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3955928707351668e-06,
+      "loss": 2.4685,
+      "step": 6417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3942017328931974e-06,
+      "loss": 2.5651,
+      "step": 6418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3928111764089957e-06,
+      "loss": 2.4222,
+      "step": 6419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3914212015067653e-06,
+      "loss": 2.378,
+      "step": 6420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3900318084106045e-06,
+      "loss": 2.4775,
+      "step": 6421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3886429973445253e-06,
+      "loss": 2.4255,
+      "step": 6422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3872547685324456e-06,
+      "loss": 2.546,
+      "step": 6423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.385867122198184e-06,
+      "loss": 2.4063,
+      "step": 6424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3844800585654721e-06,
+      "loss": 2.552,
+      "step": 6425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3830935778579407e-06,
+      "loss": 2.5308,
+      "step": 6426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.381707680299132e-06,
+      "loss": 2.4289,
+      "step": 6427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3803223661124938e-06,
+      "loss": 2.7028,
+      "step": 6428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3789376355213758e-06,
+      "loss": 2.631,
+      "step": 6429
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3775534887490365e-06,
+      "loss": 2.4656,
+      "step": 6430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3761699260186407e-06,
+      "loss": 2.434,
+      "step": 6431
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3747869475532598e-06,
+      "loss": 2.5879,
+      "step": 6432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3734045535758667e-06,
+      "loss": 2.5157,
+      "step": 6433
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3720227443093437e-06,
+      "loss": 2.4777,
+      "step": 6434
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.37064151997648e-06,
+      "loss": 2.564,
+      "step": 6435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3692608807999652e-06,
+      "loss": 2.5147,
+      "step": 6436
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3678808270024008e-06,
+      "loss": 2.5625,
+      "step": 6437
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3665013588062875e-06,
+      "loss": 2.5354,
+      "step": 6438
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3651224764340371e-06,
+      "loss": 2.451,
+      "step": 6439
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3637441801079666e-06,
+      "loss": 2.5723,
+      "step": 6440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3623664700502915e-06,
+      "loss": 2.5165,
+      "step": 6441
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3609893464831447e-06,
+      "loss": 2.321,
+      "step": 6442
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3596128096285527e-06,
+      "loss": 2.5065,
+      "step": 6443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3582368597084566e-06,
+      "loss": 2.5337,
+      "step": 6444
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3568614969446947e-06,
+      "loss": 2.3071,
+      "step": 6445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3554867215590174e-06,
+      "loss": 2.4848,
+      "step": 6446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3541125337730786e-06,
+      "loss": 2.6199,
+      "step": 6447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.352738933808434e-06,
+      "loss": 2.6431,
+      "step": 6448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3513659218865483e-06,
+      "loss": 2.4526,
+      "step": 6449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3499934982287922e-06,
+      "loss": 2.6765,
+      "step": 6450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.348621663056436e-06,
+      "loss": 2.6479,
+      "step": 6451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3472504165906614e-06,
+      "loss": 2.4188,
+      "step": 6452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3458797590525513e-06,
+      "loss": 2.5463,
+      "step": 6453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3445096906630966e-06,
+      "loss": 2.4685,
+      "step": 6454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3431402116431896e-06,
+      "loss": 2.4783,
+      "step": 6455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3417713222136297e-06,
+      "loss": 2.6402,
+      "step": 6456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.340403022595123e-06,
+      "loss": 2.5257,
+      "step": 6457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3390353130082756e-06,
+      "loss": 2.4442,
+      "step": 6458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.337668193673604e-06,
+      "loss": 2.5003,
+      "step": 6459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3363016648115246e-06,
+      "loss": 2.5911,
+      "step": 6460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3349357266423624e-06,
+      "loss": 2.4491,
+      "step": 6461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3335703793863464e-06,
+      "loss": 2.5828,
+      "step": 6462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.332205623263605e-06,
+      "loss": 2.4079,
+      "step": 6463
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.330841458494183e-06,
+      "loss": 2.6267,
+      "step": 6464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3294778852980178e-06,
+      "loss": 2.4384,
+      "step": 6465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3281149038949591e-06,
+      "loss": 2.3579,
+      "step": 6466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3267525145047555e-06,
+      "loss": 2.5781,
+      "step": 6467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.325390717347065e-06,
+      "loss": 2.3955,
+      "step": 6468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3240295126414492e-06,
+      "loss": 2.4448,
+      "step": 6469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3226689006073712e-06,
+      "loss": 2.5643,
+      "step": 6470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3213088814642023e-06,
+      "loss": 2.6067,
+      "step": 6471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3199494554312143e-06,
+      "loss": 2.6089,
+      "step": 6472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3185906227275863e-06,
+      "loss": 2.3964,
+      "step": 6473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3172323835724021e-06,
+      "loss": 2.5362,
+      "step": 6474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3158747381846481e-06,
+      "loss": 2.4736,
+      "step": 6475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3145176867832165e-06,
+      "loss": 2.3416,
+      "step": 6476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3131612295869007e-06,
+      "loss": 2.3775,
+      "step": 6477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3118053668144026e-06,
+      "loss": 2.5352,
+      "step": 6478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3104500986843239e-06,
+      "loss": 2.405,
+      "step": 6479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3090954254151733e-06,
+      "loss": 2.527,
+      "step": 6480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3077413472253647e-06,
+      "loss": 2.2763,
+      "step": 6481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3063878643332112e-06,
+      "loss": 2.6017,
+      "step": 6482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.305034976956936e-06,
+      "loss": 2.4723,
+      "step": 6483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3036826853146601e-06,
+      "loss": 2.5914,
+      "step": 6484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3023309896244129e-06,
+      "loss": 2.4898,
+      "step": 6485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3009798901041264e-06,
+      "loss": 2.512,
+      "step": 6486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2996293869716369e-06,
+      "loss": 2.6156,
+      "step": 6487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2982794804446858e-06,
+      "loss": 2.6218,
+      "step": 6488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2969301707409126e-06,
+      "loss": 2.5452,
+      "step": 6489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2955814580778686e-06,
+      "loss": 2.3198,
+      "step": 6490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2942333426730014e-06,
+      "loss": 2.5006,
+      "step": 6491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2928858247436672e-06,
+      "loss": 2.5931,
+      "step": 6492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2915389045071258e-06,
+      "loss": 2.5282,
+      "step": 6493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2901925821805372e-06,
+      "loss": 2.6882,
+      "step": 6494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2888468579809689e-06,
+      "loss": 2.4782,
+      "step": 6495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2875017321253857e-06,
+      "loss": 2.3116,
+      "step": 6496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.286157204830667e-06,
+      "loss": 2.47,
+      "step": 6497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.284813276313584e-06,
+      "loss": 2.4635,
+      "step": 6498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2834699467908184e-06,
+      "loss": 2.4826,
+      "step": 6499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2821272164789544e-06,
+      "loss": 2.6379,
+      "step": 6500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2807850855944753e-06,
+      "loss": 2.4043,
+      "step": 6501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2794435543537737e-06,
+      "loss": 2.4441,
+      "step": 6502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2781026229731425e-06,
+      "loss": 2.5198,
+      "step": 6503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2767622916687767e-06,
+      "loss": 2.459,
+      "step": 6504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2754225606567783e-06,
+      "loss": 2.6249,
+      "step": 6505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2740834301531468e-06,
+      "loss": 2.3966,
+      "step": 6506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2727449003737913e-06,
+      "loss": 2.4985,
+      "step": 6507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2714069715345195e-06,
+      "loss": 2.494,
+      "step": 6508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2700696438510445e-06,
+      "loss": 2.4309,
+      "step": 6509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2687329175389835e-06,
+      "loss": 2.6146,
+      "step": 6510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2673967928138514e-06,
+      "loss": 2.44,
+      "step": 6511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.266061269891073e-06,
+      "loss": 2.5612,
+      "step": 6512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.26472634898597e-06,
+      "loss": 2.6592,
+      "step": 6513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2633920303137708e-06,
+      "loss": 2.5469,
+      "step": 6514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2620583140896075e-06,
+      "loss": 2.3146,
+      "step": 6515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2607252005285109e-06,
+      "loss": 2.5924,
+      "step": 6516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.259392689845419e-06,
+      "loss": 2.4271,
+      "step": 6517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2580607822551677e-06,
+      "loss": 2.4513,
+      "step": 6518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2567294779725008e-06,
+      "loss": 2.516,
+      "step": 6519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2553987772120619e-06,
+      "loss": 2.5777,
+      "step": 6520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2540686801883989e-06,
+      "loss": 2.7267,
+      "step": 6521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2527391871159616e-06,
+      "loss": 2.5439,
+      "step": 6522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.251410298209101e-06,
+      "loss": 2.5735,
+      "step": 6523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2500820136820735e-06,
+      "loss": 2.5534,
+      "step": 6524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2487543337490343e-06,
+      "loss": 2.2641,
+      "step": 6525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2474272586240444e-06,
+      "loss": 2.5682,
+      "step": 6526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2461007885210685e-06,
+      "loss": 2.5803,
+      "step": 6527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2447749236539674e-06,
+      "loss": 2.5273,
+      "step": 6528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2434496642365123e-06,
+      "loss": 2.4944,
+      "step": 6529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2421250104823683e-06,
+      "loss": 2.5581,
+      "step": 6530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2408009626051137e-06,
+      "loss": 2.7278,
+      "step": 6531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2394775208182175e-06,
+      "loss": 2.4026,
+      "step": 6532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2381546853350584e-06,
+      "loss": 2.5554,
+      "step": 6533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2368324563689171e-06,
+      "loss": 2.43,
+      "step": 6534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2355108341329708e-06,
+      "loss": 2.4993,
+      "step": 6535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2341898188403068e-06,
+      "loss": 2.428,
+      "step": 6536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.232869410703907e-06,
+      "loss": 2.5324,
+      "step": 6537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2315496099366603e-06,
+      "loss": 2.4588,
+      "step": 6538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2302304167513586e-06,
+      "loss": 2.4715,
+      "step": 6539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2289118313606895e-06,
+      "loss": 2.5146,
+      "step": 6540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2275938539772485e-06,
+      "loss": 2.5291,
+      "step": 6541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2262764848135316e-06,
+      "loss": 2.5888,
+      "step": 6542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2249597240819377e-06,
+      "loss": 2.4507,
+      "step": 6543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2236435719947638e-06,
+      "loss": 2.4703,
+      "step": 6544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2223280287642115e-06,
+      "loss": 2.6027,
+      "step": 6545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2210130946023862e-06,
+      "loss": 2.4591,
+      "step": 6546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2196987697212903e-06,
+      "loss": 2.406,
+      "step": 6547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2183850543328313e-06,
+      "loss": 2.4112,
+      "step": 6548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2170719486488192e-06,
+      "loss": 2.3469,
+      "step": 6549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2157594528809619e-06,
+      "loss": 2.4943,
+      "step": 6550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2144475672408733e-06,
+      "loss": 2.3436,
+      "step": 6551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2131362919400635e-06,
+      "loss": 2.4741,
+      "step": 6552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2118256271899525e-06,
+      "loss": 2.4925,
+      "step": 6553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2105155732018532e-06,
+      "loss": 2.5788,
+      "step": 6554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.209206130186985e-06,
+      "loss": 2.5773,
+      "step": 6555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2078972983564686e-06,
+      "loss": 2.3581,
+      "step": 6556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.206589077921323e-06,
+      "loss": 2.5353,
+      "step": 6557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2052814690924724e-06,
+      "loss": 2.5795,
+      "step": 6558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2039744720807396e-06,
+      "loss": 2.5837,
+      "step": 6559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2026680870968494e-06,
+      "loss": 2.4573,
+      "step": 6560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2013623143514314e-06,
+      "loss": 2.4834,
+      "step": 6561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2000571540550092e-06,
+      "loss": 2.4639,
+      "step": 6562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1987526064180149e-06,
+      "loss": 2.5363,
+      "step": 6563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1974486716507782e-06,
+      "loss": 2.5597,
+      "step": 6564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1961453499635323e-06,
+      "loss": 2.5179,
+      "step": 6565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.194842641566406e-06,
+      "loss": 2.4419,
+      "step": 6566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1935405466694362e-06,
+      "loss": 2.4785,
+      "step": 6567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1922390654825582e-06,
+      "loss": 2.4786,
+      "step": 6568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1909381982156054e-06,
+      "loss": 2.4695,
+      "step": 6569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1896379450783174e-06,
+      "loss": 2.3857,
+      "step": 6570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1883383062803293e-06,
+      "loss": 2.5147,
+      "step": 6571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.187039282031182e-06,
+      "loss": 2.4763,
+      "step": 6572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.185740872540317e-06,
+      "loss": 2.5528,
+      "step": 6573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1844430780170697e-06,
+      "loss": 2.4968,
+      "step": 6574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1831458986706879e-06,
+      "loss": 2.622,
+      "step": 6575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1818493347103098e-06,
+      "loss": 2.393,
+      "step": 6576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1805533863449813e-06,
+      "loss": 2.4127,
+      "step": 6577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.179258053783644e-06,
+      "loss": 2.6365,
+      "step": 6578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.177963337235143e-06,
+      "loss": 2.5919,
+      "step": 6579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1766692369082255e-06,
+      "loss": 2.5602,
+      "step": 6580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1753757530115351e-06,
+      "loss": 2.4821,
+      "step": 6581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1740828857536206e-06,
+      "loss": 2.4298,
+      "step": 6582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1727906353429264e-06,
+      "loss": 2.5387,
+      "step": 6583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.171499001987802e-06,
+      "loss": 2.4051,
+      "step": 6584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1702079858964953e-06,
+      "loss": 2.5807,
+      "step": 6585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1689175872771552e-06,
+      "loss": 2.5333,
+      "step": 6586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1676278063378322e-06,
+      "loss": 2.4234,
+      "step": 6587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1663386432864725e-06,
+      "loss": 2.49,
+      "step": 6588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.16505009833093e-06,
+      "loss": 2.5732,
+      "step": 6589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.163762171678951e-06,
+      "loss": 2.4577,
+      "step": 6590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1624748635381883e-06,
+      "loss": 2.759,
+      "step": 6591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.161188174116194e-06,
+      "loss": 2.5374,
+      "step": 6592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.159902103620416e-06,
+      "loss": 2.3443,
+      "step": 6593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1586166522582094e-06,
+      "loss": 2.5762,
+      "step": 6594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1573318202368217e-06,
+      "loss": 2.6016,
+      "step": 6595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.156047607763407e-06,
+      "loss": 2.5861,
+      "step": 6596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1547640150450162e-06,
+      "loss": 2.7419,
+      "step": 6597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1534810422886016e-06,
+      "loss": 2.3385,
+      "step": 6598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1521986897010163e-06,
+      "loss": 2.7658,
+      "step": 6599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.15091695748901e-06,
+      "loss": 2.5143,
+      "step": 6600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.149635845859235e-06,
+      "loss": 2.47,
+      "step": 6601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1483553550182453e-06,
+      "loss": 2.6732,
+      "step": 6602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1470754851724897e-06,
+      "loss": 2.3553,
+      "step": 6603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.145796236528322e-06,
+      "loss": 2.5309,
+      "step": 6604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1445176092919918e-06,
+      "loss": 2.3967,
+      "step": 6605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1432396036696508e-06,
+      "loss": 2.5617,
+      "step": 6606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1419622198673512e-06,
+      "loss": 2.427,
+      "step": 6607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1406854580910426e-06,
+      "loss": 2.4354,
+      "step": 6608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1394093185465776e-06,
+      "loss": 2.3972,
+      "step": 6609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1381338014397035e-06,
+      "loss": 2.5399,
+      "step": 6610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1368589069760728e-06,
+      "loss": 2.5406,
+      "step": 6611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.135584635361232e-06,
+      "loss": 2.4687,
+      "step": 6612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.134310986800632e-06,
+      "loss": 2.6946,
+      "step": 6613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1330379614996223e-06,
+      "loss": 2.4416,
+      "step": 6614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.131765559663448e-06,
+      "loss": 2.4813,
+      "step": 6615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.13049378149726e-06,
+      "loss": 2.4994,
+      "step": 6616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1292226272061023e-06,
+      "loss": 2.509,
+      "step": 6617
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1279520969949225e-06,
+      "loss": 2.4326,
+      "step": 6618
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1266821910685671e-06,
+      "loss": 2.4436,
+      "step": 6619
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1254129096317807e-06,
+      "loss": 2.4042,
+      "step": 6620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1241442528892088e-06,
+      "loss": 2.4786,
+      "step": 6621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1228762210453926e-06,
+      "loss": 2.5781,
+      "step": 6622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1216088143047781e-06,
+      "loss": 2.5228,
+      "step": 6623
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1203420328717052e-06,
+      "loss": 2.6158,
+      "step": 6624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1190758769504156e-06,
+      "loss": 2.349,
+      "step": 6625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1178103467450512e-06,
+      "loss": 2.4841,
+      "step": 6626
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.11654544245965e-06,
+      "loss": 2.5912,
+      "step": 6627
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.115281164298153e-06,
+      "loss": 2.5283,
+      "step": 6628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1140175124643948e-06,
+      "loss": 2.4902,
+      "step": 6629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1127544871621144e-06,
+      "loss": 2.4647,
+      "step": 6630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1114920885949466e-06,
+      "loss": 2.4614,
+      "step": 6631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.110230316966427e-06,
+      "loss": 2.4321,
+      "step": 6632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.10896917247999e-06,
+      "loss": 2.4033,
+      "step": 6633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1077086553389654e-06,
+      "loss": 2.5009,
+      "step": 6634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1064487657465883e-06,
+      "loss": 2.602,
+      "step": 6635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1051895039059851e-06,
+      "loss": 2.471,
+      "step": 6636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1039308700201868e-06,
+      "loss": 2.7274,
+      "step": 6637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.102672864292122e-06,
+      "loss": 2.5045,
+      "step": 6638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1014154869246146e-06,
+      "loss": 2.5104,
+      "step": 6639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1001587381203931e-06,
+      "loss": 2.4347,
+      "step": 6640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0989026180820766e-06,
+      "loss": 2.3562,
+      "step": 6641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.097647127012193e-06,
+      "loss": 2.4224,
+      "step": 6642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0963922651131598e-06,
+      "loss": 2.5556,
+      "step": 6643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.095138032587298e-06,
+      "loss": 2.4572,
+      "step": 6644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0938844296368262e-06,
+      "loss": 2.5073,
+      "step": 6645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.092631456463859e-06,
+      "loss": 2.5863,
+      "step": 6646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0913791132704144e-06,
+      "loss": 2.3776,
+      "step": 6647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0901274002584029e-06,
+      "loss": 2.3049,
+      "step": 6648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.088876317629638e-06,
+      "loss": 2.5818,
+      "step": 6649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0876258655858307e-06,
+      "loss": 2.4197,
+      "step": 6650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0863760443285876e-06,
+      "loss": 2.5391,
+      "step": 6651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0851268540594168e-06,
+      "loss": 2.4959,
+      "step": 6652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0838782949797227e-06,
+      "loss": 2.4136,
+      "step": 6653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.08263036729081e-06,
+      "loss": 2.4707,
+      "step": 6654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0813830711938816e-06,
+      "loss": 2.4599,
