diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,28627 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999475588651738,
+  "global_step": 4767,
+  "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": 3.092,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 3.1007,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 3.1567,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 3.1164,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 3.0094,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 3.1256,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-07,
+      "loss": 3.1288,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 3.1003,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 3.0793,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.0211,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-07,
+      "loss": 3.1464,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 3.1738,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-07,
+      "loss": 3.0976,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8e-07,
+      "loss": 3.0037,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.1165,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2e-07,
+      "loss": 3.0801,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 3.0692,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 3.129,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-07,
+      "loss": 3.08,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.1163,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 3.1214,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-07,
+      "loss": 3.1527,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 3.061,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 3.094,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.1331,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2e-07,
+      "loss": 3.1798,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4e-07,
+      "loss": 3.07,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6e-07,
+      "loss": 3.1276,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 3.0871,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.0546,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 3.0465,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4e-07,
+      "loss": 3.1104,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6e-07,
+      "loss": 3.1242,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 3.048,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.0106,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-07,
+      "loss": 3.1382,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4e-07,
+      "loss": 3.0921,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6e-07,
+      "loss": 3.069,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.8e-07,
+      "loss": 3.1002,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.1066,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 3.0689,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 3.0155,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.6e-07,
+      "loss": 3.0663,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8e-07,
+      "loss": 3.0566,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.0471,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 3.0668,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 3.0367,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 3.1599,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 3.0559,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.0357,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02e-06,
+      "loss": 3.0069,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.04e-06,
+      "loss": 3.0106,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06e-06,
+      "loss": 3.0893,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.08e-06,
+      "loss": 3.0402,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1e-06,
+      "loss": 2.9643,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.12e-06,
+      "loss": 3.0251,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.14e-06,
+      "loss": 3.0735,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 3.0517,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 3.0325,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.9646,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 3.0318,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 3.0589,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26e-06,
+      "loss": 2.9803,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.28e-06,
+      "loss": 3.0203,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3e-06,
+      "loss": 3.0212,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.32e-06,
+      "loss": 2.9209,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.34e-06,
+      "loss": 3.0635,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 3.0862,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 2.91,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 3.0107,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.42e-06,
+      "loss": 2.979,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.44e-06,
+      "loss": 2.9974,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.46e-06,
+      "loss": 2.8758,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.48e-06,
+      "loss": 2.9534,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5e-06,
+      "loss": 2.9901,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.52e-06,
+      "loss": 3.0355,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.54e-06,
+      "loss": 2.8663,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.56e-06,
+      "loss": 2.9364,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 2.909,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.9126,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 2.9771,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 2.8949,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 2.8846,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 2.928,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 2.9486,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.72e-06,
+      "loss": 2.9353,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.74e-06,
+      "loss": 2.8898,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.76e-06,
+      "loss": 2.9294,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 2.9448,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 2.874,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 2.837,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 2.8484,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 2.8547,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 2.8736,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 2.8729,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 2.9013,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.94e-06,
+      "loss": 2.8442,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 2.8887,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.98e-06,
+      "loss": 2.8369,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.9085,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.02e-06,
+      "loss": 2.7856,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.04e-06,
+      "loss": 2.803,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.06e-06,
+      "loss": 2.8903,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.08e-06,
+      "loss": 2.8372,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 2.8824,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.12e-06,
+      "loss": 2.8255,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 2.8445,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.16e-06,
+      "loss": 2.8626,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 2.8125,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2e-06,
+      "loss": 2.8063,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 2.9466,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.24e-06,
+      "loss": 2.8213,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 2.8191,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.28e-06,
+      "loss": 2.8357,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 2.8093,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 2.8186,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 2.745,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 2.7376,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.38e-06,
+      "loss": 2.8125,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.7233,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.42e-06,
+      "loss": 2.7882,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 2.8079,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.46e-06,
+      "loss": 2.6885,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 2.7167,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5e-06,
+      "loss": 2.7285,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.52e-06,
+      "loss": 2.8242,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 2.7685,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.56e-06,
+      "loss": 2.7729,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 2.7742,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6e-06,
+      "loss": 2.7129,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 2.7117,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.64e-06,
+      "loss": 2.7372,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 2.7254,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.68e-06,
+      "loss": 2.726,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 2.7157,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 2.7942,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 2.7229,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 2.7787,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 2.7573,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.7267,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.82e-06,
+      "loss": 2.71,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.84e-06,
+      "loss": 2.7361,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.86e-06,
+      "loss": 2.6781,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.88e-06,
+      "loss": 2.7683,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9e-06,
+      "loss": 2.7627,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.92e-06,
+      "loss": 2.7361,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 2.739,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.96e-06,
+      "loss": 2.7379,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 2.6021,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3e-06,
+      "loss": 2.697,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 2.6825,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.04e-06,
+      "loss": 2.6351,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 2.6797,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.08e-06,
+      "loss": 2.6717,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 2.7383,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.12e-06,
+      "loss": 2.7181,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 2.796,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 2.7223,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 2.7008,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.8344,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 2.6974,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 2.6517,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 2.7337,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 2.6399,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 2.6794,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 2.7321,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 2.7192,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 2.6636,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 2.6598,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 2.6776,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 2.61,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.44e-06,
+      "loss": 2.6884,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.46e-06,
+      "loss": 2.6647,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.48e-06,
+      "loss": 2.6741,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5e-06,
+      "loss": 2.7231,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.52e-06,
+      "loss": 2.6815,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.54e-06,
+      "loss": 2.6968,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 2.6812,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.58e-06,
+      "loss": 2.7574,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 2.7588,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.62e-06,
+      "loss": 2.7068,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 2.7235,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.66e-06,
+      "loss": 2.7384,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 2.6664,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7e-06,
+      "loss": 2.7667,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 2.7026,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.74e-06,
+      "loss": 2.7062,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 2.6756,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 2.646,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 2.7622,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 2.6983,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 2.7022,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.86e-06,
+      "loss": 2.6542,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.88e-06,
+      "loss": 2.7237,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 2.7299,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 2.735,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.94e-06,
+      "loss": 2.7275,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.96e-06,
+      "loss": 2.5665,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 2.7009,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.6769,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 2.6923,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.04e-06,
+      "loss": 2.6636,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 2.7309,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.08e-06,
+      "loss": 2.6748,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1e-06,
+      "loss": 2.7055,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.12e-06,
+      "loss": 2.7203,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.14e-06,
+      "loss": 2.6797,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.16e-06,
+      "loss": 2.7229,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.18e-06,
+      "loss": 2.6673,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.7368,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.22e-06,
+      "loss": 2.7388,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.24e-06,
+      "loss": 2.6916,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.26e-06,
+      "loss": 2.7097,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 2.6832,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.3e-06,
+      "loss": 2.6329,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.32e-06,
+      "loss": 2.681,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.34e-06,
+      "loss": 2.666,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 2.7177,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.38e-06,
+      "loss": 2.6559,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.4e-06,
+      "loss": 2.6489,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.42e-06,
+      "loss": 2.6768,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 2.7167,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 2.6897,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.48e-06,
+      "loss": 2.6802,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5e-06,
+      "loss": 2.7444,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 2.6583,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 2.6481,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.56e-06,
+      "loss": 2.6342,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.58e-06,
+      "loss": 2.6935,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 2.6047,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 2.647,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 2.6242,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.66e-06,
+      "loss": 2.7531,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 2.6951,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7e-06,
+      "loss": 2.632,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 2.6911,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74e-06,
+      "loss": 2.7375,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76e-06,
+      "loss": 2.6472,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.78e-06,
+      "loss": 2.6602,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 2.7392,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 2.6402,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.84e-06,
+      "loss": 2.713,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.86e-06,
+      "loss": 2.5896,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 2.7019,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 2.6894,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92e-06,
+      "loss": 2.6391,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.94e-06,
+      "loss": 2.6874,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 2.6786,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 2.7287,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5e-06,
+      "loss": 2.7298,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.02e-06,
+      "loss": 2.7635,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.04e-06,
+      "loss": 2.6722,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 2.6845,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 2.6477,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1e-06,
+      "loss": 2.6842,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.12e-06,
+      "loss": 2.6877,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 2.7476,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 2.6254,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.18e-06,
+      "loss": 2.6529,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2e-06,
+      "loss": 2.6917,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 2.6925,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 2.7004,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 2.6631,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.28e-06,
+      "loss": 2.6592,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 2.6794,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 2.7186,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 2.694,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.36e-06,
+      "loss": 2.7096,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 2.747,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 2.7058,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 2.6247,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 2.6724,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 2.6713,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 2.6121,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 2.6669,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 2.6139,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 2.7561,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 2.7216,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 2.6066,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 2.719,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 2.6883,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.64e-06,
+      "loss": 2.7093,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.66e-06,
+      "loss": 2.7237,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.68e-06,
+      "loss": 2.6871,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.7e-06,
+      "loss": 2.7259,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.72e-06,
+      "loss": 2.708,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.74e-06,
+      "loss": 2.6399,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.76e-06,
+      "loss": 2.6727,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.78e-06,
+      "loss": 2.6684,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.8e-06,
+      "loss": 2.6605,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.82e-06,
+      "loss": 2.6623,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.84e-06,
+      "loss": 2.7235,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.86e-06,
+      "loss": 2.6955,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 2.6622,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9e-06,
+      "loss": 2.6686,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.92e-06,
+      "loss": 2.7455,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.94e-06,
+      "loss": 2.6493,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 2.6973,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.98e-06,
+      "loss": 2.6063,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6e-06,
+      "loss": 2.6333,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.02e-06,
+      "loss": 2.6961,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 2.6822,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 2.661,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.08e-06,
+      "loss": 2.7265,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1e-06,
+      "loss": 2.6347,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 2.6559,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 2.6845,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.16e-06,
+      "loss": 2.6627,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.18e-06,
+      "loss": 2.7212,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 2.6335,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 2.6973,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.24e-06,
+      "loss": 2.6362,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.26e-06,
+      "loss": 2.6919,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 2.6531,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 2.675,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 2.7016,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.34e-06,
+      "loss": 2.6671,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 2.6624,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 2.5969,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 2.7046,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.42e-06,
+      "loss": 2.6086,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 2.6046,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 2.702,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 2.6956,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 2.6409,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 2.5939,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 2.6228,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 2.678,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 2.6929,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 2.7312,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 2.5999,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 2.779,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 2.7276,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 2.6987,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 2.7404,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 2.6658,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 2.6594,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 2.7193,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 2.7091,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 2.6506,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 2.7314,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 2.6665,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 2.6355,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.88e-06,
+      "loss": 2.7268,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.9e-06,
+      "loss": 2.6226,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.92e-06,
+      "loss": 2.6505,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 2.6162,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.96e-06,
+      "loss": 2.668,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.98e-06,
+      "loss": 2.6221,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7e-06,
+      "loss": 2.6671,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 2.6769,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.04e-06,
+      "loss": 2.6362,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.06e-06,
+      "loss": 2.6491,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.08e-06,
+      "loss": 2.6822,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 2.6978,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 2.6441,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.14e-06,
+      "loss": 2.5999,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.16e-06,
+      "loss": 2.6134,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 2.6324,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 2.629,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.22e-06,
+      "loss": 2.686,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.24e-06,
+      "loss": 2.6586,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 2.675,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 2.6989,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.3e-06,
+      "loss": 2.7038,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.32e-06,
+      "loss": 2.6683,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 2.6558,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 2.7088,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 2.6915,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.4e-06,
+      "loss": 2.6242,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 2.695,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 2.6755,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 2.6543,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.48e-06,
+      "loss": 2.709,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.6571,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 2.6986,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 2.6546,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 2.6111,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.58e-06,
+      "loss": 2.6983,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 2.6343,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 2.5914,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 2.6637,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 2.7317,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 2.7443,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.7e-06,
+      "loss": 2.7185,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.72e-06,
+      "loss": 2.6855,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.74e-06,
+      "loss": 2.7489,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.76e-06,
+      "loss": 2.6577,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.78e-06,
+      "loss": 2.7105,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 2.7145,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 2.6599,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 2.6772,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 2.6592,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.88e-06,
+      "loss": 2.6568,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.9e-06,
+      "loss": 2.7224,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.92e-06,
+      "loss": 2.6833,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.94e-06,
+      "loss": 2.6479,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 2.7043,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 2.6985,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.7087,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 2.7109,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 2.6745,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.06e-06,
+      "loss": 2.7217,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.08e-06,
+      "loss": 2.6532,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.1e-06,
+      "loss": 2.717,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 2.6935,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.14e-06,
+      "loss": 2.6677,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.16e-06,
+      "loss": 2.6646,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.18e-06,
+      "loss": 2.6363,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.2e-06,
+      "loss": 2.6701,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 2.6768,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.24e-06,
+      "loss": 2.5956,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.26e-06,
+      "loss": 2.7127,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.28e-06,
+      "loss": 2.623,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.3e-06,
+      "loss": 2.6493,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.32e-06,
+      "loss": 2.6848,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.34e-06,
+      "loss": 2.6693,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.36e-06,
+      "loss": 2.6958,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 2.65,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 2.708,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.42e-06,
+      "loss": 2.6194,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.44e-06,
+      "loss": 2.6234,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.46e-06,
+      "loss": 2.6272,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.48e-06,
+      "loss": 2.65,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.5e-06,
+      "loss": 2.6509,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.52e-06,
+      "loss": 2.6685,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 2.7005,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 2.712,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 2.6477,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.6e-06,
+      "loss": 2.6886,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.62e-06,
+      "loss": 2.6384,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.64e-06,
+      "loss": 2.6381,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.66e-06,
+      "loss": 2.7119,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.68e-06,
+      "loss": 2.6355,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 2.7274,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 2.7153,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 2.5978,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.76e-06,
+      "loss": 2.5917,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.78e-06,
+      "loss": 2.6581,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.8e-06,
+      "loss": 2.6789,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.82e-06,
+      "loss": 2.6134,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.84e-06,
+      "loss": 2.646,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 2.6814,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 2.6677,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 2.7531,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 2.5807,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.94e-06,
+      "loss": 2.6369,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.96e-06,
+      "loss": 2.7019,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.98e-06,
+      "loss": 2.6746,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9e-06,
+      "loss": 2.7104,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 2.6392,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 2.7314,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 2.6959,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 2.7092,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 2.6178,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.12e-06,
+      "loss": 2.6562,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.14e-06,
+      "loss": 2.6787,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.16e-06,
+      "loss": 2.6377,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 2.6307,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 2.6426,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 2.5939,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 2.7205,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 2.6742,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 2.6567,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.3e-06,
+      "loss": 2.5991,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.32e-06,
+      "loss": 2.6078,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 2.6571,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 2.6025,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.38e-06,
+      "loss": 2.6927,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.4e-06,
+      "loss": 2.7022,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.42e-06,
+      "loss": 2.7051,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 2.6819,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 2.6613,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.48e-06,
+      "loss": 2.6323,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.5e-06,
+      "loss": 2.6013,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.52e-06,
+      "loss": 2.7117,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.54e-06,
+      "loss": 2.6806,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.56e-06,
+      "loss": 2.6685,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.58e-06,
+      "loss": 2.6483,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.6636,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 2.6899,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 2.6207,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.66e-06,
+      "loss": 2.6955,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.68e-06,
+      "loss": 2.652,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7e-06,
+      "loss": 2.6327,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.72e-06,
+      "loss": 2.6394,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.74e-06,
+      "loss": 2.7203,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 2.6816,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 2.671,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 2.6616,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 2.7088,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.84e-06,
+      "loss": 2.6852,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86e-06,
+      "loss": 2.6654,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88e-06,
+      "loss": 2.6265,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9e-06,
+      "loss": 2.6451,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 2.6624,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 2.6924,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 2.6539,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 2.6114,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 2.6568,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999998644827911e-06,
+      "loss": 2.6002,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999994579312375e-06,
+      "loss": 2.694,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999987803455598e-06,
+      "loss": 2.6409,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999978317261251e-06,
+      "loss": 2.6881,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999966120734478e-06,
+      "loss": 2.683,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99995121388189e-06,
+      "loss": 2.6212,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99993359671157e-06,
+      "loss": 2.702,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999913269233062e-06,
+      "loss": 2.6486,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999890231457389e-06,
+      "loss": 2.6342,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999864483397036e-06,
+      "loss": 2.6764,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999836025065965e-06,
+      "loss": 2.679,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999804856479599e-06,
+      "loss": 2.6428,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999770977654835e-06,
+      "loss": 2.6633,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999734388610034e-06,
+      "loss": 2.6724,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999695089365035e-06,
+      "loss": 2.6613,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999653079941137e-06,
+      "loss": 2.6902,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999608360361114e-06,
+      "loss": 2.6902,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999560930649206e-06,
+      "loss": 2.6258,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999510790831126e-06,
+      "loss": 2.6515,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999457940934048e-06,
+      "loss": 2.6984,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999402380986624e-06,
+      "loss": 2.727,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999344111018972e-06,
+      "loss": 2.6797,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999283131062676e-06,
+      "loss": 2.6181,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999219441150791e-06,
+      "loss": 2.5897,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999153041317844e-06,
+      "loss": 2.5722,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999083931599826e-06,
+      "loss": 2.7107,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999012112034202e-06,
+      "loss": 2.6413,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.9989375826599e-06,
+      "loss": 2.641,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998860343517321e-06,
+      "loss": 2.6346,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998780394648332e-06,
+      "loss": 2.5769,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998697736096276e-06,
+      "loss": 2.6819,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998612367905956e-06,
+      "loss": 2.6485,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998524290123648e-06,
+      "loss": 2.6902,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998433502797097e-06,