+      "step": 6655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0801364068900334e-06,
+      "loss": 2.4183,
+      "step": 6656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.078890374580267e-06,
+      "loss": 2.4569,
+      "step": 6657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0776449744654755e-06,
+      "loss": 2.4535,
+      "step": 6658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.076400206746454e-06,
+      "loss": 2.5795,
+      "step": 6659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0751560716238968e-06,
+      "loss": 2.5735,
+      "step": 6660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0739125692983904e-06,
+      "loss": 2.5768,
+      "step": 6661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0726696999704266e-06,
+      "loss": 2.6314,
+      "step": 6662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0714274638403865e-06,
+      "loss": 2.673,
+      "step": 6663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0701858611085602e-06,
+      "loss": 2.6326,
+      "step": 6664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.068944891975125e-06,
+      "loss": 2.5631,
+      "step": 6665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0677045566401623e-06,
+      "loss": 2.5143,
+      "step": 6666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0664648553036505e-06,
+      "loss": 2.5849,
+      "step": 6667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0652257881654625e-06,
+      "loss": 2.2979,
+      "step": 6668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0639873554253743e-06,
+      "loss": 2.6429,
+      "step": 6669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0627495572830537e-06,
+      "loss": 2.5491,
+      "step": 6670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0615123939380707e-06,
+      "loss": 2.5859,
+      "step": 6671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.060275865589892e-06,
+      "loss": 2.4764,
+      "step": 6672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0590399724378797e-06,
+      "loss": 2.5178,
+      "step": 6673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.057804714681296e-06,
+      "loss": 2.4703,
+      "step": 6674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0565700925192995e-06,
+      "loss": 2.4716,
+      "step": 6675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0553361061509482e-06,
+      "loss": 2.3997,
+      "step": 6676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0541027557751937e-06,
+      "loss": 2.5094,
+      "step": 6677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0528700415908876e-06,
+      "loss": 2.5093,
+      "step": 6678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.051637963796781e-06,
+      "loss": 2.5762,
+      "step": 6679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0504065225915166e-06,
+      "loss": 2.4511,
+      "step": 6680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0491757181736418e-06,
+      "loss": 2.4777,
+      "step": 6681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0479455507415937e-06,
+      "loss": 2.317,
+      "step": 6682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0467160204937115e-06,
+      "loss": 2.3329,
+      "step": 6683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0454871276282335e-06,
+      "loss": 2.5611,
+      "step": 6684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0442588723432872e-06,
+      "loss": 2.4642,
+      "step": 6685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0430312548369076e-06,
+      "loss": 2.4513,
+      "step": 6686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.041804275307019e-06,
+      "loss": 2.516,
+      "step": 6687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0405779339514466e-06,
+      "loss": 2.6282,
+      "step": 6688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0393522309679106e-06,
+      "loss": 2.3808,
+      "step": 6689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0381271665540294e-06,
+      "loss": 2.4726,
+      "step": 6690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0369027409073207e-06,
+      "loss": 2.4192,
+      "step": 6691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0356789542251939e-06,
+      "loss": 2.5391,
+      "step": 6692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0344558067049603e-06,
+      "loss": 2.5699,
+      "step": 6693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0332332985438248e-06,
+      "loss": 2.3571,
+      "step": 6694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0320114299388916e-06,
+      "loss": 2.5828,
+      "step": 6695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0307902010871606e-06,
+      "loss": 2.4573,
+      "step": 6696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0295696121855292e-06,
+      "loss": 2.4675,
+      "step": 6697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.028349663430792e-06,
+      "loss": 2.3509,
+      "step": 6698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0271303550196376e-06,
+      "loss": 2.4551,
+      "step": 6699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0259116871486557e-06,
+      "loss": 2.5644,
+      "step": 6700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0246936600143276e-06,
+      "loss": 2.4142,
+      "step": 6701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0234762738130354e-06,
+      "loss": 2.7077,
+      "step": 6702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0222595287410586e-06,
+      "loss": 2.4355,
+      "step": 6703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0210434249945677e-06,
+      "loss": 2.526,
+      "step": 6704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.019827962769635e-06,
+      "loss": 2.4043,
+      "step": 6705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0186131422622297e-06,
+      "loss": 2.3991,
+      "step": 6706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0173989636682109e-06,
+      "loss": 2.3965,
+      "step": 6707
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0161854271833444e-06,
+      "loss": 2.7059,
+      "step": 6708
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0149725330032827e-06,
+      "loss": 2.3652,
+      "step": 6709
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0137602813235824e-06,
+      "loss": 2.4159,
+      "step": 6710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0125486723396905e-06,
+      "loss": 2.4414,
+      "step": 6711
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0113377062469532e-06,
+      "loss": 2.5917,
+      "step": 6712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0101273832406156e-06,
+      "loss": 2.5579,
+      "step": 6713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0089177035158127e-06,
+      "loss": 2.4856,
+      "step": 6714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0077086672675823e-06,
+      "loss": 2.5277,
+      "step": 6715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0065002746908532e-06,
+      "loss": 2.4392,
+      "step": 6716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0052925259804536e-06,
+      "loss": 2.4239,
+      "step": 6717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0040854213311096e-06,
+      "loss": 2.5553,
+      "step": 6718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0028789609374356e-06,
+      "loss": 2.4595,
+      "step": 6719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0016731449939543e-06,
+      "loss": 2.6189,
+      "step": 6720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0004679736950722e-06,
+      "loss": 2.5324,
+      "step": 6721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.992634472351016e-07,
+      "loss": 2.5018,
+      "step": 6722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.980595658082426e-07,
+      "loss": 2.5164,
+      "step": 6723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.96856329608597e-07,
+      "loss": 2.475,
+      "step": 6724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.956537388301623e-07,
+      "loss": 2.4222,
+      "step": 6725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.944517936668285e-07,
+      "loss": 2.5997,
+      "step": 6726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.93250494312385e-07,
+      "loss": 2.6606,
+      "step": 6727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.92049840960514e-07,
+      "loss": 2.5187,
+      "step": 6728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.908498338047956e-07,
+      "loss": 2.3918,
+      "step": 6729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.896504730387059e-07,
+      "loss": 2.4241,
+      "step": 6730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.884517588556158e-07,
+      "loss": 2.5145,
+      "step": 6731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.87253691448794e-07,
+      "loss": 2.4539,
+      "step": 6732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.860562710113996e-07,
+      "loss": 2.72,
+      "step": 6733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.84859497736495e-07,
+      "loss": 2.5777,
+      "step": 6734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.836633718170308e-07,
+      "loss": 2.421,
+      "step": 6735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.82467893445858e-07,
+      "loss": 2.3478,
+      "step": 6736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.81273062815724e-07,
+      "loss": 2.4038,
+      "step": 6737
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.800788801192662e-07,
+      "loss": 2.5633,
+      "step": 6738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.788853455490232e-07,
+      "loss": 2.4342,
+      "step": 6739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.776924592974257e-07,
+      "loss": 2.4838,
+      "step": 6740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.765002215568015e-07,
+      "loss": 2.4236,
+      "step": 6741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.753086325193733e-07,
+      "loss": 2.5882,
+      "step": 6742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.741176923772593e-07,
+      "loss": 2.6417,
+      "step": 6743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.729274013224754e-07,
+      "loss": 2.4518,
+      "step": 6744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.717377595469269e-07,
+      "loss": 2.6279,
+      "step": 6745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.705487672424202e-07,
+      "loss": 2.6027,
+      "step": 6746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.69360424600654e-07,
+      "loss": 2.4802,
+      "step": 6747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.681727318132228e-07,
+      "loss": 2.3432,
+      "step": 6748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.669856890716188e-07,
+      "loss": 2.4321,
+      "step": 6749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.657992965672242e-07,
+      "loss": 2.4774,
+      "step": 6750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.646135544913226e-07,
+      "loss": 2.4578,
+      "step": 6751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.63428463035085e-07,
+      "loss": 2.6671,
+      "step": 6752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.622440223895868e-07,
+      "loss": 2.4264,
+      "step": 6753
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.610602327457913e-07,
+      "loss": 2.5435,
+      "step": 6754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.59877094294559e-07,
+      "loss": 2.4945,
+      "step": 6755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.586946072266479e-07,
+      "loss": 2.5343,
+      "step": 6756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.57512771732706e-07,
+      "loss": 2.4736,
+      "step": 6757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.563315880032798e-07,
+      "loss": 2.6527,
+      "step": 6758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.551510562288112e-07,
+      "loss": 2.3592,
+      "step": 6759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.539711765996334e-07,
+      "loss": 2.3183,
+      "step": 6760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.527919493059801e-07,
+      "loss": 2.3207,
+      "step": 6761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.516133745379718e-07,
+      "loss": 2.5467,
+      "step": 6762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.504354524856318e-07,
+      "loss": 2.414,
+      "step": 6763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.492581833388736e-07,
+      "loss": 2.4635,
+      "step": 6764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.480815672875077e-07,
+      "loss": 2.4049,
+      "step": 6765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.469056045212382e-07,
+      "loss": 2.5578,
+      "step": 6766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.457302952296621e-07,
+      "loss": 2.4616,
+      "step": 6767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.445556396022754e-07,
+      "loss": 2.5129,
+      "step": 6768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.433816378284638e-07,
+      "loss": 2.4566,
+      "step": 6769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.422082900975105e-07,
+      "loss": 2.4696,
+      "step": 6770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.410355965985951e-07,
+      "loss": 2.4196,
+      "step": 6771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.398635575207854e-07,
+      "loss": 2.6637,
+      "step": 6772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.386921730530513e-07,
+      "loss": 2.5521,
+      "step": 6773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.375214433842489e-07,
+      "loss": 2.4868,
+      "step": 6774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.363513687031389e-07,
+      "loss": 2.5464,
+      "step": 6775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.351819491983671e-07,
+      "loss": 2.5319,
+      "step": 6776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.340131850584777e-07,
+      "loss": 2.4377,
+      "step": 6777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.32845076471911e-07,
+      "loss": 2.4328,
+      "step": 6778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.31677623626997e-07,
+      "loss": 2.444,
+      "step": 6779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.305108267119645e-07,
+      "loss": 2.3913,
+      "step": 6780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.293446859149324e-07,
+      "loss": 2.5384,
+      "step": 6781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.281792014239171e-07,
+      "loss": 2.5745,
+      "step": 6782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.270143734268289e-07,
+      "loss": 2.6119,
+      "step": 6783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.258502021114696e-07,
+      "loss": 2.3984,
+      "step": 6784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.246866876655375e-07,
+      "loss": 2.5093,
+      "step": 6785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.235238302766247e-07,
+      "loss": 2.4536,
+      "step": 6786
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.22361630132218e-07,
+      "loss": 2.2997,
+      "step": 6787
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.212000874196953e-07,
+      "loss": 2.6088,
+      "step": 6788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.200392023263311e-07,
+      "loss": 2.5615,
+      "step": 6789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.188789750392951e-07,
+      "loss": 2.4818,
+      "step": 6790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.177194057456468e-07,
+      "loss": 2.4207,
+      "step": 6791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.165604946323436e-07,
+      "loss": 2.4474,
+      "step": 6792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.154022418862329e-07,
+      "loss": 2.4188,
+      "step": 6793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.142446476940598e-07,
+      "loss": 2.5743,
+      "step": 6794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.130877122424625e-07,
+      "loss": 2.3427,
+      "step": 6795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.119314357179687e-07,
+      "loss": 2.4942,
+      "step": 6796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.107758183070076e-07,
+      "loss": 2.5176,
+      "step": 6797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.09620860195894e-07,
+      "loss": 2.6185,
+      "step": 6798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.084665615708432e-07,
+      "loss": 2.513,
+      "step": 6799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.073129226179589e-07,
+      "loss": 2.5216,
+      "step": 6800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.06159943523241e-07,
+      "loss": 2.5273,
+      "step": 6801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.050076244725847e-07,
+      "loss": 2.6114,
+      "step": 6802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.038559656517737e-07,
+      "loss": 2.61,
+      "step": 6803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.027049672464916e-07,
+      "loss": 2.438,
+      "step": 6804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.015546294423089e-07,
+      "loss": 2.4383,
+      "step": 6805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.004049524246944e-07,
+      "loss": 2.3922,
+      "step": 6806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.992559363790093e-07,
+      "loss": 2.413,
+      "step": 6807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.981075814905077e-07,
+      "loss": 2.3703,
+      "step": 6808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.969598879443381e-07,
+      "loss": 2.4278,
+      "step": 6809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.958128559255397e-07,
+      "loss": 2.5568,
+      "step": 6810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.946664856190473e-07,
+      "loss": 2.5371,
+      "step": 6811
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.935207772096904e-07,
+      "loss": 2.4801,
+      "step": 6812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.923757308821873e-07,
+      "loss": 2.5355,
+      "step": 6813
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.912313468211547e-07,
+      "loss": 2.3753,
+      "step": 6814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.900876252110974e-07,
+      "loss": 2.2935,
+      "step": 6815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.889445662364177e-07,
+      "loss": 2.4367,
+      "step": 6816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.878021700814099e-07,
+      "loss": 2.5967,
+      "step": 6817
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.86660436930259e-07,
+      "loss": 2.6055,
+      "step": 6818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.855193669670459e-07,
+      "loss": 2.5807,
+      "step": 6819
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.843789603757446e-07,
+      "loss": 2.555,
+      "step": 6820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.83239217340221e-07,
+      "loss": 2.5251,
+      "step": 6821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.821001380442334e-07,
+      "loss": 2.4014,
+      "step": 6822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.80961722671434e-07,
+      "loss": 2.6247,
+      "step": 6823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.798239714053692e-07,
+      "loss": 2.4878,
+      "step": 6824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.786868844294755e-07,
+      "loss": 2.5047,
+      "step": 6825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.775504619270852e-07,
+      "loss": 2.3376,
+      "step": 6826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.764147040814203e-07,
+      "loss": 2.5749,
+      "step": 6827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.752796110755985e-07,
+      "loss": 2.5368,
+      "step": 6828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.741451830926301e-07,
+      "loss": 2.5531,
+      "step": 6829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.730114203154144e-07,
+      "loss": 2.4628,
+      "step": 6830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.718783229267508e-07,
+      "loss": 2.3553,
+      "step": 6831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.707458911093231e-07,
+      "loss": 2.5392,
+      "step": 6832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.696141250457152e-07,
+      "loss": 2.5145,
+      "step": 6833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.684830249183962e-07,
+      "loss": 2.6096,
+      "step": 6834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.673525909097342e-07,
+      "loss": 2.5186,
+      "step": 6835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.662228232019876e-07,
+      "loss": 2.43,
+      "step": 6836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.65093721977306e-07,
+      "loss": 2.5191,
+      "step": 6837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.639652874177346e-07,
+      "loss": 2.5696,
+      "step": 6838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.628375197052069e-07,
+      "loss": 2.4231,
+      "step": 6839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.617104190215525e-07,
+      "loss": 2.4872,
+      "step": 6840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.60583985548492e-07,
+      "loss": 2.5507,
+      "step": 6841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.594582194676393e-07,
+      "loss": 2.4487,
+      "step": 6842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.583331209605011e-07,
+      "loss": 2.5925,
+      "step": 6843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.572086902084731e-07,
+      "loss": 2.3633,
+      "step": 6844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.560849273928479e-07,
+      "loss": 2.5216,
+      "step": 6845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.549618326948056e-07,
+      "loss": 2.4321,
+      "step": 6846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.538394062954225e-07,
+      "loss": 2.4959,
+      "step": 6847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.527176483756671e-07,
+      "loss": 2.3352,
+      "step": 6848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.515965591163966e-07,
+      "loss": 2.4374,
+      "step": 6849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.504761386983651e-07,
+      "loss": 2.6216,
+      "step": 6850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.493563873022137e-07,
+      "loss": 2.6362,
+      "step": 6851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.482373051084791e-07,
+      "loss": 2.5443,
+      "step": 6852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.471188922975903e-07,
+      "loss": 2.4744,
+      "step": 6853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.460011490498671e-07,
+      "loss": 2.6094,
+      "step": 6854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.448840755455229e-07,
+      "loss": 2.5169,
+      "step": 6855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.437676719646593e-07,
+      "loss": 2.4628,
+      "step": 6856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.426519384872733e-07,
+      "loss": 2.5622,
+      "step": 6857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.415368752932551e-07,
+      "loss": 2.5443,
+      "step": 6858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.404224825623819e-07,
+      "loss": 2.5632,
+      "step": 6859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.393087604743283e-07,
+      "loss": 2.4735,
+      "step": 6860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.381957092086551e-07,
+      "loss": 2.436,
+      "step": 6861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.370833289448199e-07,
+      "loss": 2.5449,
+      "step": 6862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.359716198621698e-07,
+      "loss": 2.4819,
+      "step": 6863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.348605821399441e-07,
+      "loss": 2.5689,
+      "step": 6864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.337502159572758e-07,
+      "loss": 2.3686,
+      "step": 6865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.326405214931838e-07,
+      "loss": 2.5153,
+      "step": 6866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.315314989265866e-07,
+      "loss": 2.6249,
+      "step": 6867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.304231484362868e-07,
+      "loss": 2.5751,
+      "step": 6868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.293154702009837e-07,
+      "loss": 2.3922,
+      "step": 6869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.282084643992677e-07,
+      "loss": 2.4298,
+      "step": 6870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.271021312096178e-07,
+      "loss": 2.4978,
+      "step": 6871
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.259964708104091e-07,
+      "loss": 2.5672,
+      "step": 6872
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.248914833799027e-07,
+      "loss": 2.5524,
+      "step": 6873
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.237871690962557e-07,
+      "loss": 2.611,
+      "step": 6874
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.226835281375151e-07,
+      "loss": 2.4287,
+      "step": 6875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.215805606816191e-07,
+      "loss": 2.515,
+      "step": 6876
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.204782669063993e-07,
+      "loss": 2.398,
+      "step": 6877
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.193766469895737e-07,
+      "loss": 2.497,
+      "step": 6878
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.182757011087583e-07,
+      "loss": 2.5607,
+      "step": 6879
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.171754294414546e-07,
+      "loss": 2.3995,
+      "step": 6880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.16075832165058e-07,
+      "loss": 2.495,
+      "step": 6881
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.149769094568577e-07,
+      "loss": 2.5406,
+      "step": 6882
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.138786614940275e-07,
+      "loss": 2.5584,
+      "step": 6883
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.127810884536402e-07,
+      "loss": 2.6382,
+      "step": 6884
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.11684190512651e-07,
+      "loss": 2.4393,
+      "step": 6885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.105879678479172e-07,
+      "loss": 2.206,
+      "step": 6886
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.094924206361771e-07,
+      "loss": 2.5761,
+      "step": 6887
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.083975490540658e-07,
+      "loss": 2.5352,
+      "step": 6888
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.073033532781083e-07,
+      "loss": 2.3965,
+      "step": 6889
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.062098334847185e-07,
+      "loss": 2.531,
+      "step": 6890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.051169898502053e-07,
+      "loss": 2.4125,
+      "step": 6891
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.040248225507641e-07,
+      "loss": 2.4702,
+      "step": 6892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.029333317624838e-07,
+      "loss": 2.5914,
+      "step": 6893
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.018425176613464e-07,