+      "loss": 2.7011,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998340005975512e-06,
+      "loss": 2.6803,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998243799709582e-06,
+      "loss": 2.6414,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99814488405145e-06,
+      "loss": 2.6829,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998043259054741e-06,
+      "loss": 2.6772,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99793892477454e-06,
+      "loss": 2.6349,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997831881267403e-06,
+      "loss": 2.6979,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997722128591354e-06,
+      "loss": 2.7227,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99760966680589e-06,
+      "loss": 2.6399,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99749449597197e-06,
+      "loss": 2.6944,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997376616152026e-06,
+      "loss": 2.6157,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997256027409957e-06,
+      "loss": 2.7393,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99713272981113e-06,
+      "loss": 2.6502,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99700672342238e-06,
+      "loss": 2.6109,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.996878008312013e-06,
+      "loss": 2.6915,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996746584549796e-06,
+      "loss": 2.704,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99661245220698e-06,
+      "loss": 2.7168,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996475611356265e-06,
+      "loss": 2.7126,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996336062071831e-06,
+      "loss": 2.6848,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996193804429325e-06,
+      "loss": 2.6101,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996048838505857e-06,
+      "loss": 2.6074,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995901164380011e-06,
+      "loss": 2.6212,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995750782131836e-06,
+      "loss": 2.6692,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995597691842852e-06,
+      "loss": 2.7155,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995441893596039e-06,
+      "loss": 2.6616,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995283387475854e-06,
+      "loss": 2.639,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995122173568217e-06,
+      "loss": 2.5709,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994958251960518e-06,
+      "loss": 2.6758,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994791622741615e-06,
+      "loss": 2.6284,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99462228600183e-06,
+      "loss": 2.614,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994450241832955e-06,
+      "loss": 2.6514,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99427549032825e-06,
+      "loss": 2.6195,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994098031582446e-06,
+      "loss": 2.6823,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993917865691734e-06,
+      "loss": 2.5984,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993734992753777e-06,
+      "loss": 2.6017,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993549412867705e-06,
+      "loss": 2.6255,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993361126134116e-06,
+      "loss": 2.6241,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993170132655074e-06,
+      "loss": 2.7023,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992976432534108e-06,
+      "loss": 2.5681,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99278002587622e-06,
+      "loss": 2.6258,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992580912787876e-06,
+      "loss": 2.6261,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992379093377007e-06,
+      "loss": 2.7511,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992174567753013e-06,
+      "loss": 2.6543,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991967336026764e-06,
+      "loss": 2.6525,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99175739831059e-06,
+      "loss": 2.6664,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991544754718294e-06,
+      "loss": 2.647,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991329405365142e-06,
+      "loss": 2.7218,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99111135036787e-06,
+      "loss": 2.7051,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990890589844676e-06,
+      "loss": 2.7164,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990667123915232e-06,
+      "loss": 2.7361,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990440952700668e-06,
+      "loss": 2.5928,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990212076323587e-06,
+      "loss": 2.706,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989980494908052e-06,
+      "loss": 2.6298,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989746208579601e-06,
+      "loss": 2.6343,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989509217465227e-06,
+      "loss": 2.5708,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989269521693402e-06,
+      "loss": 2.6394,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989027121394054e-06,
+      "loss": 2.6512,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988782016698581e-06,
+      "loss": 2.663,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988534207739849e-06,
+      "loss": 2.6857,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988283694652183e-06,
+      "loss": 2.7179,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988030477571382e-06,
+      "loss": 2.6835,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.987774556634706e-06,
+      "loss": 2.6431,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.987515931980882e-06,
+      "loss": 2.6393,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.987254603750103e-06,
+      "loss": 2.6753,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.986990572084025e-06,
+      "loss": 2.5921,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.986723837125772e-06,
+      "loss": 2.6672,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.986454399019934e-06,
+      "loss": 2.6736,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.986182257912564e-06,
+      "loss": 2.6591,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.98590741395118e-06,
+      "loss": 2.6341,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.985629867284769e-06,
+      "loss": 2.6943,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.985349618063778e-06,
+      "loss": 2.646,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.985066666440123e-06,
+      "loss": 2.7127,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.984781012567182e-06,
+      "loss": 2.6116,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.9844926565998e-06,
+      "loss": 2.6784,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.984201598694286e-06,
+      "loss": 2.688,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.983907839008413e-06,
+      "loss": 2.6809,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.98361137770142e-06,
+      "loss": 2.651,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.983312214934005e-06,
+      "loss": 2.7419,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.983010350868341e-06,
+      "loss": 2.6689,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.982705785668053e-06,
+      "loss": 2.6337,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.982398519498245e-06,
+      "loss": 2.7073,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.982088552525466e-06,
+      "loss": 2.6153,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.981775884917746e-06,
+      "loss": 2.6824,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.981460516844573e-06,
+      "loss": 2.7,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.981142448476893e-06,
+      "loss": 2.6456,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.980821679987125e-06,
+      "loss": 2.6175,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.980498211549145e-06,
+      "loss": 2.6099,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.980172043338298e-06,
+      "loss": 2.7185,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97984317553139e-06,
+      "loss": 2.6752,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.979511608306686e-06,
+      "loss": 2.6065,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97917734184392e-06,
+      "loss": 2.6889,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97884037632429e-06,
+      "loss": 2.6757,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.978500711930453e-06,
+      "loss": 2.7039,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.978158348846527e-06,
+      "loss": 2.7029,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.977813287258102e-06,
+      "loss": 2.7298,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.977465527352221e-06,
+      "loss": 2.621,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.977115069317396e-06,
+      "loss": 2.6071,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.976761913343598e-06,
+      "loss": 2.63,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.976406059622263e-06,
+      "loss": 2.6824,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.976047508346287e-06,
+      "loss": 2.6667,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97568625971003e-06,
+      "loss": 2.7849,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.975322313909315e-06,
+      "loss": 2.6782,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974955671141425e-06,
+      "loss": 2.6662,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974586331605104e-06,
+      "loss": 2.7248,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974214295500561e-06,
+      "loss": 2.6697,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973839563029464e-06,
+      "loss": 2.615,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973462134394945e-06,
+      "loss": 2.6611,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973082009801596e-06,
+      "loss": 2.7345,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97269918945547e-06,
+      "loss": 2.6808,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.972313673564083e-06,
+      "loss": 2.7388,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.97192546233641e-06,
+      "loss": 2.6543,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.97153455598289e-06,
+      "loss": 2.6799,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971140954715419e-06,
+      "loss": 2.6842,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.970744658747358e-06,
+      "loss": 2.6772,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.970345668293525e-06,
+      "loss": 2.6632,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.9699439835702e-06,
+      "loss": 2.6315,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.969539604795126e-06,
+      "loss": 2.6682,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.969132532187504e-06,
+      "loss": 2.5942,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.968722765967993e-06,
+      "loss": 2.6941,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.968310306358715e-06,
+      "loss": 2.6803,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.967895153583253e-06,
+      "loss": 2.5683,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.967477307866648e-06,
+      "loss": 2.7074,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.9670567694354e-06,
+      "loss": 2.6348,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.966633538517472e-06,
+      "loss": 2.6284,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.966207615342282e-06,
+      "loss": 2.7111,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96577900014071e-06,
+      "loss": 2.6624,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.965347693145095e-06,
+      "loss": 2.6648,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.964913694589238e-06,
+      "loss": 2.6951,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96447700470839e-06,
+      "loss": 2.6774,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.964037623739273e-06,
+      "loss": 2.637,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96359555192006e-06,
+      "loss": 2.6553,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.963150789490384e-06,
+      "loss": 2.6878,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.962703336691334e-06,
+      "loss": 2.6946,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.962253193765464e-06,
+      "loss": 2.6363,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.961800360956782e-06,
+      "loss": 2.6716,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.961344838510753e-06,
+      "loss": 2.6444,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.960886626674302e-06,
+      "loss": 2.683,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.960425725695812e-06,
+      "loss": 2.7478,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.959962135825123e-06,
+      "loss": 2.6451,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.959495857313533e-06,
+      "loss": 2.7052,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.959026890413794e-06,
+      "loss": 2.6571,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.958555235380122e-06,
+      "loss": 2.7037,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.958080892468187e-06,
+      "loss": 2.7053,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.957603861935112e-06,
+      "loss": 2.7473,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.957124144039483e-06,
+      "loss": 2.6903,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.956641739041338e-06,
+      "loss": 2.7139,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.956156647202175e-06,
+      "loss": 2.6698,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.95566886878495e-06,
+      "loss": 2.6818,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.955178404054066e-06,
+      "loss": 2.6803,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.954685253275393e-06,
+      "loss": 2.623,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.954189416716253e-06,
+      "loss": 2.6657,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.953690894645423e-06,
+      "loss": 2.6601,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.953189687333133e-06,
+      "loss": 2.6954,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.952685795051078e-06,
+      "loss": 2.6698,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.952179218072397e-06,
+      "loss": 2.7103,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.951669956671691e-06,
+      "loss": 2.6213,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.951158011125018e-06,
+      "loss": 2.6367,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.950643381709882e-06,
+      "loss": 2.685,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.950126068705252e-06,
+      "loss": 2.6444,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.949606072391544e-06,
+      "loss": 2.6503,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.949083393050636e-06,
+      "loss": 2.6372,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.948558030965853e-06,
+      "loss": 2.6052,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.948029986421976e-06,
+      "loss": 2.6796,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.947499259705246e-06,
+      "loss": 2.6197,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.946965851103351e-06,
+      "loss": 2.677,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.946429760905436e-06,
+      "loss": 2.6577,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.945890989402096e-06,
+      "loss": 2.6091,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.945349536885386e-06,
+      "loss": 2.6555,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.944805403648808e-06,
+      "loss": 2.7048,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94425858998732e-06,
+      "loss": 2.6705,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.943709096197334e-06,
+      "loss": 2.6543,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.943156922576714e-06,
+      "loss": 2.6948,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.942602069424773e-06,
+      "loss": 2.7401,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.942044537042281e-06,
+      "loss": 2.6685,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94148432573146e-06,
+      "loss": 2.7075,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.940921435795982e-06,
+      "loss": 2.638,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.940355867540973e-06,
+      "loss": 2.6699,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.939787621273009e-06,
+      "loss": 2.6169,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.93921669730012e-06,
+      "loss": 2.6582,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.938643095931783e-06,
+      "loss": 2.6609,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.938066817478934e-06,
+      "loss": 2.6947,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.93748786225395e-06,
+      "loss": 2.6981,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.936906230570669e-06,
+      "loss": 2.7042,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.936321922744375e-06,
+      "loss": 2.6518,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.935734939091802e-06,
+      "loss": 2.66,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.935145279931134e-06,
+      "loss": 2.5882,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.93455294558201e-06,
+      "loss": 2.6442,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.933957936365515e-06,
+      "loss": 2.6967,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.933360252604183e-06,
+      "loss": 2.646,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.932759894622001e-06,
+      "loss": 2.6186,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.932156862744406e-06,
+      "loss": 2.707,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.93155115729828e-06,
+      "loss": 2.6289,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.93094277861196e-06,
+      "loss": 2.6607,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.930331727015228e-06,
+      "loss": 2.6396,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.929718002839312e-06,
+      "loss": 2.7165,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.9291016064169e-06,
+      "loss": 2.6432,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.928482538082115e-06,
+      "loss": 2.6646,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.92786079817054e-06,
+      "loss": 2.6502,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.927236387019197e-06,
+      "loss": 2.7286,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.92660930496656e-06,
+      "loss": 2.6722,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.925979552352552e-06,
+      "loss": 2.6559,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.92534712951854e-06,
+      "loss": 2.7649,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.924712036807345e-06,
+      "loss": 2.6487,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.924074274563229e-06,
+      "loss": 2.7249,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.9234338431319e-06,