+      "loss": 2.3309,
+      "step": 6894
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.007523804232192e-07,
+      "loss": 2.5323,
+      "step": 6895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.996629202238654e-07,
+      "loss": 2.454,
+      "step": 6896
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.985741372389361e-07,
+      "loss": 2.4252,
+      "step": 6897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.974860316439759e-07,
+      "loss": 2.4812,
+      "step": 6898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.963986036144161e-07,
+      "loss": 2.4529,
+      "step": 6899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.953118533255821e-07,
+      "loss": 2.4786,
+      "step": 6900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.94225780952691e-07,
+      "loss": 2.4672,
+      "step": 6901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.931403866708464e-07,
+      "loss": 2.5062,
+      "step": 6902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.920556706550464e-07,
+      "loss": 2.579,
+      "step": 6903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.909716330801765e-07,
+      "loss": 2.6029,
+      "step": 6904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.898882741210151e-07,
+      "loss": 2.6966,
+      "step": 6905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.888055939522327e-07,
+      "loss": 2.5529,
+      "step": 6906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.877235927483856e-07,
+      "loss": 2.5558,
+      "step": 6907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.866422706839239e-07,
+      "loss": 2.463,
+      "step": 6908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.855616279331885e-07,
+      "loss": 2.4322,
+      "step": 6909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.844816646704096e-07,
+      "loss": 2.527,
+      "step": 6910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.834023810697088e-07,
+      "loss": 2.5621,
+      "step": 6911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.823237773050963e-07,
+      "loss": 2.5232,
+      "step": 6912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.812458535504752e-07,
+      "loss": 2.386,
+      "step": 6913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.801686099796358e-07,
+      "loss": 2.6354,
+      "step": 6914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.790920467662611e-07,
+      "loss": 2.532,
+      "step": 6915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.780161640839257e-07,
+      "loss": 2.4484,
+      "step": 6916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.769409621060903e-07,
+      "loss": 2.6563,
+      "step": 6917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.758664410061095e-07,
+      "loss": 2.5626,
+      "step": 6918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.747926009572249e-07,
+      "loss": 2.5439,
+      "step": 6919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.737194421325739e-07,
+      "loss": 2.5609,
+      "step": 6920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.726469647051771e-07,
+      "loss": 2.5251,
+      "step": 6921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.7157516884795e-07,
+      "loss": 2.5114,
+      "step": 6922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.705040547336973e-07,
+      "loss": 2.6372,
+      "step": 6923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.694336225351107e-07,
+      "loss": 2.5883,
+      "step": 6924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.683638724247778e-07,
+      "loss": 2.5683,
+      "step": 6925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.672948045751699e-07,
+      "loss": 2.3527,
+      "step": 6926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.662264191586533e-07,
+      "loss": 2.5291,
+      "step": 6927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.651587163474822e-07,
+      "loss": 2.5628,
+      "step": 6928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.64091696313799e-07,
+      "loss": 2.4565,
+      "step": 6929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.630253592296388e-07,
+      "loss": 2.4799,
+      "step": 6930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.619597052669265e-07,
+      "loss": 2.5149,
+      "step": 6931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.60894734597476e-07,
+      "loss": 2.4274,
+      "step": 6932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.59830447392989e-07,
+      "loss": 2.6478,
+      "step": 6933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.587668438250612e-07,
+      "loss": 2.6902,
+      "step": 6934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.577039240651757e-07,
+      "loss": 2.5341,
+      "step": 6935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.566416882847039e-07,
+      "loss": 2.3667,
+      "step": 6936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.555801366549109e-07,
+      "loss": 2.5559,
+      "step": 6937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.545192693469467e-07,
+      "loss": 2.4354,
+      "step": 6938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.534590865318547e-07,
+      "loss": 2.3792,
+      "step": 6939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.52399588380568e-07,
+      "loss": 2.4067,
+      "step": 6940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.51340775063904e-07,
+      "loss": 2.5818,
+      "step": 6941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.502826467525787e-07,
+      "loss": 2.5167,
+      "step": 6942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.492252036171893e-07,
+      "loss": 2.4099,
+      "step": 6943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.481684458282273e-07,
+      "loss": 2.3273,
+      "step": 6944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.471123735560715e-07,
+      "loss": 2.6003,
+      "step": 6945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.460569869709905e-07,
+      "loss": 2.47,
+      "step": 6946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.450022862431449e-07,
+      "loss": 2.5858,
+      "step": 6947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.439482715425806e-07,
+      "loss": 2.571,
+      "step": 6948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.428949430392357e-07,
+      "loss": 2.3359,
+      "step": 6949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.418423009029363e-07,
+      "loss": 2.6286,
+      "step": 6950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.407903453033982e-07,
+      "loss": 2.4648,
+      "step": 6951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.39739076410228e-07,
+      "loss": 2.4452,
+      "step": 6952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.38688494392919e-07,
+      "loss": 2.5696,
+      "step": 6953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.376385994208574e-07,
+      "loss": 2.5671,
+      "step": 6954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.365893916633132e-07,
+      "loss": 2.6858,
+      "step": 6955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.355408712894508e-07,
+      "loss": 2.4907,
+      "step": 6956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.344930384683202e-07,
+      "loss": 2.6572,
+      "step": 6957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.334458933688626e-07,
+      "loss": 2.4569,
+      "step": 6958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.323994361599091e-07,
+      "loss": 2.3819,
+      "step": 6959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.313536670101762e-07,
+      "loss": 2.441,
+      "step": 6960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.303085860882725e-07,
+      "loss": 2.6341,
+      "step": 6961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.292641935626966e-07,
+      "loss": 2.686,
+      "step": 6962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.282204896018307e-07,
+      "loss": 2.4708,
+      "step": 6963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.271774743739546e-07,
+      "loss": 2.3816,
+      "step": 6964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.261351480472284e-07,
+      "loss": 2.5959,
+      "step": 6965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.250935107897078e-07,
+      "loss": 2.4079,
+      "step": 6966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.240525627693312e-07,
+      "loss": 2.3906,
+      "step": 6967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.23012304153931e-07,
+      "loss": 2.577,
+      "step": 6968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.219727351112276e-07,
+      "loss": 2.46,
+      "step": 6969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.209338558088275e-07,
+      "loss": 2.473,
+      "step": 6970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.198956664142292e-07,
+      "loss": 2.3829,
+      "step": 6971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.18858167094817e-07,
+      "loss": 2.5336,
+      "step": 6972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.178213580178655e-07,
+      "loss": 2.5946,
+      "step": 6973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.167852393505393e-07,
+      "loss": 2.4993,
+      "step": 6974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.157498112598893e-07,
+      "loss": 2.5622,
+      "step": 6975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.147150739128577e-07,
+      "loss": 2.379,
+      "step": 6976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.136810274762717e-07,
+      "loss": 2.5264,
+      "step": 6977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.126476721168513e-07,
+      "loss": 2.4595,
+      "step": 6978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.116150080011996e-07,
+      "loss": 2.7064,
+      "step": 6979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.105830352958143e-07,
+      "loss": 2.6926,
+      "step": 6980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.095517541670793e-07,
+      "loss": 2.4068,
+      "step": 6981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.085211647812645e-07,
+      "loss": 2.5204,
+      "step": 6982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.07491267304532e-07,
+      "loss": 2.3869,
+      "step": 6983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.064620619029294e-07,
+      "loss": 2.6386,
+      "step": 6984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.054335487423946e-07,
+      "loss": 2.4831,
+      "step": 6985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.04405727988754e-07,
+      "loss": 2.6016,
+      "step": 6986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.033785998077202e-07,
+      "loss": 2.5212,
+      "step": 6987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.023521643648984e-07,
+      "loss": 2.4323,
+      "step": 6988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.013264218257765e-07,
+      "loss": 2.5483,
+      "step": 6989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.003013723557356e-07,
+      "loss": 2.5896,
+      "step": 6990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.992770161200402e-07,
+      "loss": 2.5293,
+      "step": 6991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.982533532838481e-07,
+      "loss": 2.4751,
+      "step": 6992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.972303840122036e-07,
+      "loss": 2.3935,
+      "step": 6993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.962081084700361e-07,
+      "loss": 2.4752,
+      "step": 6994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.951865268221686e-07,
+      "loss": 2.4264,
+      "step": 6995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.941656392333046e-07,
+      "loss": 2.5385,
+      "step": 6996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.931454458680464e-07,
+      "loss": 2.4379,
+      "step": 6997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.921259468908742e-07,
+      "loss": 2.5093,
+      "step": 6998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.911071424661614e-07,
+      "loss": 2.5523,
+      "step": 6999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.900890327581694e-07,
+      "loss": 2.2903,
+      "step": 7000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.890716179310447e-07,
+      "loss": 2.3628,
+      "step": 7001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.880548981488261e-07,
+      "loss": 2.395,
+      "step": 7002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.870388735754358e-07,
+      "loss": 2.623,
+      "step": 7003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.86023544374686e-07,
+      "loss": 2.5501,
+      "step": 7004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.850089107102787e-07,
+      "loss": 2.6441,
+      "step": 7005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.839949727458e-07,
+      "loss": 2.4659,
+      "step": 7006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.829817306447267e-07,
+      "loss": 2.5754,
+      "step": 7007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.819691845704207e-07,
+      "loss": 2.4002,
+      "step": 7008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.809573346861364e-07,
+      "loss": 2.5553,
+      "step": 7009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.799461811550101e-07,
+      "loss": 2.5189,
+      "step": 7010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.789357241400701e-07,
+      "loss": 2.5037,
+      "step": 7011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.779259638042318e-07,
+      "loss": 2.2808,
+      "step": 7012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.769169003102954e-07,
+      "loss": 2.4919,
+      "step": 7013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.759085338209514e-07,
+      "loss": 2.6612,
+      "step": 7014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.749008644987787e-07,
+      "loss": 2.5852,
+      "step": 7015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.738938925062399e-07,
+      "loss": 2.4972,
+      "step": 7016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.728876180056904e-07,
+      "loss": 2.5964,
+      "step": 7017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.718820411593685e-07,
+      "loss": 2.4698,
+      "step": 7018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.708771621294019e-07,
+      "loss": 2.577,
+      "step": 7019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.698729810778065e-07,
+      "loss": 2.5412,
+      "step": 7020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.688694981664857e-07,
+      "loss": 2.425,
+      "step": 7021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.678667135572293e-07,
+      "loss": 2.5704,
+      "step": 7022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.668646274117135e-07,
+      "loss": 2.5388,
+      "step": 7023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.658632398915049e-07,
+      "loss": 2.6105,
+      "step": 7024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.648625511580547e-07,
+      "loss": 2.4506,
+      "step": 7025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.638625613727024e-07,
+      "loss": 2.4881,
+      "step": 7026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.62863270696677e-07,
+      "loss": 2.5272,
+      "step": 7027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.618646792910893e-07,
+      "loss": 2.4991,
+      "step": 7028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.608667873169444e-07,
+      "loss": 2.5807,
+      "step": 7029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.59869594935127e-07,
+      "loss": 2.5865,
+      "step": 7030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.588731023064177e-07,
+      "loss": 2.5757,
+      "step": 7031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.578773095914758e-07,
+      "loss": 2.5052,
+      "step": 7032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.568822169508532e-07,
+      "loss": 2.3653,
+      "step": 7033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.558878245449879e-07,
+      "loss": 2.4283,
+      "step": 7034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.548941325342024e-07,
+      "loss": 2.3236,
+      "step": 7035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.539011410787105e-07,
+      "loss": 2.4657,
+      "step": 7036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.529088503386089e-07,
+      "loss": 2.4938,
+      "step": 7037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.519172604738838e-07,
+      "loss": 2.4699,
+      "step": 7038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.509263716444092e-07,
+      "loss": 2.4351,
+      "step": 7039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.499361840099421e-07,
+      "loss": 2.4577,
+      "step": 7040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.489466977301312e-07,
+      "loss": 2.4382,
+      "step": 7041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.479579129645092e-07,
+      "loss": 2.5262,
+      "step": 7042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.469698298724974e-07,
+      "loss": 2.5562,
+      "step": 7043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.459824486134015e-07,
+      "loss": 2.3493,
+      "step": 7044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.449957693464159e-07,
+      "loss": 2.5105,
+      "step": 7045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.440097922306232e-07,
+      "loss": 2.5205,
+      "step": 7046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.430245174249889e-07,
+      "loss": 2.3766,
+      "step": 7047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.42039945088369e-07,
+      "loss": 2.5755,
+      "step": 7048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.410560753795037e-07,
+      "loss": 2.4185,
+      "step": 7049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.400729084570212e-07,
+      "loss": 2.6484,
+      "step": 7050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.390904444794372e-07,
+      "loss": 2.488,
+      "step": 7051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.381086836051498e-07,
+      "loss": 2.5949,
+      "step": 7052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.371276259924514e-07,
+      "loss": 2.3084,
+      "step": 7053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.361472717995126e-07,
+      "loss": 2.4573,
+      "step": 7054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.351676211843983e-07,
+      "loss": 2.5246,
+      "step": 7055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.341886743050523e-07,
+      "loss": 2.3508,
+      "step": 7056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.332104313193111e-07,
+      "loss": 2.5343,
+      "step": 7057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.322328923848958e-07,
+      "loss": 2.5926,
+      "step": 7058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.312560576594118e-07,
+      "loss": 2.4048,
+      "step": 7059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.302799273003546e-07,
+      "loss": 2.4688,
+      "step": 7060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.293045014651028e-07,
+      "loss": 2.7096,
+      "step": 7061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.283297803109234e-07,
+      "loss": 2.6595,
+      "step": 7062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.273557639949695e-07,
+      "loss": 2.3241,
+      "step": 7063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.263824526742807e-07,
+      "loss": 2.5094,
+      "step": 7064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.254098465057834e-07,
+      "loss": 2.5791,
+      "step": 7065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.244379456462873e-07,
+      "loss": 2.4879,
+      "step": 7066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.234667502524915e-07,
+      "loss": 2.5054,
+      "step": 7067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.22496260480982e-07,
+      "loss": 2.4852,
+      "step": 7068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.215264764882273e-07,
+      "loss": 2.5027,
+      "step": 7069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.205573984305857e-07,
+      "loss": 2.461,
+      "step": 7070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.195890264642995e-07,
+      "loss": 2.5505,
+      "step": 7071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.186213607454977e-07,
+      "loss": 2.6249,
+      "step": 7072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.176544014301978e-07,
+      "loss": 2.6036,
+      "step": 7073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.16688148674297e-07,
+      "loss": 2.3603,
+      "step": 7074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.157226026335883e-07,
+      "loss": 2.4026,
+      "step": 7075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.147577634637413e-07,
+      "loss": 2.5778,
+      "step": 7076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.13793631320318e-07,
+      "loss": 2.4887,
+      "step": 7077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.128302063587622e-07,
+      "loss": 2.4981,
+      "step": 7078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.118674887344062e-07,
+      "loss": 2.4814,
+      "step": 7079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.109054786024692e-07,
+      "loss": 2.5769,
+      "step": 7080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.099441761180519e-07,
+      "loss": 2.5252,
+      "step": 7081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.089835814361461e-07,
+      "loss": 2.4056,
+      "step": 7082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.080236947116259e-07,
+      "loss": 2.3448,
+      "step": 7083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.070645160992523e-07,
+      "loss": 2.5033,
+      "step": 7084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.061060457536727e-07,
+      "loss": 2.5494,
+      "step": 7085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.051482838294203e-07,
+      "loss": 2.5221,
+      "step": 7086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.041912304809139e-07,
+      "loss": 2.7329,
+      "step": 7087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.03234885862457e-07,
+      "loss": 2.5862,
+      "step": 7088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.022792501282409e-07,
+      "loss": 2.5092,
+      "step": 7089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.013243234323396e-07,
+      "loss": 2.6358,
+      "step": 7090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.003701059287159e-07,
+      "loss": 2.4988,
+      "step": 7091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.994165977712175e-07,
+      "loss": 2.7039,
+      "step": 7092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.98463799113575e-07,
+      "loss": 2.3833,
+      "step": 7093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.975117101094091e-07,
+      "loss": 2.5777,
+      "step": 7094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.965603309122215e-07,
+      "loss": 2.5442,
+      "step": 7095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.956096616754037e-07,
+      "loss": 2.5086,
+      "step": 7096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.946597025522294e-07,
+      "loss": 2.577,
+      "step": 7097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.937104536958598e-07,
+      "loss": 2.5645,
+      "step": 7098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.927619152593428e-07,
+      "loss": 2.4049,
+      "step": 7099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.918140873956063e-07,
+      "loss": 2.6013,
+      "step": 7100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.908669702574705e-07,
+      "loss": 2.6227,
+      "step": 7101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.899205639976347e-07,
+      "loss": 2.577,
+      "step": 7102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.889748687686886e-07,
+      "loss": 2.5048,
+      "step": 7103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.880298847231059e-07,
+      "loss": 2.6582,
+      "step": 7104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.870856120132434e-07,
+      "loss": 2.4305,
+      "step": 7105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.861420507913462e-07,
+      "loss": 2.6511,
+      "step": 7106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.851992012095414e-07,
+      "loss": 2.3853,
+      "step": 7107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.842570634198453e-07,
+      "loss": 2.6996,
+      "step": 7108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.833156375741561e-07,
+      "loss": 2.5249,
+      "step": 7109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.823749238242593e-07,
+      "loss": 2.4923,
+      "step": 7110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.814349223218263e-07,
+      "loss": 2.515,
+      "step": 7111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.804956332184092e-07,
+      "loss": 2.5192,
+      "step": 7112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.795570566654507e-07,
+      "loss": 2.4299,
+      "step": 7113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.786191928142748e-07,
+      "loss": 2.6531,
+      "step": 7114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.776820418160916e-07,
+      "loss": 2.5931,
+      "step": 7115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.767456038219987e-07,
+      "loss": 2.5592,
+      "step": 7116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.758098789829746e-07,
+      "loss": 2.3397,
+      "step": 7117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.748748674498855e-07,
+      "loss": 2.6578,
+      "step": 7118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.739405693734823e-07,
+      "loss": 2.4699,
+      "step": 7119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.730069849044007e-07,
+      "loss": 2.5224,
+      "step": 7120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.720741141931613e-07,
+      "loss": 2.6142,
+      "step": 7121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.711419573901683e-07,
+      "loss": 2.521,
+      "step": 7122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.702105146457143e-07,
+      "loss": 2.4117,
+      "step": 7123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.692797861099719e-07,
+      "loss": 2.4403,
+      "step": 7124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.683497719330022e-07,
+      "loss": 2.4956,
+      "step": 7125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.674204722647514e-07,
+      "loss": 2.5664,
+      "step": 7126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.664918872550468e-07,
+      "loss": 2.4348,
+      "step": 7127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.655640170536053e-07,
+      "loss": 2.6201,
+      "step": 7128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.64636861810024e-07,
+      "loss": 2.6298,
+      "step": 7129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.637104216737871e-07,
+      "loss": 2.3339,
+      "step": 7130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.627846967942646e-07,
+      "loss": 2.445,
+      "step": 7131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.618596873207083e-07,
+      "loss": 2.5413,
+      "step": 7132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.609353934022582e-07,
+      "loss": 2.544,