+      "loss": 2.6991,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.922790742860521e-06,
+      "loss": 2.6463,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.922144974097693e-06,
+      "loss": 2.7253,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.921496537193468e-06,
+      "loss": 2.6831,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.920845432499344e-06,
+      "loss": 2.6772,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.920191660368264e-06,
+      "loss": 2.6567,
+      "step": 743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.919535221154617e-06,
+      "loss": 2.6318,
+      "step": 744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.91887611521424e-06,
+      "loss": 2.6613,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.918214342904412e-06,
+      "loss": 2.7093,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.917549904583859e-06,
+      "loss": 2.6743,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.916882800612756e-06,
+      "loss": 2.6549,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.916213031352713e-06,
+      "loss": 2.6564,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.915540597166798e-06,
+      "loss": 2.7102,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.91486549841951e-06,
+      "loss": 2.6298,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.914187735476804e-06,
+      "loss": 2.6669,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.91350730870607e-06,
+      "loss": 2.6795,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.912824218476152e-06,
+      "loss": 2.5918,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.912138465157325e-06,
+      "loss": 2.7281,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.911450049121319e-06,
+      "loss": 2.6799,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.910758970741301e-06,
+      "loss": 2.7054,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.910065230391885e-06,
+      "loss": 2.6397,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.909368828449125e-06,
+      "loss": 2.707,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.908669765290518e-06,
+      "loss": 2.6104,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.907968041295006e-06,
+      "loss": 2.6767,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.90726365684297e-06,
+      "loss": 2.6875,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.906556612316234e-06,
+      "loss": 2.6785,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.905846908098067e-06,
+      "loss": 2.6877,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.905134544573178e-06,
+      "loss": 2.7249,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.904419522127715e-06,
+      "loss": 2.6614,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.903701841149271e-06,
+      "loss": 2.6747,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.902981502026876e-06,
+      "loss": 2.7307,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.902258505151007e-06,
+      "loss": 2.6764,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.901532850913575e-06,
+      "loss": 2.6564,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.900804539707938e-06,
+      "loss": 2.6212,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.900073571928887e-06,
+      "loss": 2.6873,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.899339947972658e-06,
+      "loss": 2.7535,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.898603668236927e-06,
+      "loss": 2.6298,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.897864733120806e-06,
+      "loss": 2.6797,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.897123143024852e-06,
+      "loss": 2.6643,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.896378898351055e-06,
+      "loss": 2.7279,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.895631999502849e-06,
+      "loss": 2.6159,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.894882446885103e-06,
+      "loss": 2.6995,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.894130240904127e-06,
+      "loss": 2.6798,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.893375381967667e-06,
+      "loss": 2.7244,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.892617870484908e-06,
+      "loss": 2.6988,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.891857706866477e-06,
+      "loss": 2.7114,
+      "step": 783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.891094891524433e-06,
+      "loss": 2.6059,
+      "step": 784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.890329424872274e-06,
+      "loss": 2.7387,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.889561307324935e-06,
+      "loss": 2.6472,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.888790539298791e-06,
+      "loss": 2.6424,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.888017121211649e-06,
+      "loss": 2.753,
+      "step": 788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.887241053482756e-06,
+      "loss": 2.6747,
+      "step": 789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.886462336532794e-06,
+      "loss": 2.6989,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.88568097078388e-06,
+      "loss": 2.6322,
+      "step": 791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.88489695665957e-06,
+      "loss": 2.579,
+      "step": 792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.884110294584852e-06,
+      "loss": 2.6055,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.883320984986153e-06,
+      "loss": 2.6401,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.88252902829133e-06,
+      "loss": 2.6282,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.88173442492968e-06,
+      "loss": 2.6711,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.880937175331932e-06,
+      "loss": 2.7433,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.880137279930254e-06,
+      "loss": 2.679,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.879334739158238e-06,
+      "loss": 2.6852,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.87852955345092e-06,
+      "loss": 2.6377,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.877721723244765e-06,
+      "loss": 2.6461,
+      "step": 801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.876911248977675e-06,
+      "loss": 2.7077,
+      "step": 802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.876098131088979e-06,
+      "loss": 2.6791,
+      "step": 803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.875282370019446e-06,
+      "loss": 2.6315,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.874463966211271e-06,
+      "loss": 2.6531,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.87364292010809e-06,
+      "loss": 2.646,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.872819232154965e-06,
+      "loss": 2.6751,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.871992902798389e-06,
+      "loss": 2.644,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.871163932486291e-06,
+      "loss": 2.6786,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.87033232166803e-06,
+      "loss": 2.7003,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.869498070794396e-06,
+      "loss": 2.6262,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.868661180317613e-06,
+      "loss": 2.6974,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.86782165069133e-06,
+      "loss": 2.6105,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.86697948237063e-06,
+      "loss": 2.7062,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.866134675812027e-06,
+      "loss": 2.6741,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.865287231473465e-06,
+      "loss": 2.7176,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.864437149814317e-06,
+      "loss": 2.6928,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.863584431295385e-06,
+      "loss": 2.6548,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.862729076378901e-06,
+      "loss": 2.6644,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.861871085528527e-06,
+      "loss": 2.6325,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.861010459209353e-06,
+      "loss": 2.699,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.860147197887897e-06,
+      "loss": 2.6947,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.859281302032107e-06,
+      "loss": 2.7095,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.858412772111358e-06,
+      "loss": 2.6345,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.857541608596453e-06,
+      "loss": 2.7162,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.85666781195962e-06,
+      "loss": 2.6549,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.85579138267452e-06,
+      "loss": 2.6238,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.85491232121624e-06,
+      "loss": 2.6419,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.854030628061286e-06,
+      "loss": 2.6017,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.853146303687602e-06,
+      "loss": 2.7023,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.85225934857455e-06,
+      "loss": 2.5757,
+      "step": 831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.851369763202919e-06,
+      "loss": 2.6746,
+      "step": 832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.850477548054927e-06,
+      "loss": 2.6571,
+      "step": 833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.849582703614218e-06,
+      "loss": 2.6631,
+      "step": 834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.848685230365857e-06,
+      "loss": 2.5819,
+      "step": 835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.847785128796337e-06,
+      "loss": 2.6669,
+      "step": 836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.846882399393575e-06,
+      "loss": 2.7268,
+      "step": 837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.845977042646912e-06,
+      "loss": 2.637,
+      "step": 838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.845069059047114e-06,
+      "loss": 2.6914,
+      "step": 839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.844158449086372e-06,
+      "loss": 2.648,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.843245213258297e-06,
+      "loss": 2.661,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.842329352057928e-06,
+      "loss": 2.653,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.841410865981722e-06,
+      "loss": 2.5952,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.840489755527565e-06,
+      "loss": 2.5974,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.839566021194758e-06,
+      "loss": 2.697,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.838639663484032e-06,
+      "loss": 2.6273,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.837710682897537e-06,
+      "loss": 2.6454,
+      "step": 847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.836779079938842e-06,
+      "loss": 2.7148,
+      "step": 848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.835844855112942e-06,
+      "loss": 2.6381,
+      "step": 849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.83490800892625e-06,
+      "loss": 2.6652,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.8339685418866e-06,
+      "loss": 2.6381,
+      "step": 851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.83302645450325e-06,
+      "loss": 2.6804,
+      "step": 852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.832081747286876e-06,
+      "loss": 2.6764,
+      "step": 853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.831134420749572e-06,
+      "loss": 2.6194,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.830184475404859e-06,
+      "loss": 2.7554,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.829231911767666e-06,
+      "loss": 2.6379,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.828276730354353e-06,
+      "loss": 2.7145,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.827318931682695e-06,
+      "loss": 2.7354,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.826358516271878e-06,
+      "loss": 2.629,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.825395484642522e-06,
+      "loss": 2.7006,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.82442983731665e-06,
+      "loss": 2.6887,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.823461574817714e-06,
+      "loss": 2.6327,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.822490697670575e-06,
+      "loss": 2.6737,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.821517206401518e-06,
+      "loss": 2.692,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.82054110153824e-06,
+      "loss": 2.637,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.819562383609857e-06,
+      "loss": 2.6317,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.818581053146905e-06,
+      "loss": 2.652,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.81759711068133e-06,
+      "loss": 2.6783,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.816610556746495e-06,
+      "loss": 2.705,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.815621391877183e-06,
+      "loss": 2.695,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.814629616609587e-06,
+      "loss": 2.7196,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.81363523148132e-06,
+      "loss": 2.6247,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.812638237031407e-06,
+      "loss": 2.7296,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.811638633800287e-06,
+      "loss": 2.7194,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.810636422329812e-06,
+      "loss": 2.7059,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.80963160316325e-06,
+      "loss": 2.6474,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.808624176845286e-06,
+      "loss": 2.7137,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.807614143922012e-06,
+      "loss": 2.6225,
+      "step": 878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.80660150494093e-06,
+      "loss": 2.6435,
+      "step": 879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.805586260450969e-06,
+      "loss": 2.6129,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.804568411002458e-06,
+      "loss": 2.6598,
+      "step": 881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.80354795714714e-06,
+      "loss": 2.7743,
+      "step": 882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.802524899438171e-06,
+      "loss": 2.6063,
+      "step": 883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.801499238430118e-06,
+      "loss": 2.6964,
+      "step": 884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.800470974678966e-06,
+      "loss": 2.6548,
+      "step": 885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.799440108742097e-06,
+      "loss": 2.624,
+      "step": 886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.798406641178317e-06,
+      "loss": 2.5838,
+      "step": 887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.797370572547835e-06,
+      "loss": 2.6342,
+      "step": 888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.796331903412268e-06,
+      "loss": 2.7021,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.795290634334651e-06,
+      "loss": 2.7334,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.794246765879421e-06,
+      "loss": 2.6683,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.793200298612426e-06,
+      "loss": 2.5763,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.792151233100926e-06,
+      "loss": 2.6234,
+      "step": 893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.791099569913585e-06,
+      "loss": 2.6509,
+      "step": 894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.790045309620476e-06,
+      "loss": 2.6701,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.78898845279308e-06,
+      "loss": 2.6649,
+      "step": 896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.78792900000429e-06,
+      "loss": 2.6946,
+      "step": 897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.7868669518284e-06,
+      "loss": 2.6601,
+      "step": 898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.785802308841112e-06,
+      "loss": 2.6969,
+      "step": 899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.784735071619538e-06,
+      "loss": 2.6373,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.783665240742193e-06,
+      "loss": 2.7593,
+      "step": 901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.782592816788999e-06,
+      "loss": 2.6462,
+      "step": 902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.781517800341282e-06,
+      "loss": 2.7157,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.780440191981779e-06,
+      "loss": 2.6829,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.779359992294625e-06,
+      "loss": 2.6541,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.778277201865363e-06,
+      "loss": 2.7344,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.777191821280939e-06,
+      "loss": 2.6226,
+      "step": 907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.776103851129706e-06,
+      "loss": 2.6567,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.775013292001418e-06,
+      "loss": 2.6588,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.773920144487232e-06,
+      "loss": 2.7051,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.77282440917971e-06,
+      "loss": 2.6446,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.771726086672815e-06,
+      "loss": 2.6279,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.770625177561917e-06,
+      "loss": 2.6293,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.76952168244378e-06,
+      "loss": 2.6959,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.768415601916577e-06,
+      "loss": 2.5996,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.767306936579878e-06,
+      "loss": 2.6106,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.766195687034659e-06,
+      "loss": 2.6407,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.765081853883291e-06,
+      "loss": 2.6855,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.763965437729549e-06,
+      "loss": 2.727,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.762846439178606e-06,
+      "loss": 2.6026,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.761724858837038e-06,
+      "loss": 2.6329,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.760600697312819e-06,
+      "loss": 2.6748,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.75947395521532e-06,
+      "loss": 2.6644,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.758344633155315e-06,
+      "loss": 2.6896,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.757212731744973e-06,
+      "loss": 2.6596,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.756078251597865e-06,
+      "loss": 2.6757,
+      "step": 926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.754941193328952e-06,
+      "loss": 2.6393,
+      "step": 927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.753801557554603e-06,
+      "loss": 2.6688,
+      "step": 928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.752659344892577e-06,
+      "loss": 2.7376,
+      "step": 929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.751514555962031e-06,
+      "loss": 2.6297,
+      "step": 930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.75036719138352e-06,
+      "loss": 2.6627,
+      "step": 931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.749217251778995e-06,
+      "loss": 2.5834,
+      "step": 932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.748064737771803e-06,
+      "loss": 2.7172,
+      "step": 933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.746909649986686e-06,
+      "loss": 2.7633,
+      "step": 934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.74575198904978e-06,
+      "loss": 2.6956,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.744591755588616e-06,
+      "loss": 2.6686,
+      "step": 936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.743428950232123e-06,
+      "loss": 2.6475,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.742263573610619e-06,
+      "loss": 2.712,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.741095626355821e-06,
+      "loss": 2.7092,
+      "step": 939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.739925109100833e-06,
+      "loss": 2.7283,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.73875202248016e-06,
+      "loss": 2.7407,
+      "step": 941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.737576367129694e-06,
+      "loss": 2.6772,
+      "step": 942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.736398143686722e-06,
+      "loss": 2.6633,
+      "step": 943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.735217352789922e-06,
+      "loss": 2.5841,
+      "step": 944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.734033995079359e-06,
+      "loss": 2.6982,
+      "step": 945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.732848071196503e-06,
+      "loss": 2.6602,
+      "step": 946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.7316595817842e-06,
+      "loss": 2.6849,
+      "step": 947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.730468527486695e-06,
+      "loss": 2.6372,
+      "step": 948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.729274908949622e-06,
+      "loss": 2.7147,
+      "step": 949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.728078726820003e-06,
+      "loss": 2.7381,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.726879981746254e-06,
+      "loss": 2.6167,
+      "step": 951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.725678674378173e-06,
+      "loss": 2.6673,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.724474805366955e-06,
+      "loss": 2.6778,
+      "step": 953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.723268375365178e-06,
+      "loss": 2.6643,
+      "step": 954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.722059385026811e-06,
+      "loss": 2.6372,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.72084783500721e-06,
+      "loss": 2.6668,
+      "step": 956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.719633725963116e-06,
+      "loss": 2.6497,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.718417058552663e-06,
+      "loss": 2.6686,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.717197833435367e-06,
+      "loss": 2.6412,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.715976051272132e-06,
+      "loss": 2.684,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.71475171272525e-06,
+      "loss": 2.7038,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.713524818458392e-06,
+      "loss": 2.6615,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.712295369136625e-06,
+      "loss": 2.7222,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.71106336542639e-06,
+      "loss": 2.6829,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.709828807995521e-06,
+      "loss": 2.6574,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.708591697513232e-06,
+      "loss": 2.6881,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.707352034650124e-06,
+      "loss": 2.6428,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.706109820078175e-06,
+      "loss": 2.6831,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.704865054470756e-06,
+      "loss": 2.6309,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.703617738502612e-06,
+      "loss": 2.6717,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.702367872849876e-06,
+      "loss": 2.6364,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.70111545819006e-06,
+      "loss": 2.6372,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.69986049520206e-06,
+      "loss": 2.7017,
+      "step": 973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.698602984566152e-06,
+      "loss": 2.6589,
+      "step": 974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.697342926963993e-06,
+      "loss": 2.6448,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.696080323078621e-06,
+      "loss": 2.5887,
+      "step": 976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.694815173594456e-06,
+      "loss": 2.6281,
+      "step": 977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.693547479197292e-06,
+      "loss": 2.5573,
+      "step": 978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.692277240574312e-06,
+      "loss": 2.7424,
+      "step": 979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.691004458414069e-06,
+      "loss": 2.6666,
+      "step": 980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.689729133406497e-06,
+      "loss": 2.636,
+      "step": 981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.688451266242916e-06,
+      "loss": 2.6554,
+      "step": 982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.687170857616011e-06,
+      "loss": 2.7156,
+      "step": 983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.685887908219857e-06,
+      "loss": 2.6373,
+      "step": 984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.6846024187499e-06,
+      "loss": 2.6675,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.68331438990296e-06,
+      "loss": 2.6659,
+      "step": 986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.682023822377242e-06,
+      "loss": 2.6141,
+      "step": 987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.680730716872319e-06,
+      "loss": 2.6629,
+      "step": 988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.679435074089145e-06,
+      "loss": 2.696,
+      "step": 989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.678136894730048e-06,
+      "loss": 2.6676,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.67683617949873e-06,
+      "loss": 2.6932,
+      "step": 991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.675532929100267e-06,
+      "loss": 2.671,
+      "step": 992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.67422714424111e-06,
+      "loss": 2.5941,
+      "step": 993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.672918825629088e-06,
+      "loss": 2.6831,
+      "step": 994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.671607973973396e-06,
+      "loss": 2.7121,
+      "step": 995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.670294589984607e-06,
+      "loss": 2.5796,
+      "step": 996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.668978674374668e-06,
+      "loss": 2.666,
+      "step": 997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.667660227856892e-06,
+      "loss": 2.6841,
+      "step": 998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.666339251145968e-06,
+      "loss": 2.6162,
+      "step": 999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.66501574495796e-06,
+      "loss": 2.7382,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.663689710010296e-06,
+      "loss": 2.6679,
+      "step": 1001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.66236114702178e-06,
+      "loss": 2.6478,
+      "step": 1002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.661030056712584e-06,
+      "loss": 2.7375,
+      "step": 1003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.659696439804249e-06,
+      "loss": 2.6908,
+      "step": 1004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.65836029701969e-06,
+      "loss": 2.6058,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.657021629083188e-06,
+      "loss": 2.7222,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.655680436720389e-06,
+      "loss": 2.6715,
+      "step": 1007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.654336720658317e-06,
+      "loss": 2.7342,
+      "step": 1008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.652990481625358e-06,
+      "loss": 2.7161,
+      "step": 1009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.651641720351262e-06,
+      "loss": 2.5814,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.650290437567153e-06,
+      "loss": 2.6689,
+      "step": 1011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.648936634005522e-06,
+      "loss": 2.6103,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.647580310400217e-06,
+      "loss": 2.6463,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.646221467486466e-06,
+      "loss": 2.68,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.644860106000851e-06,
+      "loss": 2.646,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.643496226681325e-06,
+      "loss": 2.6616,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.642129830267205e-06,
+      "loss": 2.7115,
+      "step": 1017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.640760917499171e-06,
+      "loss": 2.69,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.639389489119268e-06,
+      "loss": 2.6554,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.638015545870906e-06,
+      "loss": 2.6632,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.636639088498853e-06,
+      "loss": 2.6849,
+      "step": 1021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.635260117749248e-06,
+      "loss": 2.6168,
+      "step": 1022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.633878634369587e-06,
+      "loss": 2.6728,
+      "step": 1023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.632494639108727e-06,
+      "loss": 2.7434,
+      "step": 1024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.631108132716892e-06,
+      "loss": 2.7254,
+      "step": 1025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.62971911594566e-06,
+      "loss": 2.6336,
+      "step": 1026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.628327589547977e-06,
+      "loss": 2.6592,
+      "step": 1027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.626933554278143e-06,
+      "loss": 2.7374,
+      "step": 1028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.625537010891826e-06,
+      "loss": 2.5934,
+      "step": 1029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.624137960146043e-06,
+      "loss": 2.6488,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.622736402799178e-06,
+      "loss": 2.6303,
+      "step": 1031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.621332339610974e-06,
+      "loss": 2.6611,
+      "step": 1032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.619925771342526e-06,
+      "loss": 2.6136,
+      "step": 1033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.618516698756294e-06,
+      "loss": 2.6279,
+      "step": 1034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.61710512261609e-06,
+      "loss": 2.6247,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.615691043687087e-06,
+      "loss": 2.661,
+      "step": 1036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.614274462735813e-06,
+      "loss": 2.6896,
+      "step": 1037
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.612855380530152e-06,
+      "loss": 2.6428,
+      "step": 1038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.611433797839342e-06,
+      "loss": 2.6502,
+      "step": 1039
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.610009715433983e-06,
+      "loss": 2.752,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.608583134086022e-06,
+      "loss": 2.667,
+      "step": 1041
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.607154054568769e-06,
+      "loss": 2.6508,
+      "step": 1042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.605722477656877e-06,
+      "loss": 2.6586,
+      "step": 1043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.604288404126362e-06,
+      "loss": 2.7044,
+      "step": 1044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.602851834754593e-06,
+      "loss": 2.5892,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.601412770320289e-06,
+      "loss": 2.6598,
+      "step": 1046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.599971211603518e-06,
+      "loss": 2.7162,
+      "step": 1047
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.598527159385707e-06,
+      "loss": 2.7152,
+      "step": 1048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.597080614449632e-06,
+      "loss": 2.7647,
+      "step": 1049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.595631577579419e-06,
+      "loss": 2.6483,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.594180049560546e-06,
+      "loss": 2.6793,
+      "step": 1051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.59272603117984e-06,
+      "loss": 2.6876,
+      "step": 1052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.591269523225482e-06,
+      "loss": 2.6711,
+      "step": 1053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.589810526486996e-06,
+      "loss": 2.6958,
+      "step": 1054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.588349041755262e-06,
+      "loss": 2.7,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.586885069822502e-06,
+      "loss": 2.5865,
+      "step": 1056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.585418611482293e-06,
+      "loss": 2.7251,
+      "step": 1057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.583949667529553e-06,
+      "loss": 2.6949,
+      "step": 1058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.582478238760553e-06,
+      "loss": 2.6492,
+      "step": 1059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.58100432597291e-06,
+      "loss": 2.6714,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.579527929965581e-06,
+      "loss": 2.6474,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.57804905153888e-06,
+      "loss": 2.6433,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.576567691494456e-06,
+      "loss": 2.674,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.575083850635312e-06,
+      "loss": 2.6849,
+      "step": 1064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.57359752976579e-06,
+      "loss": 2.6469,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.572108729691579e-06,
+      "loss": 2.6795,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.57061745121971e-06,
+      "loss": 2.6828,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.569123695158562e-06,
+      "loss": 2.6437,
+      "step": 1068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.567627462317847e-06,
+      "loss": 2.643,
+      "step": 1069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.566128753508632e-06,
+      "loss": 2.7366,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.56462756954332e-06,
+      "loss": 2.5876,
+      "step": 1071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.563123911235652e-06,
+      "loss": 2.6645,
+      "step": 1072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.561617779400719e-06,
+      "loss": 2.6117,
+      "step": 1073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.560109174854944e-06,
+      "loss": 2.6408,
+      "step": 1074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.558598098416097e-06,
+      "loss": 2.712,
+      "step": 1075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.557084550903287e-06,
+      "loss": 2.7234,
+      "step": 1076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.555568533136956e-06,
+      "loss": 2.6665,
+      "step": 1077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.554050045938893e-06,
+      "loss": 2.7037,
+      "step": 1078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.552529090132223e-06,
+      "loss": 2.667,
+      "step": 1079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.551005666541408e-06,
+      "loss": 2.6959,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.54947977599225e-06,
+      "loss": 2.6319,
+      "step": 1081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.547951419311885e-06,
+      "loss": 2.6353,
+      "step": 1082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.546420597328787e-06,
+      "loss": 2.6748,
+      "step": 1083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.544887310872767e-06,
+      "loss": 2.6521,
+      "step": 1084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.543351560774976e-06,
+      "loss": 2.6827,
+      "step": 1085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.54181334786789e-06,
+      "loss": 2.6457,
+      "step": 1086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.54027267298533e-06,
+      "loss": 2.6505,
+      "step": 1087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.538729536962446e-06,
+      "loss": 2.6202,
+      "step": 1088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.537183940635725e-06,
+      "loss": 2.6764,
+      "step": 1089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.535635884842988e-06,
+      "loss": 2.6935,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.534085370423384e-06,
+      "loss": 2.608,
+      "step": 1091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.532532398217403e-06,
+      "loss": 2.6708,
+      "step": 1092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.530976969066861e-06,
+      "loss": 2.695,
+      "step": 1093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.529419083814905e-06,
+      "loss": 2.664,
+      "step": 1094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.52785874330602e-06,
+      "loss": 2.6161,
+      "step": 1095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.526295948386017e-06,
+      "loss": 2.7192,
+      "step": 1096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.524730699902036e-06,
+      "loss": 2.7253,
+      "step": 1097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.523162998702552e-06,
+      "loss": 2.6817,
+      "step": 1098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.521592845637366e-06,
+      "loss": 2.6516,
+      "step": 1099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.52002024155761e-06,
+      "loss": 2.6071,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.51844518731574e-06,
+      "loss": 2.737,
+      "step": 1101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.516867683765551e-06,
+      "loss": 2.7087,
+      "step": 1102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.515287731762152e-06,
+      "loss": 2.6621,
+      "step": 1103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.513705332161989e-06,
+      "loss": 2.6202,
+      "step": 1104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.51212048582283e-06,
+      "loss": 2.7252,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.51053319360377e-06,
+      "loss": 2.6479,
+      "step": 1106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.508943456365232e-06,
+      "loss": 2.6663,
+      "step": 1107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.507351274968964e-06,