+      "step": 7133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.600118151879347e-07,
+      "loss": 2.5083,
+      "step": 7134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.590889528266469e-07,
+      "loss": 2.4732,
+      "step": 7135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.581668064671847e-07,
+      "loss": 2.489,
+      "step": 7136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.572453762582253e-07,
+      "loss": 2.3155,
+      "step": 7137
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.563246623483309e-07,
+      "loss": 2.4213,
+      "step": 7138
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.554046648859446e-07,
+      "loss": 2.3451,
+      "step": 7139
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.544853840193981e-07,
+      "loss": 2.4915,
+      "step": 7140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.53566819896903e-07,
+      "loss": 2.4581,
+      "step": 7141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.526489726665624e-07,
+      "loss": 2.5923,
+      "step": 7142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.517318424763562e-07,
+      "loss": 2.4803,
+      "step": 7143
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.50815429474153e-07,
+      "loss": 2.5425,
+      "step": 7144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.498997338077062e-07,
+      "loss": 2.5374,
+      "step": 7145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.489847556246491e-07,
+      "loss": 2.5923,
+      "step": 7146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.480704950725057e-07,
+      "loss": 2.504,
+      "step": 7147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.471569522986775e-07,
+      "loss": 2.2859,
+      "step": 7148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.46244127450456e-07,
+      "loss": 2.4843,
+      "step": 7149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.453320206750152e-07,
+      "loss": 2.2894,
+      "step": 7150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.444206321194101e-07,
+      "loss": 2.5239,
+      "step": 7151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.43509961930585e-07,
+      "loss": 2.4396,
+      "step": 7152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.426000102553647e-07,
+      "loss": 2.3656,
+      "step": 7153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.416907772404606e-07,
+      "loss": 2.4234,
+      "step": 7154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.40782263032466e-07,
+      "loss": 2.3908,
+      "step": 7155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.398744677778595e-07,
+      "loss": 2.3035,
+      "step": 7156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.389673916230048e-07,
+      "loss": 2.499,
+      "step": 7157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.380610347141463e-07,
+      "loss": 2.3504,
+      "step": 7158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.371553971974169e-07,
+      "loss": 2.4321,
+      "step": 7159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3625047921883e-07,
+      "loss": 2.5038,
+      "step": 7160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.353462809242838e-07,
+      "loss": 2.5523,
+      "step": 7161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.344428024595633e-07,
+      "loss": 2.5417,
+      "step": 7162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.335400439703309e-07,
+      "loss": 2.588,
+      "step": 7163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.326380056021419e-07,
+      "loss": 2.3856,
+      "step": 7164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.317366875004271e-07,
+      "loss": 2.6528,
+      "step": 7165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.30836089810507e-07,
+      "loss": 2.5672,
+      "step": 7166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.299362126775814e-07,
+      "loss": 2.4255,
+      "step": 7167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.290370562467378e-07,
+      "loss": 2.5477,
+      "step": 7168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.281386206629468e-07,
+      "loss": 2.5759,
+      "step": 7169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.272409060710598e-07,
+      "loss": 2.5212,
+      "step": 7170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.263439126158149e-07,
+      "loss": 2.4678,
+      "step": 7171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.254476404418341e-07,
+      "loss": 2.5561,
+      "step": 7172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.245520896936208e-07,
+      "loss": 2.513,
+      "step": 7173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.236572605155632e-07,
+      "loss": 2.4108,
+      "step": 7174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.227631530519339e-07,
+      "loss": 2.6567,
+      "step": 7175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.218697674468898e-07,
+      "loss": 2.6726,
+      "step": 7176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.209771038444677e-07,
+      "loss": 2.5279,
+      "step": 7177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.200851623885922e-07,
+      "loss": 2.5512,
+      "step": 7178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.191939432230702e-07,
+      "loss": 2.5547,
+      "step": 7179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.183034464915898e-07,
+      "loss": 2.4334,
+      "step": 7180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.174136723377266e-07,
+      "loss": 2.4235,
+      "step": 7181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.165246209049357e-07,
+      "loss": 2.3993,
+      "step": 7182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.156362923365587e-07,
+      "loss": 2.5758,
+      "step": 7183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.147486867758201e-07,
+      "loss": 2.3624,
+      "step": 7184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.13861804365825e-07,
+      "loss": 2.4513,
+      "step": 7185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.129756452495676e-07,
+      "loss": 2.4197,
+      "step": 7186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.120902095699193e-07,
+      "loss": 2.5422,
+      "step": 7187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.112054974696395e-07,
+      "loss": 2.476,
+      "step": 7188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.103215090913676e-07,
+      "loss": 2.4772,
+      "step": 7189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.094382445776281e-07,
+      "loss": 2.5008,
+      "step": 7190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0855570407083e-07,
+      "loss": 2.5613,
+      "step": 7191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.076738877132614e-07,
+      "loss": 2.6242,
+      "step": 7192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.067927956470992e-07,
+      "loss": 2.506,
+      "step": 7193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.059124280143985e-07,
+      "loss": 2.4647,
+      "step": 7194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.050327849571002e-07,
+      "loss": 2.3254,
+      "step": 7195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.041538666170282e-07,
+      "loss": 2.3558,
+      "step": 7196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.03275673135889e-07,
+      "loss": 2.4896,
+      "step": 7197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.023982046552739e-07,
+      "loss": 2.502,
+      "step": 7198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.015214613166535e-07,
+      "loss": 2.5853,
+      "step": 7199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.006454432613867e-07,
+      "loss": 2.5238,
+      "step": 7200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.997701506307095e-07,
+      "loss": 2.5531,
+      "step": 7201
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.988955835657456e-07,
+      "loss": 2.4281,
+      "step": 7202
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.980217422075018e-07,
+      "loss": 2.5282,
+      "step": 7203
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.971486266968634e-07,
+      "loss": 2.4562,
+      "step": 7204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.962762371746038e-07,
+      "loss": 2.5405,
+      "step": 7205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.954045737813762e-07,
+      "loss": 2.4427,
+      "step": 7206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.945336366577169e-07,
+      "loss": 2.5608,
+      "step": 7207
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.93663425944047e-07,
+      "loss": 2.5373,
+      "step": 7208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.927939417806693e-07,
+      "loss": 2.6382,
+      "step": 7209
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.919251843077699e-07,
+      "loss": 2.5288,
+      "step": 7210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.910571536654163e-07,
+      "loss": 2.2957,
+      "step": 7211
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.901898499935609e-07,
+      "loss": 2.5172,
+      "step": 7212
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.893232734320358e-07,
+      "loss": 2.4648,
+      "step": 7213
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.884574241205603e-07,
+      "loss": 2.653,
+      "step": 7214
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.875923021987333e-07,
+      "loss": 2.6679,
+      "step": 7215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.867279078060366e-07,
+      "loss": 2.5117,
+      "step": 7216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858642410818354e-07,
+      "loss": 2.5194,
+      "step": 7217
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.850013021653788e-07,
+      "loss": 2.5305,
+      "step": 7218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.841390911957939e-07,
+      "loss": 2.5615,
+      "step": 7219
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.832776083120983e-07,
+      "loss": 2.6335,
+      "step": 7220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.824168536531848e-07,
+      "loss": 2.5592,
+      "step": 7221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.815568273578325e-07,
+      "loss": 2.4856,
+      "step": 7222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.806975295647015e-07,
+      "loss": 2.2908,
+      "step": 7223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.798389604123355e-07,
+      "loss": 2.3066,
+      "step": 7224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.789811200391619e-07,
+      "loss": 2.2988,
+      "step": 7225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.781240085834865e-07,
+      "loss": 2.5438,
+      "step": 7226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.772676261835024e-07,
+      "loss": 2.5536,
+      "step": 7227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.764119729772809e-07,
+      "loss": 2.5195,
+      "step": 7228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7555704910277887e-07,
+      "loss": 2.568,
+      "step": 7229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7470285469783493e-07,
+      "loss": 2.4764,
+      "step": 7230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7384938990016836e-07,
+      "loss": 2.6047,
+      "step": 7231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.729966548473841e-07,
+      "loss": 2.5623,
+      "step": 7232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.721446496769655e-07,
+      "loss": 2.5745,
+      "step": 7233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.712933745262815e-07,
+      "loss": 2.4777,
+      "step": 7234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.704428295325808e-07,
+      "loss": 2.5396,
+      "step": 7235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.695930148329958e-07,
+      "loss": 2.5459,
+      "step": 7236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.687439305645419e-07,
+      "loss": 2.4172,
+      "step": 7237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.678955768641147e-07,
+      "loss": 2.5723,
+      "step": 7238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.670479538684941e-07,
+      "loss": 2.5921,
+      "step": 7239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.662010617143392e-07,
+      "loss": 2.3685,
+      "step": 7240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.653549005381952e-07,
+      "loss": 2.4473,
+      "step": 7241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.645094704764863e-07,
+      "loss": 2.471,
+      "step": 7242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6366477166552073e-07,
+      "loss": 2.4685,
+      "step": 7243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.628208042414889e-07,
+      "loss": 2.5505,
+      "step": 7244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6197756834046037e-07,
+      "loss": 2.4189,
+      "step": 7245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6113506409839135e-07,
+      "loss": 2.5042,
+      "step": 7246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.602932916511149e-07,
+      "loss": 2.6627,
+      "step": 7247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5945225113435024e-07,
+      "loss": 2.5217,
+      "step": 7248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.586119426836977e-07,
+      "loss": 2.3901,
+      "step": 7249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5777236643463787e-07,
+      "loss": 2.5806,
+      "step": 7250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5693352252253576e-07,
+      "loss": 2.6154,
+      "step": 7251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5609541108263377e-07,
+      "loss": 2.38,
+      "step": 7252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5525803225006437e-07,
+      "loss": 2.3915,
+      "step": 7253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5442138615983344e-07,
+      "loss": 2.6117,
+      "step": 7254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.535854729468331e-07,
+      "loss": 2.5032,
+      "step": 7255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5275029274583715e-07,
+      "loss": 2.6501,
+      "step": 7256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.519158456914996e-07,
+      "loss": 2.5988,
+      "step": 7257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5108213191835835e-07,
+      "loss": 2.5965,
+      "step": 7258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.502491515608298e-07,
+      "loss": 2.5541,
+      "step": 7259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.494169047532154e-07,
+      "loss": 2.5433,
+      "step": 7260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4858539162969894e-07,
+      "loss": 2.4509,
+      "step": 7261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.477546123243409e-07,
+      "loss": 2.6333,
+      "step": 7262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.469245669710881e-07,
+      "loss": 2.4801,
+      "step": 7263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4609525570376786e-07,
+      "loss": 2.485,
+      "step": 7264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.452666786560894e-07,
+      "loss": 2.518,
+      "step": 7265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4443883596164083e-07,
+      "loss": 2.7408,
+      "step": 7266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4361172775389596e-07,
+      "loss": 2.5248,
+      "step": 7267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4278535416620914e-07,
+      "loss": 2.5354,
+      "step": 7268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4195971533181223e-07,
+      "loss": 2.5358,
+      "step": 7269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.411348113838243e-07,
+      "loss": 2.6381,
+      "step": 7270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.403106424552439e-07,
+      "loss": 2.5136,
+      "step": 7271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.394872086789481e-07,
+      "loss": 2.3875,
+      "step": 7272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3866451018770074e-07,
+      "loss": 2.5735,
+      "step": 7273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3784254711414064e-07,
+      "loss": 2.559,
+      "step": 7274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.370213195907963e-07,
+      "loss": 2.6276,
+      "step": 7275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.362008277500701e-07,
+      "loss": 2.4778,
+      "step": 7276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.353810717242496e-07,
+      "loss": 2.5268,
+      "step": 7277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3456205164550345e-07,
+      "loss": 2.5978,
+      "step": 7278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3374376764588e-07,
+      "loss": 2.5567,
+      "step": 7279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.329262198573109e-07,
+      "loss": 2.4766,
+      "step": 7280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.321094084116073e-07,
+      "loss": 2.3987,
+      "step": 7281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3129333344046345e-07,
+      "loss": 2.3892,
+      "step": 7282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3047799507545397e-07,
+      "loss": 2.4345,
+      "step": 7283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2966339344803376e-07,
+      "loss": 2.5241,
+      "step": 7284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.288495286895405e-07,
+      "loss": 2.3935,
+      "step": 7285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.280364009311927e-07,
+      "loss": 2.4291,
+      "step": 7286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.272240103040903e-07,
+      "loss": 2.6228,
+      "step": 7287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2641235693921257e-07,
+      "loss": 2.5014,
+      "step": 7288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.256014409674214e-07,
+      "loss": 2.5744,
+      "step": 7289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.247912625194611e-07,
+      "loss": 2.4496,
+      "step": 7290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2398182172595393e-07,
+      "loss": 2.3382,
+      "step": 7291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.231731187174065e-07,
+      "loss": 2.5298,
+      "step": 7292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.223651536242029e-07,
+      "loss": 2.47,
+      "step": 7293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2155792657661066e-07,
+      "loss": 2.4589,
+      "step": 7294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2075143770478e-07,
+      "loss": 2.4062,
+      "step": 7295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.199456871387364e-07,
+      "loss": 2.4248,
+      "step": 7296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.191406750083937e-07,
+      "loss": 2.5215,
+      "step": 7297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.183364014435398e-07,
+      "loss": 2.4307,
+      "step": 7298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.175328665738487e-07,
+      "loss": 2.4109,
+      "step": 7299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.167300705288718e-07,
+      "loss": 2.4653,
+      "step": 7300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1592801343804334e-07,
+      "loss": 2.5132,
+      "step": 7301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1512669543067873e-07,
+      "loss": 2.4861,
+      "step": 7302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.143261166359713e-07,
+      "loss": 2.4211,
+      "step": 7303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1352627718299953e-07,
+      "loss": 2.4667,
+      "step": 7304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1272717720071795e-07,
+      "loss": 2.398,
+      "step": 7305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.119288168179658e-07,
+      "loss": 2.3611,
+      "step": 7306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.111311961634623e-07,
+      "loss": 2.7293,
+      "step": 7307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.10334315365804e-07,
+      "loss": 2.4768,
+      "step": 7308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.095381745534749e-07,
+      "loss": 2.4837,
+      "step": 7309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.087427738548322e-07,
+      "loss": 2.4832,
+      "step": 7310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.079481133981189e-07,
+      "loss": 2.5513,
+      "step": 7311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0715419331145646e-07,
+      "loss": 2.458,
+      "step": 7312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.063610137228474e-07,
+      "loss": 2.5604,
+      "step": 7313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0556857476017566e-07,
+      "loss": 2.6307,
+      "step": 7314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.047768765512039e-07,
+      "loss": 2.5499,
+      "step": 7315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0398591922357787e-07,
+      "loss": 2.26,
+      "step": 7316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.031957029048211e-07,
+      "loss": 2.4035,
+      "step": 7317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0240622772234005e-07,
+      "loss": 2.4886,
+      "step": 7318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.016174938034201e-07,
+      "loss": 2.6167,
+      "step": 7319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.008295012752278e-07,
+      "loss": 2.5344,
+      "step": 7320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.000422502648116e-07,
+      "loss": 2.6365,
+      "step": 7321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9925574089909603e-07,
+      "loss": 2.5785,
+      "step": 7322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.984699733048908e-07,
+      "loss": 2.4383,
+      "step": 7323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9768494760888455e-07,
+      "loss": 2.5733,
+      "step": 7324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.969006639376444e-07,
+      "loss": 2.5851,
+      "step": 7325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9611712241762024e-07,
+      "loss": 2.4319,
+      "step": 7326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.953343231751405e-07,
+      "loss": 2.5397,
+      "step": 7327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.945522663364154e-07,
+      "loss": 2.4683,
+      "step": 7328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.937709520275351e-07,
+      "loss": 2.5308,
+      "step": 7329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9299038037446837e-07,
+      "loss": 2.4036,
+      "step": 7330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.922105515030683e-07,
+      "loss": 2.5268,
+      "step": 7331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.914314655390633e-07,
+      "loss": 2.5696,
+      "step": 7332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9065312260806623e-07,
+      "loss": 2.3478,
+      "step": 7333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.898755228355661e-07,
+      "loss": 2.4529,
+      "step": 7334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8909866634693484e-07,
+      "loss": 2.3545,
+      "step": 7335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8832255326742563e-07,
+      "loss": 2.519,
+      "step": 7336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.875471837221678e-07,
+      "loss": 2.5441,
+      "step": 7337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8677255783617475e-07,
+      "loss": 2.4038,
+      "step": 7338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.85998675734337e-07,
+      "loss": 2.5866,
+      "step": 7339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.852255375414271e-07,
+      "loss": 2.4925,
+      "step": 7340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8445314338209693e-07,
+      "loss": 2.5016,
+      "step": 7341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8368149338087855e-07,
+      "loss": 2.6007,
+      "step": 7342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8291058766218526e-07,
+      "loss": 2.4856,
+      "step": 7343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.821404263503065e-07,
+      "loss": 2.4805,
+      "step": 7344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.813710095694168e-07,
+      "loss": 2.5351,
+      "step": 7345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8060233744356634e-07,
+      "loss": 2.5091,
+      "step": 7346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7983441009668774e-07,
+      "loss": 2.2845,
+      "step": 7347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7906722765259364e-07,
+      "loss": 2.5157,
+      "step": 7348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7830079023497393e-07,
+      "loss": 2.4045,
+      "step": 7349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.775350979674025e-07,
+      "loss": 2.5561,
+      "step": 7350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.767701509733285e-07,
+      "loss": 2.5471,
+      "step": 7351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.760059493760848e-07,
+      "loss": 2.5303,
+      "step": 7352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7524249329888183e-07,
+      "loss": 2.6072,
+      "step": 7353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.744797828648111e-07,
+      "loss": 2.4956,
+      "step": 7354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.737178181968437e-07,
+      "loss": 2.4836,
+      "step": 7355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7295659941782856e-07,
+      "loss": 2.2705,
+      "step": 7356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.721961266504975e-07,
+      "loss": 2.3644,
+      "step": 7357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7143640001745916e-07,
+      "loss": 2.5015,
+      "step": 7358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.706774196412044e-07,
+      "loss": 2.4285,
+      "step": 7359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.69919185644102e-07,
+      "loss": 2.6323,
+      "step": 7360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.691616981484009e-07,
+      "loss": 2.6013,
+      "step": 7361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.684049572762299e-07,
+      "loss": 2.4026,
+      "step": 7362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6764896314959606e-07,
+      "loss": 2.5333,
+      "step": 7363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6689371589039013e-07,
+      "loss": 2.4805,
+      "step": 7364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6613921562037633e-07,
+      "loss": 2.492,
+      "step": 7365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6538546246120353e-07,
+      "loss": 2.3869,
+      "step": 7366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6463245653439893e-07,
+      "loss": 2.5174,
+      "step": 7367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6388019796136654e-07,
+      "loss": 2.5276,
+      "step": 7368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6312868686339375e-07,
+      "loss": 2.3055,
+      "step": 7369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6237792336164424e-07,
+      "loss": 2.6116,
+      "step": 7370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.616279075771628e-07,
+      "loss": 2.504,
+      "step": 7371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.60878639630875e-07,