+      "loss": 2.6522,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.505756650278037e-06,
+      "loss": 2.6501,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.504159583156847e-06,
+      "loss": 2.664,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.502560074471115e-06,
+      "loss": 2.6754,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.500958125087882e-06,
+      "loss": 2.661,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.49935373587552e-06,
+      "loss": 2.6826,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.497746907703715e-06,
+      "loss": 2.7433,
+      "step": 1114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.496137641443479e-06,
+      "loss": 2.6137,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.494525937967144e-06,
+      "loss": 2.6297,
+      "step": 1116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.492911798148365e-06,
+      "loss": 2.6527,
+      "step": 1117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.491295222862119e-06,
+      "loss": 2.7038,
+      "step": 1118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.4896762129847e-06,
+      "loss": 2.7373,
+      "step": 1119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.48805476939372e-06,
+      "loss": 2.6565,
+      "step": 1120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.486430892968115e-06,
+      "loss": 2.7096,
+      "step": 1121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.484804584588139e-06,
+      "loss": 2.6116,
+      "step": 1122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.483175845135362e-06,
+      "loss": 2.6941,
+      "step": 1123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.481544675492673e-06,
+      "loss": 2.7092,
+      "step": 1124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.479911076544278e-06,
+      "loss": 2.6114,
+      "step": 1125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.478275049175701e-06,
+      "loss": 2.7135,
+      "step": 1126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.47663659427378e-06,
+      "loss": 2.7793,
+      "step": 1127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.474995712726671e-06,
+      "loss": 2.621,
+      "step": 1128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.473352405423845e-06,
+      "loss": 2.7026,
+      "step": 1129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.471706673256087e-06,
+      "loss": 2.6816,
+      "step": 1130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.470058517115498e-06,
+      "loss": 2.7721,
+      "step": 1131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.468407937895491e-06,
+      "loss": 2.6463,
+      "step": 1132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.466754936490797e-06,
+      "loss": 2.67,
+      "step": 1133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.46509951379745e-06,
+      "loss": 2.671,
+      "step": 1134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.46344167071281e-06,
+      "loss": 2.7159,
+      "step": 1135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.461781408135537e-06,
+      "loss": 2.7143,
+      "step": 1136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.46011872696561e-06,
+      "loss": 2.6297,
+      "step": 1137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.458453628104316e-06,
+      "loss": 2.5957,
+      "step": 1138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.456786112454255e-06,
+      "loss": 2.6201,
+      "step": 1139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.455116180919333e-06,
+      "loss": 2.6897,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.453443834404768e-06,
+      "loss": 2.5128,
+      "step": 1141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.451769073817087e-06,
+      "loss": 2.6699,
+      "step": 1142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.450091900064125e-06,
+      "loss": 2.6637,
+      "step": 1143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.44841231405503e-06,
+      "loss": 2.7282,
+      "step": 1144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.446730316700246e-06,
+      "loss": 2.6222,
+      "step": 1145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.445045908911536e-06,
+      "loss": 2.7076,
+      "step": 1146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.443359091601966e-06,
+      "loss": 2.6618,
+      "step": 1147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.441669865685904e-06,
+      "loss": 2.631,
+      "step": 1148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.439978232079027e-06,
+      "loss": 2.6715,
+      "step": 1149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.43828419169832e-06,
+      "loss": 2.6198,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.436587745462065e-06,
+      "loss": 2.6503,
+      "step": 1151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.434888894289856e-06,
+      "loss": 2.6579,
+      "step": 1152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.433187639102587e-06,
+      "loss": 2.7277,
+      "step": 1153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.431483980822453e-06,
+      "loss": 2.6244,
+      "step": 1154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.429777920372957e-06,
+      "loss": 2.7109,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.428069458678899e-06,
+      "loss": 2.659,
+      "step": 1156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.426358596666385e-06,
+      "loss": 2.703,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.424645335262818e-06,
+      "loss": 2.6038,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.422929675396903e-06,
+      "loss": 2.6736,
+      "step": 1159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.421211617998649e-06,
+      "loss": 2.6915,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.419491163999359e-06,
+      "loss": 2.6409,
+      "step": 1161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.417768314331637e-06,
+      "loss": 2.6884,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.416043069929389e-06,
+      "loss": 2.6652,
+      "step": 1163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.414315431727813e-06,
+      "loss": 2.6841,
+      "step": 1164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.412585400663407e-06,
+      "loss": 2.6836,
+      "step": 1165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.410852977673971e-06,
+      "loss": 2.568,
+      "step": 1166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.409118163698596e-06,
+      "loss": 2.6689,
+      "step": 1167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.40738095967767e-06,
+      "loss": 2.6412,
+      "step": 1168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.405641366552878e-06,
+      "loss": 2.6786,
+      "step": 1169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.403899385267197e-06,
+      "loss": 2.6536,
+      "step": 1170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.402155016764903e-06,
+      "loss": 2.6887,
+      "step": 1171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.400408261991564e-06,
+      "loss": 2.654,
+      "step": 1172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.39865912189404e-06,
+      "loss": 2.6279,
+      "step": 1173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.396907597420483e-06,
+      "loss": 2.6523,
+      "step": 1174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.395153689520346e-06,
+      "loss": 2.6254,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.393397399144363e-06,
+      "loss": 2.6219,
+      "step": 1176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.391638727244566e-06,
+      "loss": 2.6271,
+      "step": 1177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.389877674774276e-06,
+      "loss": 2.7017,
+      "step": 1178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.388114242688105e-06,
+      "loss": 2.5923,
+      "step": 1179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.386348431941953e-06,
+      "loss": 2.6387,
+      "step": 1180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.384580243493014e-06,
+      "loss": 2.6522,
+      "step": 1181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.382809678299763e-06,
+      "loss": 2.6462,
+      "step": 1182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.381036737321973e-06,
+      "loss": 2.6988,
+      "step": 1183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.379261421520698e-06,
+      "loss": 2.6562,
+      "step": 1184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.377483731858282e-06,
+      "loss": 2.639,
+      "step": 1185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.375703669298354e-06,
+      "loss": 2.7192,
+      "step": 1186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.373921234805833e-06,
+      "loss": 2.7026,
+      "step": 1187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.372136429346918e-06,
+      "loss": 2.5981,
+      "step": 1188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.370349253889098e-06,
+      "loss": 2.63,
+      "step": 1189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.368559709401145e-06,
+      "loss": 2.6356,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.366767796853116e-06,
+      "loss": 2.5826,
+      "step": 1191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.36497351721635e-06,
+      "loss": 2.713,
+      "step": 1192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.36317687146347e-06,
+      "loss": 2.7351,
+      "step": 1193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.36137786056838e-06,
+      "loss": 2.5918,
+      "step": 1194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.35957648550627e-06,
+      "loss": 2.652,
+      "step": 1195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.357772747253612e-06,
+      "loss": 2.6813,
+      "step": 1196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.355966646788152e-06,
+      "loss": 2.6646,
+      "step": 1197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.354158185088921e-06,
+      "loss": 2.6356,
+      "step": 1198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.352347363136232e-06,
+      "loss": 2.6227,
+      "step": 1199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.350534181911672e-06,
+      "loss": 2.641,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.348718642398114e-06,
+      "loss": 2.651,
+      "step": 1201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.346900745579703e-06,
+      "loss": 2.6348,
+      "step": 1202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.345080492441864e-06,
+      "loss": 2.7449,
+      "step": 1203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.3432578839713e-06,
+      "loss": 2.6832,
+      "step": 1204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.34143292115599e-06,
+      "loss": 2.6087,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.339605604985192e-06,
+      "loss": 2.6602,
+      "step": 1206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.337775936449432e-06,
+      "loss": 2.7243,
+      "step": 1207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.335943916540521e-06,
+      "loss": 2.6373,
+      "step": 1208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.334109546251538e-06,
+      "loss": 2.5984,
+      "step": 1209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.332272826576838e-06,
+      "loss": 2.6639,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.330433758512048e-06,
+      "loss": 2.6785,
+      "step": 1211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.328592343054072e-06,
+      "loss": 2.6515,
+      "step": 1212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.326748581201084e-06,
+      "loss": 2.6696,
+      "step": 1213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.324902473952529e-06,
+      "loss": 2.6615,
+      "step": 1214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.323054022309123e-06,
+      "loss": 2.7008,
+      "step": 1215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.321203227272854e-06,
+      "loss": 2.6277,
+      "step": 1216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.319350089846983e-06,
+      "loss": 2.6494,
+      "step": 1217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.317494611036038e-06,
+      "loss": 2.7001,
+      "step": 1218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.315636791845813e-06,
+      "loss": 2.6727,
+      "step": 1219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.313776633283376e-06,
+      "loss": 2.7211,
+      "step": 1220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.31191413635706e-06,
+      "loss": 2.7129,
+      "step": 1221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.310049302076467e-06,
+      "loss": 2.637,
+      "step": 1222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.308182131452465e-06,
+      "loss": 2.6218,
+      "step": 1223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.306312625497191e-06,
+      "loss": 2.6127,
+      "step": 1224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.304440785224043e-06,
+      "loss": 2.6578,
+      "step": 1225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.30256661164769e-06,
+      "loss": 2.7476,
+      "step": 1226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.300690105784063e-06,
+      "loss": 2.6439,
+      "step": 1227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.298811268650353e-06,
+      "loss": 2.6052,
+      "step": 1228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.296930101265026e-06,
+      "loss": 2.6038,
+      "step": 1229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.295046604647799e-06,
+      "loss": 2.6792,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.293160779819658e-06,
+      "loss": 2.6958,
+      "step": 1231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.291272627802852e-06,
+      "loss": 2.6651,
+      "step": 1232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.289382149620886e-06,
+      "loss": 2.6759,
+      "step": 1233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.28748934629853e-06,
+      "loss": 2.6505,
+      "step": 1234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.285594218861816e-06,
+      "loss": 2.6534,
+      "step": 1235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.283696768338033e-06,
+      "loss": 2.6533,
+      "step": 1236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.281796995755728e-06,
+      "loss": 2.6228,
+      "step": 1237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.279894902144708e-06,
+      "loss": 2.6188,
+      "step": 1238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.277990488536042e-06,
+      "loss": 2.7145,
+      "step": 1239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.276083755962051e-06,
+      "loss": 2.7581,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.274174705456315e-06,
+      "loss": 2.7293,
+      "step": 1241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.272263338053673e-06,
+      "loss": 2.775,
+      "step": 1242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.270349654790215e-06,
+      "loss": 2.6306,
+      "step": 1243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.26843365670329e-06,
+      "loss": 2.6562,
+      "step": 1244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.266515344831502e-06,
+      "loss": 2.644,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.264594720214706e-06,
+      "loss": 2.6683,
+      "step": 1246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.262671783894015e-06,
+      "loss": 2.6183,
+      "step": 1247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.260746536911792e-06,
+      "loss": 2.6272,
+      "step": 1248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.258818980311653e-06,
+      "loss": 2.6997,
+      "step": 1249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.256889115138465e-06,
+      "loss": 2.6775,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.25495694243835e-06,
+      "loss": 2.6681,
+      "step": 1251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.25302246325868e-06,
+      "loss": 2.6457,
+      "step": 1252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.251085678648072e-06,
+      "loss": 2.6836,
+      "step": 1253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.249146589656398e-06,
+      "loss": 2.6754,
+      "step": 1254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.247205197334779e-06,
+      "loss": 2.6087,
+      "step": 1255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.245261502735582e-06,
+      "loss": 2.6594,
+      "step": 1256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.243315506912423e-06,
+      "loss": 2.668,
+      "step": 1257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.241367210920166e-06,
+      "loss": 2.5823,
+      "step": 1258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.239416615814925e-06,
+      "loss": 2.6892,
+      "step": 1259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.23746372265405e-06,
+      "loss": 2.6667,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.23550853249615e-06,
+      "loss": 2.6721,
+      "step": 1261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.233551046401068e-06,
+      "loss": 2.6834,
+      "step": 1262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.231591265429897e-06,
+      "loss": 2.6184,
+      "step": 1263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.229629190644978e-06,
+      "loss": 2.7311,
+      "step": 1264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.227664823109884e-06,
+      "loss": 2.5989,
+      "step": 1265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.22569816388944e-06,
+      "loss": 2.703,
+      "step": 1266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.223729214049712e-06,
+      "loss": 2.7196,
+      "step": 1267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.221757974658006e-06,
+      "loss": 2.7018,
+      "step": 1268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.219784446782867e-06,
+      "loss": 2.6462,
+      "step": 1269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.217808631494086e-06,
+      "loss": 2.6672,
+      "step": 1270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.215830529862688e-06,
+      "loss": 2.6452,
+      "step": 1271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.213850142960943e-06,
+      "loss": 2.662,
+      "step": 1272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.211867471862357e-06,
+      "loss": 2.6483,
+      "step": 1273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.20988251764167e-06,
+      "loss": 2.7041,
+      "step": 1274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.20789528137487e-06,
+      "loss": 2.6623,
+      "step": 1275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.205905764139172e-06,
+      "loss": 2.7028,
+      "step": 1276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.203913967013032e-06,
+      "loss": 2.6445,
+      "step": 1277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.201919891076139e-06,
+      "loss": 2.6763,
+      "step": 1278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.199923537409424e-06,
+      "loss": 2.6872,
+      "step": 1279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.197924907095045e-06,