+      "loss": 2.4892,
+      "step": 7372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.601301196435819e-07,
+      "loss": 2.4271,
+      "step": 7373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.593823477359676e-07,
+      "loss": 2.5548,
+      "step": 7374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.586353240285945e-07,
+      "loss": 2.5003,
+      "step": 7375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5788904864190346e-07,
+      "loss": 2.4124,
+      "step": 7376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5714352169621656e-07,
+      "loss": 2.3785,
+      "step": 7377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.563987433117327e-07,
+      "loss": 2.5888,
+      "step": 7378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.556547136085331e-07,
+      "loss": 2.2479,
+      "step": 7379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5491143270657445e-07,
+      "loss": 2.3572,
+      "step": 7380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.541689007256965e-07,
+      "loss": 2.5052,
+      "step": 7381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.534271177856163e-07,
+      "loss": 2.5544,
+      "step": 7382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5268608400592987e-07,
+      "loss": 2.5853,
+      "step": 7383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519457995061143e-07,
+      "loss": 2.6683,
+      "step": 7384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51206264405522e-07,
+      "loss": 2.6758,
+      "step": 7385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.504674788233908e-07,
+      "loss": 2.4982,
+      "step": 7386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.497294428788317e-07,
+      "loss": 2.3822,
+      "step": 7387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.489921566908372e-07,
+      "loss": 2.667,
+      "step": 7388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4825562037828054e-07,
+      "loss": 2.5653,
+      "step": 7389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4751983405991007e-07,
+      "loss": 2.5551,
+      "step": 7390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4678479785435816e-07,
+      "loss": 2.5126,
+      "step": 7391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.460505118801316e-07,
+      "loss": 2.5487,
+      "step": 7392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4531697625561857e-07,
+      "loss": 2.4059,
+      "step": 7393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.445841910990877e-07,
+      "loss": 2.4765,
+      "step": 7394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.438521565286823e-07,
+      "loss": 2.642,
+      "step": 7395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4312087266242964e-07,
+      "loss": 2.4736,
+      "step": 7396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.42390339618231e-07,
+      "loss": 2.4759,
+      "step": 7397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.416605575138726e-07,
+      "loss": 2.5898,
+      "step": 7398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.409315264670132e-07,
+      "loss": 2.5504,
+      "step": 7399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4020324659519487e-07,
+      "loss": 2.5619,
+      "step": 7400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3947571801583756e-07,
+      "loss": 2.4192,
+      "step": 7401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.38748940846238e-07,
+      "loss": 2.5985,
+      "step": 7402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.380229152035758e-07,
+      "loss": 2.3326,
+      "step": 7403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3729764120490447e-07,
+      "loss": 2.4205,
+      "step": 7404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3657311896716004e-07,
+      "loss": 2.4164,
+      "step": 7405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.358493486071568e-07,
+      "loss": 2.6297,
+      "step": 7406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.351263302415864e-07,
+      "loss": 2.5179,
+      "step": 7407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3440406398702055e-07,
+      "loss": 2.4233,
+      "step": 7408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.336825499599089e-07,
+      "loss": 2.3842,
+      "step": 7409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.329617882765807e-07,
+      "loss": 2.3199,
+      "step": 7410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.322417790532423e-07,
+      "loss": 2.4857,
+      "step": 7411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.315225224059809e-07,
+      "loss": 2.6377,
+      "step": 7412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3080401845076106e-07,
+      "loss": 2.6067,
+      "step": 7413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3008626730342496e-07,
+      "loss": 2.4915,
+      "step": 7414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.293692690796968e-07,
+      "loss": 2.4119,
+      "step": 7415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.286530238951746e-07,
+      "loss": 2.4619,
+      "step": 7416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.279375318653388e-07,
+      "loss": 2.5698,
+      "step": 7417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2722279310554825e-07,
+      "loss": 2.4443,
+      "step": 7418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2650880773103686e-07,
+      "loss": 2.4182,
+      "step": 7419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.25795575856922e-07,
+      "loss": 2.4719,
+      "step": 7420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.250830975981956e-07,
+      "loss": 2.4857,
+      "step": 7421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.243713730697301e-07,
+      "loss": 2.4766,
+      "step": 7422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2366040238627493e-07,
+      "loss": 2.48,
+      "step": 7423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2295018566246007e-07,
+      "loss": 2.503,
+      "step": 7424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.222407230127922e-07,
+      "loss": 2.4238,
+      "step": 7425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.215320145516565e-07,
+      "loss": 2.5642,
+      "step": 7426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.208240603933177e-07,
+      "loss": 2.4594,
+      "step": 7427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2011686065191894e-07,
+      "loss": 2.4236,
+      "step": 7428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1941041544147897e-07,
+      "loss": 2.587,
+      "step": 7429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1870472487589845e-07,
+      "loss": 2.4274,
+      "step": 7430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1799978906895456e-07,
+      "loss": 2.5842,
+      "step": 7431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.172956081343043e-07,
+      "loss": 2.601,
+      "step": 7432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.165921821854795e-07,
+      "loss": 2.5265,
+      "step": 7433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.158895113358934e-07,
+      "loss": 2.4854,
+      "step": 7434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1518759569883814e-07,
+      "loss": 2.3677,
+      "step": 7435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1448643538748045e-07,
+      "loss": 2.5613,
+      "step": 7436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1378603051486877e-07,
+      "loss": 2.1597,
+      "step": 7437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.130863811939272e-07,
+      "loss": 2.4903,
+      "step": 7438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1238748753746e-07,
+      "loss": 2.4745,
+      "step": 7439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1168934965814966e-07,
+      "loss": 2.5619,
+      "step": 7440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1099196766855357e-07,
+      "loss": 2.5186,
+      "step": 7441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1029534168111217e-07,
+      "loss": 2.4908,
+      "step": 7442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.095994718081402e-07,
+      "loss": 2.6297,
+      "step": 7443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0890435816183226e-07,
+      "loss": 2.6244,
+      "step": 7444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.082100008542599e-07,
+      "loss": 2.6643,
+      "step": 7445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0751639999737346e-07,
+      "loss": 2.4812,
+      "step": 7446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0682355570300306e-07,
+      "loss": 2.636,
+      "step": 7447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.06131468082852e-07,
+      "loss": 2.4256,
+      "step": 7448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0544013724850664e-07,
+      "loss": 2.2894,
+      "step": 7449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0474956331142837e-07,
+      "loss": 2.6668,
+      "step": 7450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.040597463829581e-07,
+      "loss": 2.5251,
+      "step": 7451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.03370686574313e-07,
+      "loss": 2.5542,
+      "step": 7452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0268238399659035e-07,
+      "loss": 2.5387,
+      "step": 7453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0199483876076465e-07,
+      "loss": 2.404,
+      "step": 7454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.013080509776861e-07,
+      "loss": 2.5297,
+      "step": 7455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.006220207580862e-07,
+      "loss": 2.4893,
+      "step": 7456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.999367482125709e-07,
+      "loss": 2.5595,
+      "step": 7457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.992522334516268e-07,
+      "loss": 2.6839,
+      "step": 7458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.985684765856184e-07,
+      "loss": 2.4838,
+      "step": 7459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9788547772478416e-07,
+      "loss": 2.5541,
+      "step": 7460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9720323697924545e-07,
+      "loss": 2.5761,
+      "step": 7461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.965217544589971e-07,
+      "loss": 2.5328,
+      "step": 7462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9584103027391453e-07,
+      "loss": 2.5022,
+      "step": 7463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.951610645337499e-07,
+      "loss": 2.5085,
+      "step": 7464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.944818573481334e-07,
+      "loss": 2.5811,
+      "step": 7465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.938034088265723e-07,
+      "loss": 2.6842,
+      "step": 7466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.93125719078452e-07,
+      "loss": 2.511,
+      "step": 7467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9244878821303556e-07,
+      "loss": 2.5078,
+      "step": 7468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9177261633946294e-07,
+      "loss": 2.3229,
+      "step": 7469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.910972035667531e-07,
+      "loss": 2.6174,
+      "step": 7470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9042255000380213e-07,
+      "loss": 2.4438,
+      "step": 7471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.897486557593826e-07,
+      "loss": 2.4391,
+      "step": 7472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8907552094214643e-07,
+      "loss": 2.6034,
+      "step": 7473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.884031456606201e-07,
+      "loss": 2.4423,
+      "step": 7474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.877315300232131e-07,
+      "loss": 2.4917,
+      "step": 7475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.870606741382059e-07,
+      "loss": 2.5672,
+      "step": 7476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.863905781137616e-07,
+      "loss": 2.5695,
+      "step": 7477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8572124205791916e-07,
+      "loss": 2.567,
+      "step": 7478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8505266607859297e-07,
+      "loss": 2.58,
+      "step": 7479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.843848502835772e-07,
+      "loss": 2.6181,
+      "step": 7480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.837177947805442e-07,
+      "loss": 2.5806,
+      "step": 7481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8305149967704e-07,
+      "loss": 2.5277,
+      "step": 7482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8238596508049275e-07,
+      "loss": 2.6321,
+      "step": 7483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.817211910982037e-07,
+      "loss": 2.5709,
+      "step": 7484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.81057177837355e-07,
+      "loss": 2.5485,
+      "step": 7485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.803939254050031e-07,
+      "loss": 2.5134,
+      "step": 7486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.797314339080842e-07,
+      "loss": 2.6694,
+      "step": 7487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7906970345341177e-07,
+      "loss": 2.4697,
+      "step": 7488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7840873414767377e-07,
+      "loss": 2.3712,
+      "step": 7489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7774852609743887e-07,
+      "loss": 2.5892,
+      "step": 7490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.770890794091502e-07,
+      "loss": 2.5274,
+      "step": 7491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7643039418912996e-07,
+      "loss": 2.453,
+      "step": 7492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.757724705435777e-07,
+      "loss": 2.416,
+      "step": 7493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7511530857856794e-07,
+      "loss": 2.4695,
+      "step": 7494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.744589084000565e-07,
+      "loss": 2.4957,
+      "step": 7495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7380327011387096e-07,
+      "loss": 2.4016,
+      "step": 7496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7314839382572067e-07,
+      "loss": 2.5278,
+      "step": 7497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7249427964119e-07,
+      "loss": 2.6148,
+      "step": 7498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.718409276657408e-07,
+      "loss": 2.5561,
+      "step": 7499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.711883380047131e-07,
+      "loss": 2.4227,
+      "step": 7500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.705365107633212e-07,
+      "loss": 2.4384,
+      "step": 7501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.698854460466599e-07,
+      "loss": 2.4844,
+      "step": 7502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.692351439596985e-07,
+      "loss": 2.6736,
+      "step": 7503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6858560460728433e-07,
+      "loss": 2.4976,
+      "step": 7504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6793682809414245e-07,
+      "loss": 2.6499,
+      "step": 7505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.672888145248731e-07,
+      "loss": 2.4516,
+      "step": 7506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6664156400395545e-07,
+      "loss": 2.5231,
+      "step": 7507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6599507663574387e-07,
+      "loss": 2.6003,
+      "step": 7508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.653493525244721e-07,
+      "loss": 2.4628,
+      "step": 7509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6470439177424754e-07,
+      "loss": 2.5134,
+      "step": 7510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.640601944890575e-07,
+      "loss": 2.532,
+      "step": 7511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.634167607727656e-07,
+      "loss": 2.5282,
+      "step": 7512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6277409072910984e-07,
+      "loss": 2.5174,
+      "step": 7513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.621321844617092e-07,
+      "loss": 2.4503,
+      "step": 7514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6149104207405515e-07,
+      "loss": 2.5001,
+      "step": 7515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6085066366951907e-07,
+      "loss": 2.5748,
+      "step": 7516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6021104935134934e-07,
+      "loss": 2.4817,
+      "step": 7517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5957219922266794e-07,
+      "loss": 2.5297,
+      "step": 7518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5893411338647757e-07,
+      "loss": 2.4065,
+      "step": 7519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.582967919456547e-07,
+      "loss": 2.3784,
+      "step": 7520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5766023500295503e-07,
+      "loss": 2.6452,
+      "step": 7521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5702444266100913e-07,
+      "loss": 2.5534,
+      "step": 7522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5638941502232395e-07,
+      "loss": 2.6655,
+      "step": 7523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.557551521892859e-07,
+      "loss": 2.4299,
+      "step": 7524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5512165426415494e-07,
+      "loss": 2.5942,
+      "step": 7525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.544889213490698e-07,
+      "loss": 2.5404,
+      "step": 7526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.53856953546045e-07,
+      "loss": 2.5293,
+      "step": 7527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5322575095697077e-07,
+      "loss": 2.4715,
+      "step": 7528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.525953136836168e-07,
+      "loss": 2.6186,
+      "step": 7529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5196564182762506e-07,
+      "loss": 2.6049,
+      "step": 7530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5133673549051994e-07,
+      "loss": 2.5306,
+      "step": 7531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5070859477369645e-07,
+      "loss": 2.4617,
+      "step": 7532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.500812197784297e-07,
+      "loss": 2.4174,
+      "step": 7533
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.494546106058715e-07,
+      "loss": 2.4929,
+      "step": 7534
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4882876735704775e-07,
+      "loss": 2.5337,
+      "step": 7535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4820369013286326e-07,
+      "loss": 2.4312,
+      "step": 7536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.475793790340969e-07,
+      "loss": 2.4037,
+      "step": 7537
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.46955834161407e-07,
+      "loss": 2.6468,
+      "step": 7538
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.463330556153265e-07,
+      "loss": 2.4074,
+      "step": 7539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.457110434962645e-07,
+      "loss": 2.3817,
+      "step": 7540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.450897979045069e-07,
+      "loss": 2.5595,
+      "step": 7541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.444693189402175e-07,
+      "loss": 2.5788,
+      "step": 7542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4384960670343514e-07,
+      "loss": 2.538,
+      "step": 7543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.432306612940738e-07,
+      "loss": 2.4828,
+      "step": 7544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4261248281192584e-07,
+      "loss": 2.3047,
+      "step": 7545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4199507135666046e-07,
+      "loss": 2.5859,
+      "step": 7546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4137842702782023e-07,
+      "loss": 2.4836,
+      "step": 7547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.407625499248273e-07,
+      "loss": 2.6588,
+      "step": 7548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4014744014697723e-07,
+      "loss": 2.6479,
+      "step": 7549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.39533097793444e-07,
+      "loss": 2.4312,
+      "step": 7550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3891952296327846e-07,
+      "loss": 2.4688,
+      "step": 7551
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3830671575540308e-07,
+      "loss": 2.5428,
+      "step": 7552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3769467626862385e-07,
+      "loss": 2.5049,
+      "step": 7553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3708340460161628e-07,
+      "loss": 2.5956,
+      "step": 7554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3647290085293707e-07,
+      "loss": 2.5924,
+      "step": 7555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3586316512101416e-07,
+      "loss": 2.6818,
+      "step": 7556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3525419750415612e-07,
+      "loss": 2.5844,
+      "step": 7557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.346459981005461e-07,
+      "loss": 2.4585,
+      "step": 7558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3403856700824236e-07,
+      "loss": 2.5234,
+      "step": 7559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3343190432518092e-07,
+      "loss": 2.4789,
+      "step": 7560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3282601014917193e-07,
+      "loss": 2.4733,
+      "step": 7561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3222088457790448e-07,
+      "loss": 2.4107,
+      "step": 7562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3161652770894048e-07,
+      "loss": 2.3885,
+      "step": 7563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3101293963972094e-07,
+      "loss": 2.5325,
+      "step": 7564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3041012046756128e-07,
+      "loss": 2.615,
+      "step": 7565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2980807028965213e-07,
+      "loss": 2.5826,
+      "step": 7566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2920678920306304e-07,
+      "loss": 2.4776,
+      "step": 7567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.286062773047354e-07,
+      "loss": 2.556,
+      "step": 7568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2800653469149058e-07,
+      "loss": 2.4157,
+      "step": 7569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2740756146002409e-07,
+      "loss": 2.5058,
+      "step": 7570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2680935770690694e-07,
+      "loss": 2.5982,
+      "step": 7571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2621192352858702e-07,
+      "loss": 2.4008,
+      "step": 7572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2561525902138725e-07,
+      "loss": 2.3623,
+      "step": 7573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2501936428150794e-07,
+      "loss": 2.4943,
+      "step": 7574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2442423940502333e-07,
+      "loss": 2.2234,
+      "step": 7575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2382988448788557e-07,
+      "loss": 2.4438,
+      "step": 7576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.232362996259213e-07,
+      "loss": 2.5758,
+      "step": 7577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.226434849148329e-07,
+      "loss": 2.6327,
+      "step": 7578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2205144045019943e-07,
+      "loss": 2.6428,
+      "step": 7579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2146016632747624e-07,
+      "loss": 2.504,
+      "step": 7580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2086966264199206e-07,
+      "loss": 2.5781,
+      "step": 7581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.202799294889546e-07,
+      "loss": 2.4054,
+      "step": 7582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.196909669634434e-07,
+      "loss": 2.6063,
+      "step": 7583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.19102775160418e-07,
+      "loss": 2.5317,
+      "step": 7584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1851535417471093e-07,
+      "loss": 2.6552,
+      "step": 7585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1792870410103195e-07,
+      "loss": 2.58,
+      "step": 7586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1734282503396543e-07,
+      "loss": 2.5128,
+      "step": 7587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1675771706797132e-07,
+      "loss": 2.4231,
+      "step": 7588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1617338029738698e-07,
+      "loss": 2.6604,
+      "step": 7589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.155898148164226e-07,
+      "loss": 2.4688,
+      "step": 7590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.150070207191668e-07,
+      "loss": 2.5273,
+      "step": 7591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1442499809958273e-07,
+      "loss": 2.3634,
+      "step": 7592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1384374705150813e-07,
+      "loss": 2.5199,
+      "step": 7593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.13263267668658e-07,
+      "loss": 2.5234,
+      "step": 7594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1268356004462198e-07,
+      "loss": 2.5142,
+      "step": 7595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1210462427286528e-07,
+      "loss": 2.7263,
+      "step": 7596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1152646044672876e-07,
+      "loss": 2.6416,
+      "step": 7597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.109490686594301e-07,
+      "loss": 2.4821,
+      "step": 7598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.103724490040604e-07,
+      "loss": 2.5278,
+      "step": 7599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.097966015735875e-07,
+      "loss": 2.4321,
+      "step": 7600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0922152646085492e-07,
+      "loss": 2.4793,
+      "step": 7601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0864722375858015e-07,
+      "loss": 2.5183,
+      "step": 7602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0807369355935746e-07,
+      "loss": 2.4297,
+      "step": 7603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0750093595565735e-07,
+      "loss": 2.4949,
+      "step": 7604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.069289510398237e-07,
+      "loss": 2.5981,
+      "step": 7605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0635773890407718e-07,
+      "loss": 2.5156,
+      "step": 7606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0578729964051357e-07,
+      "loss": 2.4259,
+      "step": 7607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0521763334110324e-07,
+      "loss": 2.537,
+      "step": 7608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0464874009769377e-07,
+      "loss": 2.4241,
+      "step": 7609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0408062000200623e-07,
+      "loss": 2.3109,
+      "step": 7610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0351327314563907e-07,