+      "loss": 2.6706,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.195924001216399e-06,
+      "loss": 2.7143,
+      "step": 1281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.193920820858113e-06,
+      "loss": 2.5832,
+      "step": 1282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.191915367106048e-06,
+      "loss": 2.753,
+      "step": 1283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.1899076410473e-06,
+      "loss": 2.6811,
+      "step": 1284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.187897643770194e-06,
+      "loss": 2.6276,
+      "step": 1285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.185885376364288e-06,
+      "loss": 2.6964,
+      "step": 1286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.183870839920366e-06,
+      "loss": 2.7006,
+      "step": 1287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.18185403553045e-06,
+      "loss": 2.6865,
+      "step": 1288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.179834964287784e-06,
+      "loss": 2.6495,
+      "step": 1289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.177813627286842e-06,
+      "loss": 2.725,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.175790025623333e-06,
+      "loss": 2.6958,
+      "step": 1291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.173764160394184e-06,
+      "loss": 2.6634,
+      "step": 1292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.171736032697556e-06,
+      "loss": 2.6582,
+      "step": 1293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.169705643632832e-06,
+      "loss": 2.6358,
+      "step": 1294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.167672994300622e-06,
+      "loss": 2.7266,
+      "step": 1295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.165638085802764e-06,
+      "loss": 2.6496,
+      "step": 1296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.163600919242318e-06,
+      "loss": 2.6555,
+      "step": 1297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.161561495723566e-06,
+      "loss": 2.6633,
+      "step": 1298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.159519816352021e-06,
+      "loss": 2.6946,
+      "step": 1299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.157475882234408e-06,
+      "loss": 2.6907,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.155429694478685e-06,
+      "loss": 2.7675,
+      "step": 1301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.153381254194023e-06,
+      "loss": 2.6656,
+      "step": 1302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.15133056249082e-06,
+      "loss": 2.6396,
+      "step": 1303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.14927762048069e-06,
+      "loss": 2.6461,
+      "step": 1304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.14722242927647e-06,
+      "loss": 2.7014,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.145164989992215e-06,
+      "loss": 2.6401,
+      "step": 1306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.143105303743199e-06,
+      "loss": 2.7465,
+      "step": 1307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.141043371645914e-06,
+      "loss": 2.6642,
+      "step": 1308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.138979194818068e-06,
+      "loss": 2.705,
+      "step": 1309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.136912774378588e-06,
+      "loss": 2.6822,
+      "step": 1310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.134844111447616e-06,
+      "loss": 2.6764,
+      "step": 1311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.13277320714651e-06,
+      "loss": 2.6046,
+      "step": 1312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.130700062597841e-06,
+      "loss": 2.637,
+      "step": 1313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.128624678925396e-06,
+      "loss": 2.6698,
+      "step": 1314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.126547057254177e-06,
+      "loss": 2.7087,
+      "step": 1315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.124467198710401e-06,
+      "loss": 2.6323,
+      "step": 1316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.12238510442149e-06,
+      "loss": 2.6743,
+      "step": 1317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.120300775516082e-06,
+      "loss": 2.7371,
+      "step": 1318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.11821421312403e-06,
+      "loss": 2.6626,
+      "step": 1319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.116125418376394e-06,
+      "loss": 2.6441,
+      "step": 1320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.114034392405442e-06,
+      "loss": 2.6152,
+      "step": 1321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.111941136344656e-06,
+      "loss": 2.6862,
+      "step": 1322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.109845651328724e-06,
+      "loss": 2.6722,
+      "step": 1323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.107747938493545e-06,
+      "loss": 2.6495,
+      "step": 1324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.105647998976221e-06,
+      "loss": 2.6276,
+      "step": 1325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.103545833915065e-06,
+      "loss": 2.6774,
+      "step": 1326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.101441444449596e-06,
+      "loss": 2.6399,
+      "step": 1327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.099334831720537e-06,
+      "loss": 2.6814,
+      "step": 1328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.097225996869818e-06,
+      "loss": 2.6907,
+      "step": 1329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.09511494104057e-06,
+      "loss": 2.664,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.093001665377135e-06,
+      "loss": 2.6973,
+      "step": 1331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.09088617102505e-06,
+      "loss": 2.6835,
+      "step": 1332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.08876845913106e-06,
+      "loss": 2.6139,
+      "step": 1333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.08664853084311e-06,
+      "loss": 2.6691,
+      "step": 1334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.084526387310349e-06,
+      "loss": 2.6907,
+      "step": 1335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.082402029683125e-06,
+      "loss": 2.7834,
+      "step": 1336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.08027545911298e-06,
+      "loss": 2.6461,
+      "step": 1337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.07814667675267e-06,
+      "loss": 2.6984,
+      "step": 1338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.076015683756136e-06,
+      "loss": 2.6803,
+      "step": 1339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.073882481278527e-06,
+      "loss": 2.7098,
+      "step": 1340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.071747070476182e-06,
+      "loss": 2.6947,
+      "step": 1341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.069609452506642e-06,
+      "loss": 2.6745,
+      "step": 1342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.067469628528643e-06,
+      "loss": 2.671,
+      "step": 1343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.065327599702117e-06,
+      "loss": 2.6879,
+      "step": 1344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.06318336718819e-06,
+      "loss": 2.693,
+      "step": 1345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.061036932149186e-06,
+      "loss": 2.6813,
+      "step": 1346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.058888295748617e-06,
+      "loss": 2.6414,
+      "step": 1347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.056737459151198e-06,
+      "loss": 2.7117,
+      "step": 1348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.054584423522822e-06,
+      "loss": 2.6582,
+      "step": 1349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.052429190030589e-06,
+      "loss": 2.6261,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.050271759842782e-06,
+      "loss": 2.6312,
+      "step": 1351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.048112134128877e-06,
+      "loss": 2.656,
+      "step": 1352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.04595031405954e-06,
+      "loss": 2.6526,
+      "step": 1353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.043786300806622e-06,
+      "loss": 2.6351,
+      "step": 1354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.041620095543175e-06,
+      "loss": 2.6964,
+      "step": 1355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.039451699443426e-06,
+      "loss": 2.6271,
+      "step": 1356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.037281113682796e-06,
+      "loss": 2.7167,
+      "step": 1357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.03510833943789e-06,
+      "loss": 2.6571,
+      "step": 1358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.032933377886505e-06,
+      "loss": 2.6244,
+      "step": 1359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.030756230207619e-06,
+      "loss": 2.6816,
+      "step": 1360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.028576897581393e-06,
+      "loss": 2.6642,
+      "step": 1361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.026395381189179e-06,
+      "loss": 2.6037,
+      "step": 1362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.024211682213508e-06,
+      "loss": 2.732,
+      "step": 1363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.022025801838093e-06,
+      "loss": 2.8118,
+      "step": 1364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.019837741247833e-06,
+      "loss": 2.6739,
+      "step": 1365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.017647501628809e-06,
+      "loss": 2.6458,
+      "step": 1366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.015455084168279e-06,
+      "loss": 2.6682,
+      "step": 1367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.013260490054686e-06,
+      "loss": 2.6476,
+      "step": 1368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.011063720477651e-06,
+      "loss": 2.6516,
+      "step": 1369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.008864776627973e-06,
+      "loss": 2.6177,
+      "step": 1370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.006663659697632e-06,
+      "loss": 2.643,
+      "step": 1371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.004460370879783e-06,
+      "loss": 2.6527,
+      "step": 1372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.002254911368765e-06,
+      "loss": 2.7449,
+      "step": 1373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.000047282360085e-06,
+      "loss": 2.6732,
+      "step": 1374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.997837485050429e-06,
+      "loss": 2.6487,
+      "step": 1375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.995625520637662e-06,
+      "loss": 2.6606,
+      "step": 1376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.993411390320822e-06,
+      "loss": 2.6341,
+      "step": 1377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.991195095300113e-06,
+      "loss": 2.6482,
+      "step": 1378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.988976636776928e-06,
+      "loss": 2.6427,
+      "step": 1379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.98675601595382e-06,
+      "loss": 2.6803,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.984533234034518e-06,
+      "loss": 2.6919,
+      "step": 1381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.982308292223925e-06,
+      "loss": 2.7373,
+      "step": 1382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.98008119172811e-06,
+      "loss": 2.6105,
+      "step": 1383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.977851933754317e-06,
+      "loss": 2.6254,
+      "step": 1384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.975620519510957e-06,
+      "loss": 2.6839,
+      "step": 1385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.973386950207608e-06,
+      "loss": 2.5925,
+      "step": 1386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.97115122705502e-06,
+      "loss": 2.6762,
+      "step": 1387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.968913351265109e-06,
+      "loss": 2.6575,
+      "step": 1388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.966673324050958e-06,
+      "loss": 2.6532,
+      "step": 1389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.964431146626816e-06,
+      "loss": 2.6344,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.962186820208097e-06,
+      "loss": 2.6224,
+      "step": 1391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.959940346011378e-06,
+      "loss": 2.6746,
+      "step": 1392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.957691725254407e-06,
+      "loss": 2.6374,
+      "step": 1393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.955440959156088e-06,
+      "loss": 2.7068,
+      "step": 1394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.953188048936493e-06,
+      "loss": 2.8022,
+      "step": 1395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.950932995816855e-06,
+      "loss": 2.6948,
+      "step": 1396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.948675801019564e-06,
+      "loss": 2.7275,
+      "step": 1397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.94641646576818e-06,
+      "loss": 2.6732,
+      "step": 1398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.944154991287414e-06,
+      "loss": 2.7147,
+      "step": 1399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.941891378803145e-06,
+      "loss": 2.6271,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.939625629542401e-06,
+      "loss": 2.6808,
+      "step": 1401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.93735774473338e-06,
+      "loss": 2.6729,
+      "step": 1402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.93508772560543e-06,
+      "loss": 2.6954,
+      "step": 1403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.932815573389058e-06,
+      "loss": 2.6258,
+      "step": 1404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.930541289315923e-06,
+      "loss": 2.7604,
+      "step": 1405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.928264874618849e-06,
+      "loss": 2.7296,
+      "step": 1406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.925986330531804e-06,
+      "loss": 2.6066,
+      "step": 1407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.923705658289921e-06,
+      "loss": 2.7471,
+      "step": 1408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.921422859129478e-06,
+      "loss": 2.6561,
+      "step": 1409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.91913793428791e-06,
+      "loss": 2.6976,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.916850885003804e-06,
+      "loss": 2.6061,
+      "step": 1411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.914561712516898e-06,
+      "loss": 2.6118,
+      "step": 1412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.91227041806808e-06,
+      "loss": 2.6105,
+      "step": 1413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.909977002899392e-06,
+      "loss": 2.6504,
+      "step": 1414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.90768146825402e-06,
+      "loss": 2.6628,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.905383815376303e-06,
+      "loss": 2.6727,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.903084045511728e-06,
+      "loss": 2.6777,
+      "step": 1417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.900782159906927e-06,
+      "loss": 2.7082,
+      "step": 1418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.89847815980968e-06,
+      "loss": 2.6628,
+      "step": 1419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.896172046468917e-06,
+      "loss": 2.6493,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.893863821134705e-06,
+      "loss": 2.7032,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.891553485058267e-06,
+      "loss": 2.6146,
+      "step": 1422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.889241039491957e-06,
+      "loss": 2.7197,
+      "step": 1423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.886926485689287e-06,
+      "loss": 2.7049,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.8846098249049e-06,
+      "loss": 2.6984,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.882291058394587e-06,
+      "loss": 2.6482,
+      "step": 1426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.87997018741528e-06,
+      "loss": 2.7285,
+      "step": 1427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.877647213225049e-06,
+      "loss": 2.6748,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.875322137083107e-06,
+      "loss": 2.738,
+      "step": 1429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.872994960249804e-06,
+      "loss": 2.5715,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.87066568398663e-06,
+      "loss": 2.662,
+      "step": 1431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.868334309556216e-06,
+      "loss": 2.6273,
+      "step": 1432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.866000838222327e-06,
+      "loss": 2.6716,
+      "step": 1433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.863665271249859e-06,
+      "loss": 2.6051,
+      "step": 1434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.861327609904859e-06,
+      "loss": 2.6118,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.858987855454493e-06,
+      "loss": 2.6373,
+      "step": 1436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.856646009167073e-06,
+      "loss": 2.606,
+      "step": 1437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.85430207231204e-06,
+      "loss": 2.6596,
+      "step": 1438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.85195604615997e-06,
+      "loss": 2.6213,
+      "step": 1439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.84960793198257e-06,
+      "loss": 2.5985,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.847257731052676e-06,
+      "loss": 2.6495,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.844905444644265e-06,
+      "loss": 2.6498,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.842551074032433e-06,
+      "loss": 2.7026,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.840194620493414e-06,
+      "loss": 2.6859,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.837836085304568e-06,
+      "loss": 2.7629,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.83547546974438e-06,
+      "loss": 2.6012,
+      "step": 1446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.83311277509247e-06,
+      "loss": 2.6702,
+      "step": 1447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.83074800262958e-06,
+      "loss": 2.6835,
+      "step": 1448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.828381153637577e-06,
+      "loss": 2.6605,
+      "step": 1449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.82601222939946e-06,
+      "loss": 2.6116,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.823641231199347e-06,
+      "loss": 2.5983,
+      "step": 1451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.821268160322482e-06,