+      "loss": 2.4229,
+      "step": 7611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0294669962006352e-07,
+      "loss": 2.5782,
+      "step": 7612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0238089951662765e-07,
+      "loss": 2.2915,
+      "step": 7613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0181587292655512e-07,
+      "loss": 2.5249,
+      "step": 7614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0125161994094312e-07,
+      "loss": 2.5208,
+      "step": 7615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.006881406507677e-07,
+      "loss": 2.6225,
+      "step": 7616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0012543514687566e-07,
+      "loss": 2.5495,
+      "step": 7617
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.995635035199922e-07,
+      "loss": 2.5515,
+      "step": 7618
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9900234586071542e-07,
+      "loss": 2.5095,
+      "step": 7619
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.984419622595224e-07,
+      "loss": 2.6258,
+      "step": 7620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9788235280676138e-07,
+      "loss": 2.3513,
+      "step": 7621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9732351759265743e-07,
+      "loss": 2.5115,
+      "step": 7622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9676545670731128e-07,
+      "loss": 2.4361,
+      "step": 7623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9620817024069816e-07,
+      "loss": 2.5774,
+      "step": 7624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9565165828266896e-07,
+      "loss": 2.6376,
+      "step": 7625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9509592092294805e-07,
+      "loss": 2.5592,
+      "step": 7626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9454095825113706e-07,
+      "loss": 2.3954,
+      "step": 7627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9398677035671222e-07,
+      "loss": 2.5003,
+      "step": 7628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9343335732902313e-07,
+      "loss": 2.5519,
+      "step": 7629
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9288071925729735e-07,
+      "loss": 2.5303,
+      "step": 7630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.923288562306347e-07,
+      "loss": 2.4354,
+      "step": 7631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9177776833801233e-07,
+      "loss": 2.4906,
+      "step": 7632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.912274556682814e-07,
+      "loss": 2.5275,
+      "step": 7633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9067791831016647e-07,
+      "loss": 2.4508,
+      "step": 7634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9012915635227058e-07,
+      "loss": 2.4331,
+      "step": 7635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8958116988306852e-07,
+      "loss": 2.5353,
+      "step": 7636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8903395899091182e-07,
+      "loss": 2.7129,
+      "step": 7637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8848752376402658e-07,
+      "loss": 2.5556,
+      "step": 7638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8794186429051397e-07,
+      "loss": 2.4212,
+      "step": 7639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8739698065834977e-07,
+      "loss": 2.3587,
+      "step": 7640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8685287295538368e-07,
+      "loss": 2.5139,
+      "step": 7641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.863095412693433e-07,
+      "loss": 2.5936,
+      "step": 7642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8576698568782802e-07,
+      "loss": 2.5167,
+      "step": 7643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8522520629831396e-07,
+      "loss": 2.4825,
+      "step": 7644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8468420318815128e-07,
+      "loss": 2.4161,
+      "step": 7645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.841439764445646e-07,
+      "loss": 2.4846,
+      "step": 7646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8360452615465486e-07,
+      "loss": 2.4969,
+      "step": 7647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8306585240539576e-07,
+      "loss": 2.4819,
+      "step": 7648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8252795528363732e-07,
+      "loss": 2.4894,
+      "step": 7649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8199083487610514e-07,
+      "loss": 2.6562,
+      "step": 7650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8145449126939662e-07,
+      "loss": 2.5424,
+      "step": 7651
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8091892454998595e-07,
+      "loss": 2.4792,
+      "step": 7652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.803841348042229e-07,
+      "loss": 2.4824,
+      "step": 7653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.798501221183302e-07,
+      "loss": 2.5073,
+      "step": 7654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7931688657840563e-07,
+      "loss": 2.6783,
+      "step": 7655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7878442827042264e-07,
+      "loss": 2.5077,
+      "step": 7656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7825274728022868e-07,
+      "loss": 2.5522,
+      "step": 7657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7772184369354517e-07,
+      "loss": 2.4323,
+      "step": 7658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.771917175959703e-07,
+      "loss": 2.2891,
+      "step": 7659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7666236907297407e-07,
+      "loss": 2.513,
+      "step": 7660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7613379820990317e-07,
+      "loss": 2.3849,
+      "step": 7661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.756060050919789e-07,
+      "loss": 2.4031,
+      "step": 7662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7507898980429539e-07,
+      "loss": 2.3924,
+      "step": 7663
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7455275243182413e-07,
+      "loss": 2.577,
+      "step": 7664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7402729305940835e-07,
+      "loss": 2.5801,
+      "step": 7665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7350261177176807e-07,
+      "loss": 2.5976,
+      "step": 7666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7297870865349564e-07,
+      "loss": 2.6736,
+      "step": 7667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7245558378906012e-07,
+      "loss": 2.3839,
+      "step": 7668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.719332372628052e-07,
+      "loss": 2.5803,
+      "step": 7669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.714116691589457e-07,
+      "loss": 2.4605,
+      "step": 7670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7089087956157546e-07,
+      "loss": 2.5767,
+      "step": 7671
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7037086855465902e-07,
+      "loss": 2.6585,
+      "step": 7672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.698516362220376e-07,
+      "loss": 2.3702,
+      "step": 7673
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6933318264742648e-07,
+      "loss": 2.4733,
+      "step": 7674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6881550791441548e-07,
+      "loss": 2.567,
+      "step": 7675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.682986121064689e-07,
+      "loss": 2.4837,
+      "step": 7676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6778249530692348e-07,
+      "loss": 2.4458,
+      "step": 7677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.672671575989937e-07,
+      "loss": 2.526,
+      "step": 7678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.667525990657659e-07,
+      "loss": 2.4089,
+      "step": 7679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.662388197902015e-07,
+      "loss": 2.4471,
+      "step": 7680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6572581985513703e-07,
+      "loss": 2.4027,
+      "step": 7681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6521359934328185e-07,
+      "loss": 2.4196,
+      "step": 7682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6470215833722103e-07,
+      "loss": 2.4775,
+      "step": 7683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.641914969194147e-07,
+      "loss": 2.5306,
+      "step": 7684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6368161517219317e-07,
+      "loss": 2.4456,
+      "step": 7685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6317251317776673e-07,
+      "loss": 2.5982,
+      "step": 7686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626641910182164e-07,
+      "loss": 2.5303,
+      "step": 7687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6215664877549774e-07,
+      "loss": 2.4269,
+      "step": 7688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6164988653144086e-07,
+      "loss": 2.4469,
+      "step": 7689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6114390436775095e-07,
+      "loss": 2.432,
+      "step": 7690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6063870236600777e-07,
+      "loss": 2.4317,
+      "step": 7691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6013428060766168e-07,
+      "loss": 2.5498,
+      "step": 7692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5963063917404263e-07,
+      "loss": 2.3484,
+      "step": 7693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5912777814635018e-07,
+      "loss": 2.389,
+      "step": 7694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5862569760565994e-07,
+      "loss": 2.4911,
+      "step": 7695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.581243976329233e-07,
+      "loss": 2.4631,
+      "step": 7696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5762387830896174e-07,
+      "loss": 2.5091,
+      "step": 7697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.571241397144757e-07,
+      "loss": 2.4213,
+      "step": 7698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5662518193003628e-07,
+      "loss": 2.4258,
+      "step": 7699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.561270050360897e-07,
+      "loss": 2.6934,
+      "step": 7700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5562960911295556e-07,
+      "loss": 2.5769,
+      "step": 7701
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5513299424082973e-07,
+      "loss": 2.464,
+      "step": 7702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.546371604997804e-07,
+      "loss": 2.5145,
+      "step": 7703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.541421079697486e-07,
+      "loss": 2.5964,
+      "step": 7704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5364783673055328e-07,
+      "loss": 2.5034,
+      "step": 7705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5315434686188345e-07,
+      "loss": 2.4922,
+      "step": 7706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5266163844330439e-07,
+      "loss": 2.5152,
+      "step": 7707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5216971155425474e-07,
+      "loss": 2.562,
+      "step": 7708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5167856627404664e-07,
+      "loss": 2.4872,
+      "step": 7709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.511882026818684e-07,
+      "loss": 2.5838,
+      "step": 7710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5069862085677845e-07,
+      "loss": 2.5677,
+      "step": 7711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5020982087771362e-07,
+      "loss": 2.5326,
+      "step": 7712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.497218028234798e-07,
+      "loss": 2.3291,
+      "step": 7713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4923456677276183e-07,
+      "loss": 2.5942,
+      "step": 7714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4874811280411518e-07,
+      "loss": 2.3984,
+      "step": 7715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4826244099596986e-07,
+      "loss": 2.3906,
+      "step": 7716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4777755142663054e-07,
+      "loss": 2.5648,
+      "step": 7717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.472934441742746e-07,
+      "loss": 2.5364,
+      "step": 7718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4681011931695466e-07,
+      "loss": 2.6065,
+      "step": 7719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4632757693259615e-07,
+      "loss": 2.4225,
+      "step": 7720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4584581709899848e-07,
+      "loss": 2.3998,
+      "step": 7721
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4536483989383676e-07,
+      "loss": 2.5006,
+      "step": 7722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4488464539465563e-07,
+      "loss": 2.593,
+      "step": 7723
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.444052336788787e-07,
+      "loss": 2.5751,
+      "step": 7724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.439266048237986e-07,
+      "loss": 2.6022,
+      "step": 7725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4344875890658526e-07,
+      "loss": 2.5078,
+      "step": 7726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4297169600428096e-07,
+      "loss": 2.48,
+      "step": 7727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.424954161938019e-07,
+      "loss": 2.456,
+      "step": 7728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.420199195519384e-07,
+      "loss": 2.4964,
+      "step": 7729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4154520615535185e-07,
+      "loss": 2.4319,
+      "step": 7730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4107127608058325e-07,
+      "loss": 2.5435,
+      "step": 7731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4059812940404093e-07,
+      "loss": 2.4659,
+      "step": 7732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4012576620201057e-07,
+      "loss": 2.584,
+      "step": 7733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3965418655065122e-07,
+      "loss": 2.5757,
+      "step": 7734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3918339052599372e-07,
+      "loss": 2.6635,
+      "step": 7735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3871337820394404e-07,
+      "loss": 2.3891,
+      "step": 7736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3824414966028321e-07,
+      "loss": 2.2799,
+      "step": 7737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.377757049706624e-07,
+      "loss": 2.5217,
+      "step": 7738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3730804421061006e-07,
+      "loss": 2.4137,
+      "step": 7739
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3684116745552423e-07,
+      "loss": 2.4882,
+      "step": 7740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3637507478068023e-07,
+      "loss": 2.4651,
+      "step": 7741
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.359097662612252e-07,
+      "loss": 2.4582,
+      "step": 7742
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3544524197218078e-07,
+      "loss": 2.5012,
+      "step": 7743
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3498150198844094e-07,
+      "loss": 2.6082,
+      "step": 7744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3451854638477367e-07,
+      "loss": 2.5283,
+      "step": 7745
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.340563752358215e-07,
+      "loss": 2.4916,
+      "step": 7746
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3359498861609866e-07,
+      "loss": 2.4254,
+      "step": 7747
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.33134386599994e-07,
+      "loss": 2.4628,
+      "step": 7748
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3267456926177091e-07,
+      "loss": 2.5108,
+      "step": 7749
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3221553667556342e-07,
+      "loss": 2.4836,
+      "step": 7750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3175728891538175e-07,
+      "loss": 2.5129,
+      "step": 7751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3129982605510738e-07,
+      "loss": 2.5184,
+      "step": 7752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3084314816849853e-07,
+      "loss": 2.5712,
+      "step": 7753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.303872553291824e-07,
+      "loss": 2.3172,
+      "step": 7754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2993214761066354e-07,
+      "loss": 2.4905,
+      "step": 7755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2947782508631823e-07,
+      "loss": 2.4209,
+      "step": 7756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2902428782939514e-07,
+      "loss": 2.574,
+      "step": 7757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2857153591301907e-07,
+      "loss": 2.5354,
+      "step": 7758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.28119569410185e-07,
+      "loss": 2.3728,
+      "step": 7759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2766838839376294e-07,
+      "loss": 2.5698,
+      "step": 7760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2721799293649807e-07,
+      "loss": 2.4668,
+      "step": 7761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.267683831110045e-07,
+      "loss": 2.4992,
+      "step": 7762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2631955898977366e-07,
+      "loss": 2.3982,
+      "step": 7763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2587152064516828e-07,
+      "loss": 2.4879,
+      "step": 7764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.254242681494261e-07,
+      "loss": 2.5005,
+      "step": 7765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.24977801574655e-07,
+      "loss": 2.5777,
+      "step": 7766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2453212099283962e-07,
+      "loss": 2.487,
+      "step": 7767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2408722647583692e-07,
+      "loss": 2.451,
+      "step": 7768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2364311809537456e-07,
+      "loss": 2.2638,
+      "step": 7769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2319979592305798e-07,
+      "loss": 2.3659,
+      "step": 7770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2275726003036114e-07,
+      "loss": 2.5344,
+      "step": 7771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2231551048863421e-07,
+      "loss": 2.5267,
+      "step": 7772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2187454736910077e-07,
+      "loss": 2.3894,
+      "step": 7773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2143437074285503e-07,
+      "loss": 2.6083,
+      "step": 7774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2099498068086802e-07,
+      "loss": 2.4971,
+      "step": 7775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.205563772539803e-07,
+      "loss": 2.4029,
+      "step": 7776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.201185605329086e-07,
+      "loss": 2.4083,
+      "step": 7777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.196815305882404e-07,
+      "loss": 2.5014,
+      "step": 7778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1924528749043818e-07,
+      "loss": 2.4891,
+      "step": 7779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1880983130983626e-07,
+      "loss": 2.3971,
+      "step": 7780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.183751621166429e-07,
+      "loss": 2.2928,
+      "step": 7781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1794127998093929e-07,
+      "loss": 2.5015,
+      "step": 7782
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1750818497267946e-07,
+      "loss": 2.3475,
+      "step": 7783
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1707587716169033e-07,
+      "loss": 2.505,
+      "step": 7784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1664435661767281e-07,
+      "loss": 2.431,
+      "step": 7785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1621362341020014e-07,
+      "loss": 2.5514,
+      "step": 7786
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1578367760871956e-07,
+      "loss": 2.4214,
+      "step": 7787
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1535451928254948e-07,
+      "loss": 2.3808,
+      "step": 7788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.149261485008829e-07,
+      "loss": 2.5442,
+      "step": 7789
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1449856533278624e-07,
+      "loss": 2.5828,
+      "step": 7790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1407176984719658e-07,
+      "loss": 2.435,
+      "step": 7791
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1364576211292722e-07,
+      "loss": 2.3238,
+      "step": 7792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1322054219866096e-07,
+      "loss": 2.4528,
+      "step": 7793
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1279611017295689e-07,
+      "loss": 2.562,
+      "step": 7794
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1237246610424579e-07,
+      "loss": 2.4713,
+      "step": 7795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1194961006082972e-07,
+      "loss": 2.4013,
+      "step": 7796
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.115275421108858e-07,
+      "loss": 2.3863,
+      "step": 7797
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1110626232246347e-07,
+      "loss": 2.5661,
+      "step": 7798
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1068577076348619e-07,
+      "loss": 2.6861,
+      "step": 7799
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.102660675017475e-07,
+      "loss": 2.4688,
+      "step": 7800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0984715260491663e-07,
+      "loss": 2.3044,
+      "step": 7801
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0942902614053453e-07,
+      "loss": 2.5173,
+      "step": 7802
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0901168817601448e-07,
+      "loss": 2.6059,
+      "step": 7803
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0859513877864381e-07,
+      "loss": 2.4244,
+      "step": 7804
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.081793780155821e-07,
+      "loss": 2.512,
+      "step": 7805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0776440595386184e-07,
+      "loss": 2.4523,
+      "step": 7806
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0735022266038898e-07,
+      "loss": 2.5974,
+      "step": 7807
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0693682820194062e-07,
+      "loss": 2.4321,
+      "step": 7808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0652422264516848e-07,
+      "loss": 2.5084,
+      "step": 7809
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0611240605659601e-07,
+      "loss": 2.4907,
+      "step": 7810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0570137850262064e-07,
+      "loss": 2.4921,
+      "step": 7811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0529114004951047e-07,
+      "loss": 2.7171,
+      "step": 7812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.048816907634087e-07,
+      "loss": 2.4455,
+      "step": 7813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0447303071032977e-07,
+      "loss": 2.5449,
+      "step": 7814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0406515995616152e-07,
+      "loss": 2.5396,
+      "step": 7815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0365807856666465e-07,
+      "loss": 2.3989,
+      "step": 7816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0325178660747171e-07,
+      "loss": 2.5713,
+      "step": 7817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.028462841440886e-07,
+      "loss": 2.541,
+      "step": 7818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.024415712418947e-07,
+      "loss": 2.5355,
+      "step": 7819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0203764796614057e-07,
+      "loss": 2.3239,
+      "step": 7820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0163451438195083e-07,
+      "loss": 2.4575,
+      "step": 7821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0123217055432121e-07,
+      "loss": 2.493,
+      "step": 7822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0083061654812154e-07,
+      "loss": 2.617,
+      "step": 7823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0042985242809389e-07,
+      "loss": 2.6842,
+      "step": 7824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0002987825885324e-07,
+      "loss": 2.339,
+      "step": 7825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.963069410488635e-08,
+      "loss": 2.532,
+      "step": 7826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.923230003055284e-08,
+      "loss": 2.4053,
+      "step": 7827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.883469610008578e-08,
+      "loss": 2.5451,
+      "step": 7828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.843788237759e-08,
+      "loss": 2.4263,
+      "step": 7829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.804185892704376e-08,
+      "loss": 2.5106,
+      "step": 7830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.7646625812296e-08,
+      "loss": 2.4816,
+      "step": 7831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.725218309707129e-08,
+      "loss": 2.6836,
+      "step": 7832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.685853084496433e-08,
+      "loss": 2.4785,
+      "step": 7833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.646566911944266e-08,
+      "loss": 2.4941,
+      "step": 7834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.607359798384785e-08,
+      "loss": 2.4946,
+      "step": 7835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.568231750139212e-08,
+      "loss": 2.6624,
+      "step": 7836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.52918277351611e-08,
+      "loss": 2.4036,
+      "step": 7837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.490212874811388e-08,
+      "loss": 2.4882,
+      "step": 7838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.451322060307966e-08,
+      "loss": 2.639,
+      "step": 7839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.412510336276214e-08,
+      "loss": 2.5369,
+      "step": 7840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.373777708973741e-08,
+      "loss": 2.6328,
+      "step": 7841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.335124184645272e-08,
+      "loss": 2.4092,
+      "step": 7842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.296549769522933e-08,
+      "loss": 2.5696,
+      "step": 7843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.258054469825972e-08,
+      "loss": 2.5333,
+      "step": 7844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.219638291761035e-08,
+      "loss": 2.5723,
+      "step": 7845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.181301241521723e-08,
+      "loss": 2.3422,
+      "step": 7846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.143043325289202e-08,
+      "loss": 2.4904,
+      "step": 7847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.104864549231706e-08,
+      "loss": 2.5112,
+      "step": 7848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.066764919504756e-08,
+      "loss": 2.5179,
+      "step": 7849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.028744442251103e-08,
+      "loss": 2.4185,
+      "step": 7850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.990803123600677e-08,
+      "loss": 2.6487,
+      "step": 7851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.952940969670809e-08,
+      "loss": 2.6766,
+      "step": 7852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.915157986565837e-08,
+      "loss": 2.4392,
+      "step": 7853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.877454180377553e-08,
+      "loss": 2.4316,
+      "step": 7854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.839829557184876e-08,
+      "loss": 2.4406,
+      "step": 7855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.802284123053894e-08,
+      "loss": 2.5106,
+      "step": 7856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.764817884038157e-08,
+      "loss": 2.4984,
+      "step": 7857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.727430846178109e-08,
+      "loss": 2.5036,
+      "step": 7858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.690123015501705e-08,
+      "loss": 2.5099,
+      "step": 7859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.652894398024137e-08,
+      "loss": 2.5362,
+      "step": 7860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.615744999747489e-08,
+      "loss": 2.6241,
+      "step": 7861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.578674826661527e-08,
+      "loss": 2.3797,
+      "step": 7862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.541683884742802e-08,
+      "loss": 2.5823,
+      "step": 7863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.504772179955546e-08,
+      "loss": 2.1053,
+      "step": 7864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.467939718250828e-08,
+      "loss": 2.5822,
+      "step": 7865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.431186505567179e-08,
+      "loss": 2.4693,
+      "step": 7866
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.394512547830191e-08,
+      "loss": 2.251,
+      "step": 7867
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.357917850952802e-08,
+      "loss": 2.6286,
+      "step": 7868
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.321402420835123e-08,
+      "loss": 2.5579,
+      "step": 7869
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.284966263364447e-08,
+      "loss": 2.4881,
+      "step": 7870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.248609384415352e-08,
+      "loss": 2.4689,
+      "step": 7871
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.212331789849648e-08,
+      "loss": 2.6223,
+      "step": 7872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.176133485516213e-08,
+      "loss": 2.5208,
+      "step": 7873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.140014477251379e-08,
+      "loss": 2.5481,
+      "step": 7874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.10397477087843e-08,
+      "loss": 2.4135,
+      "step": 7875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.06801437220811e-08,
+      "loss": 2.49,
+      "step": 7876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.032133287038224e-08,
+      "loss": 2.3021,
+      "step": 7877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.99633152115381e-08,
+      "loss": 2.6466,
+      "step": 7878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.960609080327141e-08,
+      "loss": 2.5567,
+      "step": 7879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.924965970317722e-08,
+      "loss": 2.4246,
+      "step": 7880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.889402196872231e-08,
+      "loss": 2.7354,
+      "step": 7881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.853917765724528e-08,
+      "loss": 2.7613,
+      "step": 7882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.818512682595702e-08,
+      "loss": 2.548,
+      "step": 7883
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.783186953194189e-08,
+      "loss": 2.4719,
+      "step": 7884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.74794058321543e-08,
+      "loss": 2.4591,
+      "step": 7885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.712773578342047e-08,
+      "loss": 2.5747,
+      "step": 7886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.677685944244118e-08,
+      "loss": 2.5344,
+      "step": 7887
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.642677686578726e-08,
+      "loss": 2.4681,
+      "step": 7888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.607748810990246e-08,
+      "loss": 2.4397,
+      "step": 7889
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.572899323110116e-08,
+      "loss": 2.3186,
+      "step": 7890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.538129228557123e-08,
+      "loss": 2.5402,
+      "step": 7891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.503438532937169e-08,
+      "loss": 2.6091,
+      "step": 7892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.46882724184339e-08,
+      "loss": 2.5396,
+      "step": 7893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.434295360856214e-08,
+      "loss": 2.4622,
+      "step": 7894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.399842895543019e-08,
+      "loss": 2.6297,
+      "step": 7895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.36546985145864e-08,
+      "loss": 2.7383,
+      "step": 7896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.331176234144921e-08,
+      "loss": 2.4209,
+      "step": 7897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.296962049131051e-08,
+      "loss": 2.5079,
+      "step": 7898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.262827301933229e-08,
+      "loss": 2.4959,
+      "step": 7899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.228771998054995e-08,
+      "loss": 2.5505,
+      "step": 7900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.194796142987127e-08,
+      "loss": 2.465,
+      "step": 7901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.160899742207406e-08,
+      "loss": 2.5698,
+      "step": 7902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.127082801180907e-08,
+      "loss": 2.4795,
+      "step": 7903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.093345325359935e-08,
+      "loss": 2.4868,
+      "step": 7904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.059687320183862e-08,
+      "loss": 2.4002,
+      "step": 7905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.0261087910794e-08,
+      "loss": 2.6033,
+      "step": 7906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.992609743460332e-08,
+      "loss": 2.5269,
+      "step": 7907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.959190182727616e-08,
+      "loss": 2.5741,
+      "step": 7908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.925850114269494e-08,
+      "loss": 2.505,
+      "step": 7909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.892589543461392e-08,
+      "loss": 2.5686,
+      "step": 7910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.859408475665796e-08,
+      "loss": 2.5264,
+      "step": 7911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.826306916232428e-08,
+      "loss": 2.6221,
+      "step": 7912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.79328487049824e-08,
+      "loss": 2.4873,
+      "step": 7913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.760342343787307e-08,
+      "loss": 2.3416,
+      "step": 7914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.727479341410937e-08,
+      "loss": 2.4067,
+      "step": 7915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.694695868667556e-08,
+      "loss": 2.641,
+      "step": 7916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.661991930842826e-08,
+      "loss": 2.4694,
+      "step": 7917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.629367533209529e-08,
+      "loss": 2.5393,
+      "step": 7918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.596822681027626e-08,
+      "loss": 2.5646,
+      "step": 7919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.564357379544361e-08,
+      "loss": 2.4901,
+      "step": 7920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.531971633993994e-08,
+      "loss": 2.4935,
+      "step": 7921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.49966544959807e-08,
+      "loss": 2.37,
+      "step": 7922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.467438831565254e-08,
+      "loss": 2.5289,
+      "step": 7923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.43529178509139e-08,
+      "loss": 2.4308,
+      "step": 7924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.403224315359558e-08,
+      "loss": 2.5588,
+      "step": 7925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.371236427539895e-08,
+      "loss": 2.4023,
+      "step": 7926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.339328126789778e-08,
+      "loss": 2.4589,
+      "step": 7927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.307499418253705e-08,
+      "loss": 2.542,
+      "step": 7928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.2757503070634e-08,
+      "loss": 2.5726,
+      "step": 7929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.244080798337715e-08,
+      "loss": 2.5009,
+      "step": 7930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.212490897182732e-08,
+      "loss": 2.3909,
+      "step": 7931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.180980608691656e-08,
+      "loss": 2.5154,
+      "step": 7932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.149549937944698e-08,
+      "loss": 2.4099,
+      "step": 7933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.118198890009586e-08,
+      "loss": 2.5711,
+      "step": 7934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.086927469940829e-08,
+      "loss": 2.4319,
+      "step": 7935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.055735682780395e-08,
+      "loss": 2.5646,
+      "step": 7936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.024623533557206e-08,
+      "loss": 2.3425,
+      "step": 7937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.993591027287471e-08,
+      "loss": 2.4871,
+      "step": 7938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.962638168974522e-08,
+      "loss": 2.4894,
+      "step": 7939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9317649636088656e-08,
+      "loss": 2.4981,
+      "step": 7940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9009714161680774e-08,
+      "loss": 2.6045,
+      "step": 7941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.870257531617019e-08,
+      "loss": 2.5136,
+      "step": 7942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8396233149076165e-08,
+      "loss": 2.4809,
+      "step": 7943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.809068770979032e-08,
+      "loss": 2.4325,
+      "step": 7944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.778593904757435e-08,
+      "loss": 2.5616,
+      "step": 7945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.748198721156284e-08,
+      "loss": 2.4539,
+      "step": 7946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.717883225076215e-08,
+      "loss": 2.5822,
+      "step": 7947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.687647421404874e-08,
+      "loss": 2.4633,
+      "step": 7948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.657491315017194e-08,
+      "loss": 2.5609,
+      "step": 7949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.627414910775175e-08,
+      "loss": 2.594,
+      "step": 7950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.597418213527939e-08,
+      "loss": 2.3858,
+      "step": 7951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.5675012281119486e-08,
+      "loss": 2.4699,
+      "step": 7952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.537663959350514e-08,
+      "loss": 2.3078,
+      "step": 7953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.507906412054398e-08,
+      "loss": 2.553,
+      "step": 7954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.478228591021262e-08,
+      "loss": 2.5887,
+      "step": 7955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.4486305010361116e-08,
+      "loss": 2.4975,
+      "step": 7956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.419112146870964e-08,
+      "loss": 2.4709,
+      "step": 7957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.389673533284956e-08,
+      "loss": 2.5744,
+      "step": 7958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.360314665024513e-08,
+      "loss": 2.3592,
+      "step": 7959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3310355468231266e-08,
+      "loss": 2.5321,
+      "step": 7960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3018361834014095e-08,
+      "loss": 2.4584,
+      "step": 7961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.272716579467097e-08,
+      "loss": 2.5314,
+      "step": 7962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.243676739715153e-08,
+      "loss": 2.4053,
+      "step": 7963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.214716668827558e-08,
+      "loss": 2.6091,
+      "step": 7964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1858363714736294e-08,
+      "loss": 2.4558,
+      "step": 7965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.157035852309589e-08,
+      "loss": 2.3415,
+      "step": 7966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.128315115978943e-08,
+      "loss": 2.5186,
+      "step": 7967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0996741671123226e-08,
+      "loss": 2.4359,
+      "step": 7968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.071113010327366e-08,
+      "loss": 2.5279,
+      "step": 7969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.042631650229057e-08,
+      "loss": 2.4521,
+      "step": 7970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.014230091409389e-08,
+      "loss": 2.6516,
+      "step": 7971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.985908338447476e-08,
+      "loss": 2.2857,
+      "step": 7972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9576663959096104e-08,
+      "loss": 2.5611,
+      "step": 7973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.929504268349095e-08,
+      "loss": 2.4419,
+      "step": 7974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.901421960306685e-08,
+      "loss": 2.45,
+      "step": 7975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8734194763098706e-08,
+      "loss": 2.5355,
+      "step": 7976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.845496820873541e-08,
+      "loss": 2.4989,
+      "step": 7977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.817653998499594e-08,
+      "loss": 2.5958,
+      "step": 7978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7898910136770505e-08,
+      "loss": 2.4397,
+      "step": 7979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7622078708822184e-08,
+      "loss": 2.4935,
+      "step": 7980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.734604574578305e-08,
+      "loss": 2.4365,
+      "step": 7981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7070811292157513e-08,
+      "loss": 2.6389,
+      "step": 7982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6796375392321734e-08,
+      "loss": 2.5946,
+      "step": 7983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6522738090522545e-08,
+      "loss": 2.5094,
+      "step": 7984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.624989943087799e-08,
+      "loss": 2.6543,
+      "step": 7985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.597785945737732e-08,
+      "loss": 2.5055,
+      "step": 7986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5706618213881005e-08,
+      "loss": 2.4786,
+      "step": 7987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.543617574412185e-08,
+      "loss": 2.6198,
+      "step": 7988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.516653209170163e-08,
+      "loss": 2.4092,
+      "step": 7989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.489768730009558e-08,
+      "loss": 2.5965,
+      "step": 7990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.462964141264847e-08,
+      "loss": 2.4933,
+      "step": 7991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.43623944725774e-08,
+      "loss": 2.5014,
+      "step": 7992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4095946522970664e-08,
+      "loss": 2.445,
+      "step": 7993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.383029760678614e-08,
+      "loss": 2.4396,
+      "step": 7994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.356544776685456e-08,
+      "loss": 2.7581,
+      "step": 7995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.330139704587788e-08,
+      "loss": 2.6582,
+      "step": 7996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.30381454864276e-08,
+      "loss": 2.6077,
+      "step": 7997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2775693130948094e-08,
+      "loss": 2.4998,
+      "step": 7998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2514040021753856e-08,
+      "loss": 2.5785,
+      "step": 7999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2253186201031135e-08,
+      "loss": 2.6848,
+      "step": 8000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1993131710836855e-08,
+      "loss": 2.5215,
+      "step": 8001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.173387659309913e-08,
+      "loss": 2.4724,
+      "step": 8002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.147542088961787e-08,
+      "loss": 2.4318,
+      "step": 8003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1217764642062505e-08,
+      "loss": 2.5185,
+      "step": 8004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0960907891975376e-08,
+      "loss": 2.5431,
+      "step": 8005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.07048506807689e-08,
+      "loss": 2.3466,
+      "step": 8006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.044959304972673e-08,
+      "loss": 2.4217,
+      "step": 8007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.019513504000372e-08,
+      "loss": 2.4986,
+      "step": 8008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.994147669262649e-08,
+      "loss": 2.5625,
+      "step": 8009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.968861804849067e-08,
+      "loss": 2.5648,
+      "step": 8010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9436559148365304e-08,
+      "loss": 2.4397,
+      "step": 8011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9185300032889005e-08,
+      "loss": 2.411,
+      "step": 8012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8934840742572145e-08,
+      "loss": 2.3682,
+      "step": 8013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8685181317796306e-08,
+      "loss": 2.5185,
+      "step": 8014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.843632179881318e-08,
+      "loss": 2.4517,
+      "step": 8015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.818826222574568e-08,
+      "loss": 2.5366,
+      "step": 8016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7941002638589577e-08,
+      "loss": 2.5976,
+      "step": 8017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7694543077208546e-08,
+      "loss": 2.5096,
+      "step": 8018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.744888358133913e-08,
+      "loss": 2.3945,
+      "step": 8019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.720402419058966e-08,
+      "loss": 2.5915,
+      "step": 8020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.695996494443799e-08,
+      "loss": 2.5585,
+      "step": 8021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6716705882233194e-08,
+      "loss": 2.5558,
+      "step": 8022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.647424704319558e-08,
+      "loss": 2.4261,
+      "step": 8023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6232588466417195e-08,
+      "loss": 2.5114,
+      "step": 8024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5991730190859666e-08,
+      "loss": 2.555,
+      "step": 8025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5751672255355806e-08,
+      "loss": 2.5462,
+      "step": 8026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.551241469861077e-08,
+      "loss": 2.4342,
+      "step": 8027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5273957559199265e-08,
+      "loss": 2.3793,
+      "step": 8028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.50363008755672e-08,
+      "loss": 2.4913,
+      "step": 8029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.479944468603169e-08,
+      "loss": 2.4706,
+      "step": 8030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.456338902878109e-08,
+      "loss": 2.4413,
+      "step": 8031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4328133941873844e-08,
+      "loss": 2.5269,
+      "step": 8032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.409367946324071e-08,
+      "loss": 2.726,
+      "step": 8033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386002563068147e-08,
+      "loss": 2.4281,
+      "step": 8034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.362717248186764e-08,
+      "loss": 2.5992,
+      "step": 8035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.339512005434309e-08,
+      "loss": 2.3747,
+      "step": 8036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.316386838552011e-08,
+      "loss": 2.5118,
+      "step": 8037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.293341751268386e-08,
+      "loss": 2.5261,
+      "step": 8038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.270376747298853e-08,
+      "loss": 2.4839,
+      "step": 8039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2474918303461165e-08,
+      "loss": 2.3934,
+      "step": 8040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.224687004099891e-08,
+      "loss": 2.5952,
+      "step": 8041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2019622722369024e-08,
+      "loss": 2.5002,
+      "step": 8042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.179317638421109e-08,
+      "loss": 2.492,
+      "step": 8043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.156753106303367e-08,
+      "loss": 2.425,
+      "step": 8044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.134268679521824e-08,
+      "loss": 2.5371,
+      "step": 8045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1118643617015775e-08,
+      "loss": 2.4483,
+      "step": 8046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0895401564547954e-08,
+      "loss": 2.4316,
+      "step": 8047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0672960673808205e-08,
+      "loss": 2.5201,
+      "step": 8048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0451320980660615e-08,
+      "loss": 2.4943,
+      "step": 8049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0230482520839376e-08,
+      "loss": 2.4389,
+      "step": 8050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0010445329950456e-08,
+      "loss": 2.4476,
+      "step": 8051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.979120944346936e-08,
+      "loss": 2.5211,
+      "step": 8052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9572774896743928e-08,
+      "loss": 2.6114,
+      "step": 8053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.93551417249921e-08,
+      "loss": 2.5514,
+      "step": 8054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9138309963301913e-08,
+      "loss": 2.4543,
+      "step": 8055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.892227964663319e-08,
+      "loss": 2.4639,
+      "step": 8056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8707050809816394e-08,
+      "loss": 2.2849,
+      "step": 8057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8492623487552106e-08,
+      "loss": 2.4419,
+      "step": 8058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.827899771441267e-08,
+      "loss": 2.7205,
+      "step": 8059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8066173524839978e-08,
+      "loss": 2.2535,
+      "step": 8060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7854150953148253e-08,
+      "loss": 2.3218,
+      "step": 8061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.76429300335207e-08,
+      "loss": 2.4108,
+      "step": 8062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7432510800012855e-08,
+      "loss": 2.4714,
+      "step": 8063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.722289328654981e-08,
+      "loss": 2.4038,
+      "step": 8064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7014077526928418e-08,
+      "loss": 2.6233,
+      "step": 8065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6806063554815632e-08,
+      "loss": 2.4373,
+      "step": 8066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.659885140374907e-08,
+      "loss": 2.593,
+      "step": 8067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6392441107137013e-08,
+      "loss": 2.4281,
+      "step": 8068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6186832698258946e-08,
+      "loss": 2.4735,
+      "step": 8069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5982026210265575e-08,
+      "loss": 2.4899,
+      "step": 8070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5778021676176047e-08,
+      "loss": 2.4278,
+      "step": 8071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5574819128882933e-08,
+      "loss": 2.4526,
+      "step": 8072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5372418601147807e-08,
+      "loss": 2.6938,
+      "step": 8073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5170820125602902e-08,
+      "loss": 2.3191,
+      "step": 8074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.497002373475277e-08,
+      "loss": 2.5696,
+      "step": 8075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4770029460970956e-08,
+      "loss": 2.4943,
+      "step": 8076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4570837336502784e-08,
+      "loss": 2.475,
+      "step": 8077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4372447393462562e-08,
+      "loss": 2.5633,
+      "step": 8078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.417485966383748e-08,
+      "loss": 2.4631,
+      "step": 8079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3978074179483722e-08,
+      "loss": 2.5786,
+      "step": 8080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3782090972129244e-08,
+      "loss": 2.5463,
+      "step": 8081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3586910073371548e-08,
+      "loss": 2.6023,
+      "step": 8082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3392531514679906e-08,
+      "loss": 2.4445,
+      "step": 8083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.319895532739369e-08,
+      "loss": 2.6117,
+      "step": 8084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3006181542722384e-08,
+      "loss": 2.4663,
+      "step": 8085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.281421019174668e-08,
+      "loss": 2.3659,
+      "step": 8086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.262304130541848e-08,
+      "loss": 2.2293,
+      "step": 8087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.243267491455925e-08,
+      "loss": 2.4403,
+      "step": 8088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2243111049861655e-08,
+      "loss": 2.4967,
+      "step": 8089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.205434974188847e-08,
+      "loss": 2.3527,
+      "step": 8090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.186639102107424e-08,
+      "loss": 2.4984,
+      "step": 8091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1679234917721946e-08,
+      "loss": 2.4879,
+      "step": 8092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1492881462007452e-08,
+      "loss": 2.602,
+      "step": 8093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1307330683976722e-08,
+      "loss": 2.4107,
+      "step": 8094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1122582613544718e-08,
+      "loss": 2.468,
+      "step": 8095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.093863728049872e-08,
+      "loss": 2.3734,
+      "step": 8096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.075549471449556e-08,
+      "loss": 2.4779,
+      "step": 8097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.057315494506329e-08,
+      "loss": 2.5257,
+      "step": 8098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.039161800160061e-08,
+      "loss": 2.501,
+      "step": 8099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0210883913376334e-08,
+      "loss": 2.5441,
+      "step": 8100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0030952709529927e-08,
+      "loss": 2.5297,
+      "step": 8101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9851824419070965e-08,
+      "loss": 2.513,
+      "step": 8102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967349907088023e-08,
+      "loss": 2.5316,
+      "step": 8103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9495976693709727e-08,
+      "loss": 2.4145,
+      "step": 8104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9319257316179894e-08,
+      "loss": 2.3765,
+      "step": 8105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.914334096678405e-08,
+      "loss": 2.4922,
+      "step": 8106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8968227673883956e-08,
+      "loss": 2.5679,
+      "step": 8107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8793917465713686e-08,
+      "loss": 2.4788,
+      "step": 8108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8620410370376875e-08,
+      "loss": 2.5197,
+      "step": 8109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8447706415847255e-08,
+      "loss": 2.5933,
+      "step": 8110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8275805629969778e-08,
+      "loss": 2.5009,
+      "step": 8111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8104708040460605e-08,
+      "loss": 2.4543,
+      "step": 8112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7934413674904894e-08,
+      "loss": 2.4384,
+      "step": 8113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7764922560759014e-08,
+      "loss": 2.4955,
+      "step": 8114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7596234725349437e-08,
+      "loss": 2.4736,
+      "step": 8115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.742835019587441e-08,
+      "loss": 2.4556,
+      "step": 8116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7261268999400616e-08,
+      "loss": 2.4773,
+      "step": 8117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7094991162867057e-08,
+      "loss": 2.4661,
+      "step": 8118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6929516713082295e-08,
+      "loss": 2.3319,
+      "step": 8119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.676484567672554e-08,
+      "loss": 2.5926,
+      "step": 8120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6600978080346108e-08,
+      "loss": 2.398,
+      "step": 8121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6437913950364538e-08,
+      "loss": 2.4773,
+      "step": 8122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6275653313072015e-08,
+      "loss": 2.6735,
+      "step": 8123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6114196194628174e-08,
+      "loss": 2.4705,
+      "step": 8124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5953542621066075e-08,
+      "loss": 2.5295,
+      "step": 8125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5793692618286115e-08,
+      "loss": 2.5514,
+      "step": 8126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.563464621206212e-08,
+      "loss": 2.4929,
+      "step": 8127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5476403428035803e-08,
+      "loss": 2.3954,
+      "step": 8128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5318964291721196e-08,
+      "loss": 2.6618,
+      "step": 8129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5162328828501328e-08,
+      "loss": 2.6144,
+      "step": 8130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5006497063631e-08,
+      "loss": 2.7542,
+      "step": 8131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4851469022234e-08,
+      "loss": 2.2789,
+      "step": 8132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4697244729305894e-08,
+      "loss": 2.5679,
+      "step": 8133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4543824209711788e-08,
+      "loss": 2.3904,
+      "step": 8134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4391207488187453e-08,
+      "loss": 2.5307,
+      "step": 8135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4239394589339317e-08,
+      "loss": 2.3717,
+      "step": 8136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4088385537643912e-08,
+      "loss": 2.4622,
+      "step": 8137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3938180357447318e-08,
+      "loss": 2.5982,
+      "step": 8138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3788779072968495e-08,
+      "loss": 2.5639,
+      "step": 8139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3640181708293731e-08,
+      "loss": 2.7357,
+      "step": 8140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3492388287382196e-08,
+      "loss": 2.3608,
+      "step": 8141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3345398834062051e-08,
+      "loss": 2.3723,
+      "step": 8142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3199213372031561e-08,
+      "loss": 2.4542,
+      "step": 8143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3053831924861315e-08,
+      "loss": 2.3731,
+      "step": 8144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2909254515989788e-08,
+      "loss": 2.5814,
+      "step": 8145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2765481168727779e-08,
+      "loss": 2.649,
+      "step": 8146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2622511906255075e-08,
+      "loss": 2.5635,
+      "step": 8147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2480346751622686e-08,
+      "loss": 2.4383,
+      "step": 8148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2338985727751719e-08,
+      "loss": 2.5835,
+      "step": 8149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2198428857433941e-08,
+      "loss": 2.5639,
+      "step": 8150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2058676163330119e-08,
+      "loss": 2.5097,
+      "step": 8151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1919727667973335e-08,
+      "loss": 2.6043,
+      "step": 8152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1781583393765673e-08,
+      "loss": 2.6138,
+      "step": 8153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1644243362980423e-08,
+      "loss": 2.4462,
+      "step": 8154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1507707597759876e-08,
+      "loss": 2.5627,
+      "step": 8155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.137197612011809e-08,
+      "loss": 2.4642,
+      "step": 8156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1237048951938666e-08,
+      "loss": 2.563,
+      "step": 8157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1102926114975877e-08,
+      "loss": 2.4635,
+      "step": 8158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0969607630854084e-08,
+      "loss": 2.4442,
+      "step": 8159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0837093521067765e-08,
+      "loss": 2.5965,
+      "step": 8160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0705383806982606e-08,
+      "loss": 2.5469,
+      "step": 8161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0574478509833286e-08,
+      "loss": 2.5637,
+      "step": 8162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0444377650725702e-08,
+      "loss": 2.5458,
+      "step": 8163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0315081250636405e-08,
+      "loss": 2.4314,
+      "step": 8164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0186589330410945e-08,
+      "loss": 2.5568,
+      "step": 8165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0058901910766083e-08,
+      "loss": 2.5783,
+      "step": 8166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.932019012288685e-09,
+      "loss": 2.55,
+      "step": 8167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.805940655436274e-09,
+      "loss": 2.3963,
+      "step": 8168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.68066686053537e-09,
+      "loss": 2.482,
+      "step": 8169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.556197647784814e-09,
+      "loss": 2.4012,
+      "step": 8170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.43253303725189e-09,
+      "loss": 2.465,
+      "step": 8171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.30967304887509e-09,
+      "loss": 2.5418,
+      "step": 8172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.187617702463014e-09,
+      "loss": 2.5482,
+      "step": 8173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.066367017694366e-09,
+      "loss": 2.381,
+      "step": 8174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.945921014117954e-09,
+      "loss": 2.5039,
+      "step": 8175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.826279711153795e-09,
+      "loss": 2.5449,
+      "step": 8176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.707443128090908e-09,
+      "loss": 2.5104,
+      "step": 8177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.589411284089522e-09,
+      "loss": 2.6798,
+      "step": 8178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.472184198179412e-09,
+      "loss": 2.4502,
+      "step": 8179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.35576188926046e-09,
+      "loss": 2.4924,
+      "step": 8180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.240144376104876e-09,
+      "loss": 2.5149,
+      "step": 8181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.125331677352188e-09,
+      "loss": 2.3395,
+      "step": 8182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.011323811513704e-09,
+      "loss": 2.4903,
+      "step": 8183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.898120796970831e-09,
+      "loss": 2.6128,
+      "step": 8184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.78572265197508e-09,
+      "loss": 2.5722,
+      "step": 8185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.674129394648623e-09,
+      "loss": 2.4804,
+      "step": 8186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.563341042983174e-09,
+      "loss": 2.4066,
+      "step": 8187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.453357614841116e-09,
+      "loss": 2.401,
+      "step": 8188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.344179127954376e-09,
+      "loss": 2.5883,
+      "step": 8189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.235805599926093e-09,
+      "loss": 2.5615,
+      "step": 8190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.128237048229514e-09,
+      "loss": 2.4466,
+      "step": 8191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.0214734902074314e-09,
+      "loss": 2.4612,
+      "step": 8192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.9155149430732985e-09,
+      "loss": 2.4596,
+      "step": 8193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.810361423910117e-09,
+      "loss": 2.5326,
+      "step": 8194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.706012949672103e-09,
+      "loss": 2.3822,
+      "step": 8195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.60246953718302e-09,
+      "loss": 2.5622,
+      "step": 8196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.499731203137294e-09,
+      "loss": 2.6148,
+      "step": 8197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.3977979640994504e-09,
+      "loss": 2.4227,
+      "step": 8198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.296669836503566e-09,
+      "loss": 2.3367,
+      "step": 8199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.196346836654932e-09,
+      "loss": 2.571,
+      "step": 8200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.096828980727832e-09,
+      "loss": 2.3922,
+      "step": 8201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9981162847672085e-09,
+      "loss": 2.4372,
+      "step": 8202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.900208764689774e-09,
+      "loss": 2.6134,
+      "step": 8203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.803106436279571e-09,
+      "loss": 2.4157,
+      "step": 8204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.706809315192963e-09,
+      "loss": 2.5498,
+      "step": 8205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.611317416955308e-09,
+      "loss": 2.4451,
+      "step": 8206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.516630756963182e-09,
+      "loss": 2.3918,
+      "step": 8207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.422749350482148e-09,
+      "loss": 2.3648,
+      "step": 8208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.329673212649544e-09,
+      "loss": 2.4885,
+      "step": 8209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.237402358471144e-09,
+      "loss": 2.3634,
+      "step": 8210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.145936802823937e-09,
+      "loss": 2.4086,
+      "step": 8211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.055276560454459e-09,
+      "loss": 2.5217,
+      "step": 8212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.965421645979907e-09,
+      "loss": 2.4708,
+      "step": 8213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.876372073887581e-09,
+      "loss": 2.3934,
+      "step": 8214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.788127858534886e-09,
+      "loss": 2.4085,
+      "step": 8215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.700689014149329e-09,
+      "loss": 2.528,
+      "step": 8216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.614055554827967e-09,
+      "loss": 2.557,
+      "step": 8217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5282274945396276e-09,
+      "loss": 2.497,
+      "step": 8218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.443204847121019e-09,
+      "loss": 2.5316,
+      "step": 8219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.358987626281175e-09,
+      "loss": 2.4091,
+      "step": 8220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.275575845597568e-09,
+      "loss": 2.4405,
+      "step": 8221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.192969518519441e-09,
+      "loss": 2.4955,
+      "step": 8222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.111168658365028e-09,
+      "loss": 2.4421,
+      "step": 8223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0301732783226685e-09,
+      "loss": 2.4354,
+      "step": 8224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.949983391450807e-09,
+      "loss": 2.4029,
+      "step": 8225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.870599010679099e-09,
+      "loss": 2.6248,
+      "step": 8226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.792020148806197e-09,
+      "loss": 2.3476,
+      "step": 8227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.71424681850141e-09,
+      "loss": 2.462,
+      "step": 8228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.637279032304153e-09,
+      "loss": 2.5265,
+      "step": 8229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5611168026228326e-09,
+      "loss": 2.6097,
+      "step": 8230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4857601417387365e-09,
+      "loss": 2.4939,
+      "step": 8231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.411209061799925e-09,
+      "loss": 2.3645,
+      "step": 8232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3374635748267827e-09,
+      "loss": 2.2656,
+      "step": 8233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2645236927092426e-09,
+      "loss": 2.5451,
+      "step": 8234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.192389427207343e-09,
+      "loss": 2.5124,
+      "step": 8235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1210607899512244e-09,
+      "loss": 2.6119,
+      "step": 8236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0505377924405778e-09,
+      "loss": 2.428,
+      "step": 8237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9808204460463064e-09,
+      "loss": 2.3868,
+      "step": 8238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.911908762008864e-09,
+      "loss": 2.4861,
+      "step": 8239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8438027514382517e-09,
+      "loss": 2.5844,
+      "step": 8240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7765024253156856e-09,
+      "loss": 2.6198,
+      "step": 8241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.710007794491376e-09,
+      "loss": 2.4432,
+      "step": 8242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6443188696867462e-09,
+      "loss": 2.4262,
+      "step": 8243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.579435661492213e-09,
+      "loss": 2.5472,
+      "step": 8244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5153581803688544e-09,
+      "loss": 2.5657,
+      "step": 8245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.452086436648404e-09,
+      "loss": 2.4245,
+      "step": 8246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3896204405310373e-09,
+      "loss": 2.4933,
+      "step": 8247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3279602020892522e-09,
+      "loss": 2.6031,
+      "step": 8248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2671057312634305e-09,
+      "loss": 2.4978,
+      "step": 8249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.207057037865168e-09,
+      "loss": 2.4621,
+      "step": 8250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.14781413157672e-09,
+      "loss": 2.6318,
+      "step": 8251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0893770219493347e-09,
+      "loss": 2.5875,
+      "step": 8252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0317457184043654e-09,
+      "loss": 2.4807,
+      "step": 8253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9749202302343784e-09,
+      "loss": 2.5129,
+      "step": 8254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.918900566600379e-09,
+      "loss": 2.5663,
+      "step": 8255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.863686736535697e-09,
+      "loss": 2.4534,
+      "step": 8256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8092787489404352e-09,
+      "loss": 2.6267,
+      "step": 8257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.755676612588686e-09,
+      "loss": 2.3115,
+      "step": 8258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7028803361213152e-09,
+      "loss": 2.4257,
+      "step": 8259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6508899280515134e-09,
+      "loss": 2.5104,
+      "step": 8260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5997053967603538e-09,
+      "loss": 2.5339,
+      "step": 8261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5493267505017895e-09,
+      "loss": 2.3132,
+      "step": 8262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.499753997397102e-09,
+      "loss": 2.448,
+      "step": 8263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.450987145439342e-09,
+      "loss": 2.5244,
+      "step": 8264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.403026202491109e-09,
+      "loss": 2.355,
+      "step": 8265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3558711762851063e-09,
+      "loss": 2.5497,
+      "step": 8266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.309522074424141e-09,
+      "loss": 2.4854,
+      "step": 8267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2639789043805695e-09,
+      "loss": 2.4989,
+      "step": 8268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2192416734979617e-09,
+      "loss": 2.3725,
+      "step": 8269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1753103889883267e-09,
+      "loss": 2.5154,
+      "step": 8270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.132185057935442e-09,
+      "loss": 2.3742,
+      "step": 8271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0898656872920798e-09,
+      "loss": 2.5806,
+      "step": 8272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0483522838811155e-09,
+      "loss": 2.4613,
+      "step": 8273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0076448543960838e-09,
+      "loss": 2.4708,
+      "step": 8274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.677434053995127e-10,
+      "loss": 2.4681,
+      "step": 8275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.286479433257e-10,
+      "loss": 2.5984,
+      "step": 8276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.903584744768268e-10,
+      "loss": 2.494,
+      "step": 8277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.528750050273982e-10,
+      "loss": 2.5161,
+      "step": 8278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.161975410198031e-10,
+      "loss": 2.3601,
+      "step": 8279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.803260883681996e-10,
+      "loss": 2.5558,
+      "step": 8280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.452606528557393e-10,
+      "loss": 2.5375,
+      "step": 8281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.110012401362332e-10,
+      "loss": 2.5502,
+      "step": 8282
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.775478557324855e-10,
+      "loss": 2.6348,
+      "step": 8283
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.4490050503907e-10,
+      "loss": 2.5907,
+      "step": 8284
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.13059193319554e-10,
+      "loss": 2.5646,
+      "step": 8285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.820239257070537e-10,
+      "loss": 2.281,
+      "step": 8286
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.517947072064544e-10,
+      "loss": 2.5137,
+      "step": 8287
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.2237154268997e-10,
+      "loss": 2.714,
+      "step": 8288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.937544369032488e-10,
+      "loss": 2.6214,
+      "step": 8289
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.659433944587122e-10,
+      "loss": 2.5226,
+      "step": 8290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.389384198405511e-10,
+      "loss": 2.6566,
+      "step": 8291
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.127395174036153e-10,
+      "loss": 2.3989,
+      "step": 8292
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8734669137119316e-10,
+      "loss": 2.3277,
+      "step": 8293
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.627599458377873e-10,
+      "loss": 2.5224,
+      "step": 8294
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.389792847668938e-10,
+      "loss": 2.6145,
+      "step": 8295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1600471199322304e-10,
+      "loss": 2.4101,
+      "step": 8296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.938362312199239e-10,
+      "loss": 2.4066,
+      "step": 8297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7247384602246964e-10,
+      "loss": 2.3178,
+      "step": 8298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5191755984477204e-10,
+      "loss": 2.4465,
+      "step": 8299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.321673760002918e-10,
+      "loss": 2.392,
+      "step": 8300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1322329767425875e-10,
+      "loss": 2.497,
+      "step": 8301
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.950853279203413e-10,
+      "loss": 2.365,
+      "step": 8302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.77753469663422e-10,
+      "loss": 2.5538,
+      "step": 8303
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6122772569793221e-10,
+      "loss": 2.5754,
+      "step": 8304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4550809868729698e-10,
+      "loss": 2.3786,
+      "step": 8305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.305945911672657e-10,
+      "loss": 2.2847,
+      "step": 8306
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.164872055414712e-10,
+      "loss": 2.3156,
+      "step": 8307
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0318594408476045e-10,
+      "loss": 2.5269,
+      "step": 8308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.069080894152926e-11,
+      "loss": 2.4647,
+      "step": 8309
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.90018021268324e-11,
+      "loss": 2.339,
+      "step": 8310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.811892552471832e-11,
+      "loss": 2.3976,
+      "step": 8311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.804218088933944e-11,
+      "loss": 2.4379,
+      "step": 8312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.877156984661735e-11,
+      "loss": 2.5229,
+      "step": 8313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.030709389035714e-11,
+      "loss": 2.5463,
+      "step": 8314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.264875438613313e-11,
+      "loss": 2.5688,
+      "step": 8315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.57965525674031e-11,
+      "loss": 2.4653,
+      "step": 8316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.975048953994918e-11,
+      "loss": 2.4684,
+      "step": 8317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4510566277992077e-11,
+      "loss": 2.5291,
+      "step": 8318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0076783626966625e-11,
+      "loss": 2.4207,
+      "step": 8319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.449142300746225e-12,
+      "loss": 2.5677,
+      "step": 8320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6276428855286373e-12,
+      "loss": 2.5879,
+      "step": 8321
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6122858353950776e-12,
+      "loss": 2.5597,
+      "step": 8322
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.030714750857811e-13,
+      "loss": 2.5695,
+      "step": 8323
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 2.521,
+      "step": 8324
+    },
+    {
+      "epoch": 1.0,
+      "step": 8324,
+      "total_flos": 1.0732074279370752e+18,
+      "train_loss": 2.5607235752465,
+      "train_runtime": 9425.8937,
+      "train_samples_per_second": 56.515,
+      "train_steps_per_second": 0.883
+    }
+  ],
+  "max_steps": 8324,
+  "num_train_epochs": 1,
+  "total_flos": 1.0732074279370752e+18,
+  "trial_name": null,
+  "trial_params": null
+}