+      "loss": 2.6506,
+      "step": 1452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.818893018055234e-06,
+      "loss": 2.6183,
+      "step": 1453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.816515805685094e-06,
+      "loss": 2.6505,
+      "step": 1454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.814136524500671e-06,
+      "loss": 2.6911,
+      "step": 1455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.811755175791703e-06,
+      "loss": 2.639,
+      "step": 1456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.809371760849044e-06,
+      "loss": 2.6918,
+      "step": 1457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.806986280964668e-06,
+      "loss": 2.6201,
+      "step": 1458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.80459873743167e-06,
+      "loss": 2.6729,
+      "step": 1459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.802209131544264e-06,
+      "loss": 2.672,
+      "step": 1460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.799817464597778e-06,
+      "loss": 2.648,
+      "step": 1461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.797423737888662e-06,
+      "loss": 2.7209,
+      "step": 1462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.795027952714482e-06,
+      "loss": 2.7344,
+      "step": 1463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.792630110373917e-06,
+      "loss": 2.7296,
+      "step": 1464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.79023021216676e-06,
+      "loss": 2.6873,
+      "step": 1465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.787828259393927e-06,
+      "loss": 2.6339,
+      "step": 1466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.785424253357436e-06,
+      "loss": 2.5271,
+      "step": 1467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.783018195360426e-06,
+      "loss": 2.6558,
+      "step": 1468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.780610086707149e-06,
+      "loss": 2.6779,
+      "step": 1469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.77819992870296e-06,
+      "loss": 2.7185,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.775787722654336e-06,
+      "loss": 2.6765,
+      "step": 1471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.773373469868853e-06,
+      "loss": 2.6331,
+      "step": 1472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.770957171655208e-06,
+      "loss": 2.6428,
+      "step": 1473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.768538829323196e-06,
+      "loss": 2.7035,
+      "step": 1474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.766118444183728e-06,
+      "loss": 2.754,
+      "step": 1475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.763696017548821e-06,
+      "loss": 2.5954,
+      "step": 1476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.761271550731592e-06,
+      "loss": 2.6181,
+      "step": 1477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.758845045046274e-06,
+      "loss": 2.7423,
+      "step": 1478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.756416501808197e-06,
+      "loss": 2.6996,
+      "step": 1479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.753985922333797e-06,
+      "loss": 2.6623,
+      "step": 1480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.75155330794062e-06,
+      "loss": 2.6714,
+      "step": 1481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.749118659947306e-06,
+      "loss": 2.6587,
+      "step": 1482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.746681979673604e-06,
+      "loss": 2.7045,
+      "step": 1483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.744243268440365e-06,
+      "loss": 2.6487,
+      "step": 1484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.741802527569533e-06,
+      "loss": 2.6699,
+      "step": 1485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.739359758384162e-06,
+      "loss": 2.7491,
+      "step": 1486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.736914962208399e-06,
+      "loss": 2.7266,
+      "step": 1487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.734468140367491e-06,
+      "loss": 2.6592,
+      "step": 1488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.732019294187786e-06,
+      "loss": 2.7114,
+      "step": 1489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.729568424996726e-06,
+      "loss": 2.7128,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.727115534122852e-06,
+      "loss": 2.6631,
+      "step": 1491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.724660622895797e-06,
+      "loss": 2.6534,
+      "step": 1492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.722203692646294e-06,
+      "loss": 2.6957,
+      "step": 1493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.719744744706169e-06,
+      "loss": 2.7018,
+      "step": 1494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.717283780408338e-06,
+      "loss": 2.7114,
+      "step": 1495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.714820801086817e-06,
+      "loss": 2.6762,
+      "step": 1496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.712355808076708e-06,
+      "loss": 2.6392,
+      "step": 1497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.709888802714206e-06,
+      "loss": 2.6712,
+      "step": 1498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.7074197863366e-06,
+      "loss": 2.6381,
+      "step": 1499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.704948760282265e-06,
+      "loss": 2.6523,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.70247572589067e-06,
+      "loss": 2.7332,
+      "step": 1501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.700000684502366e-06,
+      "loss": 2.6975,
+      "step": 1502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.697523637458997e-06,
+      "loss": 2.6678,
+      "step": 1503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.695044586103297e-06,
+      "loss": 2.6539,
+      "step": 1504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.692563531779078e-06,
+      "loss": 2.6177,
+      "step": 1505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.690080475831241e-06,
+      "loss": 2.6783,
+      "step": 1506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.687595419605776e-06,
+      "loss": 2.6902,
+      "step": 1507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.685108364449757e-06,
+      "loss": 2.6495,
+      "step": 1508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.682619311711334e-06,
+      "loss": 2.6527,
+      "step": 1509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.680128262739747e-06,
+      "loss": 2.701,
+      "step": 1510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.677635218885318e-06,
+      "loss": 2.617,
+      "step": 1511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.675140181499445e-06,
+      "loss": 2.633,
+      "step": 1512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.672643151934614e-06,
+      "loss": 2.6509,
+      "step": 1513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.670144131544382e-06,
+      "loss": 2.5992,
+      "step": 1514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.667643121683394e-06,
+      "loss": 2.6028,
+      "step": 1515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.665140123707367e-06,
+      "loss": 2.7006,
+      "step": 1516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.662635138973099e-06,
+      "loss": 2.613,
+      "step": 1517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.660128168838466e-06,
+      "loss": 2.698,
+      "step": 1518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.657619214662417e-06,
+      "loss": 2.6123,
+      "step": 1519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.655108277804975e-06,
+      "loss": 2.6534,
+      "step": 1520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.652595359627245e-06,
+      "loss": 2.6914,
+      "step": 1521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.6500804614914e-06,
+      "loss": 2.7337,
+      "step": 1522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.647563584760688e-06,
+      "loss": 2.653,
+      "step": 1523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.64504473079943e-06,
+      "loss": 2.778,
+      "step": 1524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.642523900973017e-06,
+      "loss": 2.6736,
+      "step": 1525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.640001096647912e-06,
+      "loss": 2.7049,
+      "step": 1526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.637476319191651e-06,
+      "loss": 2.7113,
+      "step": 1527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.634949569972834e-06,
+      "loss": 2.6911,
+      "step": 1528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.632420850361136e-06,
+      "loss": 2.7148,
+      "step": 1529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.629890161727295e-06,
+      "loss": 2.6324,
+      "step": 1530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.627357505443119e-06,
+      "loss": 2.7331,
+      "step": 1531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.624822882881483e-06,
+      "loss": 2.6577,
+      "step": 1532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.622286295416324e-06,
+      "loss": 2.613,
+      "step": 1533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.619747744422651e-06,
+      "loss": 2.7145,
+      "step": 1534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.617207231276532e-06,
+      "loss": 2.6564,
+      "step": 1535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.614664757355098e-06,
+      "loss": 2.7064,
+      "step": 1536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.612120324036548e-06,
+      "loss": 2.717,
+      "step": 1537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.609573932700138e-06,
+      "loss": 2.6624,
+      "step": 1538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.607025584726187e-06,
+      "loss": 2.7029,
+      "step": 1539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.604475281496076e-06,
+      "loss": 2.6576,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.601923024392244e-06,
+      "loss": 2.6764,
+      "step": 1541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.599368814798192e-06,
+      "loss": 2.6414,
+      "step": 1542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.596812654098476e-06,
+      "loss": 2.6247,
+      "step": 1543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.594254543678711e-06,
+      "loss": 2.6527,
+      "step": 1544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.591694484925567e-06,
+      "loss": 2.7001,
+      "step": 1545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.589132479226777e-06,
+      "loss": 2.5875,
+      "step": 1546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.586568527971122e-06,
+      "loss": 2.6461,
+      "step": 1547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.584002632548438e-06,
+      "loss": 2.676,
+      "step": 1548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.581434794349619e-06,
+      "loss": 2.6258,
+      "step": 1549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.578865014766611e-06,
+      "loss": 2.6882,
+      "step": 1550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.576293295192409e-06,
+      "loss": 2.7454,
+      "step": 1551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.573719637021063e-06,
+      "loss": 2.6508,
+      "step": 1552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.571144041647674e-06,
+      "loss": 2.6134,
+      "step": 1553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.568566510468392e-06,
+      "loss": 2.6368,
+      "step": 1554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.565987044880413e-06,
+      "loss": 2.619,
+      "step": 1555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.563405646281989e-06,
+      "loss": 2.6691,
+      "step": 1556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.560822316072414e-06,
+      "loss": 2.686,
+      "step": 1557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.55823705565203e-06,
+      "loss": 2.7224,
+      "step": 1558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.555649866422227e-06,
+      "loss": 2.6593,
+      "step": 1559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.55306074978544e-06,
+      "loss": 2.5605,
+      "step": 1560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.550469707145148e-06,
+      "loss": 2.627,
+      "step": 1561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.547876739905874e-06,
+      "loss": 2.6356,
+      "step": 1562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.545281849473187e-06,
+      "loss": 2.664,
+      "step": 1563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.542685037253693e-06,
+      "loss": 2.6619,
+      "step": 1564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.540086304655045e-06,
+      "loss": 2.68,
+      "step": 1565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.537485653085934e-06,
+      "loss": 2.65,
+      "step": 1566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.534883083956093e-06,
+      "loss": 2.664,
+      "step": 1567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.532278598676293e-06,
+      "loss": 2.6295,
+      "step": 1568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.529672198658344e-06,
+      "loss": 2.6903,
+      "step": 1569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.527063885315096e-06,
+      "loss": 2.6812,
+      "step": 1570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.524453660060434e-06,
+      "loss": 2.7074,
+      "step": 1571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.521841524309279e-06,
+      "loss": 2.6716,
+      "step": 1572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.519227479477584e-06,
+      "loss": 2.6936,
+      "step": 1573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.516611526982349e-06,
+      "loss": 2.72,
+      "step": 1574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.513993668241596e-06,
+      "loss": 2.6247,
+      "step": 1575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.511373904674385e-06,
+      "loss": 2.7139,
+      "step": 1576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.50875223770081e-06,
+      "loss": 2.7333,
+      "step": 1577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.506128668741993e-06,
+      "loss": 2.6082,
+      "step": 1578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.503503199220088e-06,
+      "loss": 2.6571,
+      "step": 1579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.500875830558284e-06,
+      "loss": 2.6715,
+      "step": 1580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.498246564180793e-06,
+      "loss": 2.6698,
+      "step": 1581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.49561540151286e-06,
+      "loss": 2.5837,
+      "step": 1582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.492982343980753e-06,
+      "loss": 2.598,
+      "step": 1583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.490347393011772e-06,
+      "loss": 2.687,
+      "step": 1584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.487710550034245e-06,
+      "loss": 2.6542,
+      "step": 1585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.48507181647752e-06,
+      "loss": 2.7626,
+      "step": 1586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.48243119377197e-06,
+      "loss": 2.6556,
+      "step": 1587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.479788683348996e-06,
+      "loss": 2.6449,
+      "step": 1588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.47714428664102e-06,
+      "loss": 2.6797,
+      "step": 1589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.47449800508149e-06,
+      "loss": 2.7321,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.471849840104871e-06,
+      "loss": 2.6969,
+      "step": 1591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.46919979314665e-06,
+      "loss": 2.6974,
+      "step": 1592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.466547865643335e-06,
+      "loss": 2.6639,
+      "step": 1593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.463894059032453e-06,
+      "loss": 2.6792,
+      "step": 1594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.46123837475255e-06,
+      "loss": 2.6534,
+      "step": 1595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.458580814243191e-06,
+      "loss": 2.6624,
+      "step": 1596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.455921378944955e-06,
+      "loss": 2.7106,
+      "step": 1597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.45326007029944e-06,
+      "loss": 2.6003,
+      "step": 1598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.450596889749258e-06,
+      "loss": 2.6843,
+      "step": 1599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.447931838738037e-06,
+      "loss": 2.6554,
+      "step": 1600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.445264918710417e-06,
+      "loss": 2.6698,
+      "step": 1601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.442596131112052e-06,
+      "loss": 2.6093,
+      "step": 1602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.439925477389611e-06,
+      "loss": 2.6333,
+      "step": 1603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.43725295899077e-06,
+      "loss": 2.7173,
+      "step": 1604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.434578577364218e-06,
+      "loss": 2.6105,
+      "step": 1605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.431902333959655e-06,
+      "loss": 2.6719,
+      "step": 1606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.429224230227788e-06,
+      "loss": 2.676,
+      "step": 1607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.426544267620334e-06,
+      "loss": 2.7163,
+      "step": 1608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.42386244759002e-06,
+      "loss": 2.6704,
+      "step": 1609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.421178771590568e-06,
+      "loss": 2.6502,
+      "step": 1610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.418493241076725e-06,
+      "loss": 2.7562,
+      "step": 1611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.415805857504231e-06,
+      "loss": 2.6821,
+      "step": 1612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.41311662232983e-06,
+      "loss": 2.605,
+      "step": 1613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.410425537011277e-06,
+      "loss": 2.6674,
+      "step": 1614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.40773260300732e-06,
+      "loss": 2.6694,
+      "step": 1615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.40503782177772e-06,
+      "loss": 2.6513,
+      "step": 1616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.40234119478323e-06,
+      "loss": 2.6336,
+      "step": 1617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.399642723485607e-06,
+      "loss": 2.6359,
+      "step": 1618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.396942409347612e-06,
+      "loss": 2.717,
+      "step": 1619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.394240253832999e-06,
+      "loss": 2.5842,
+      "step": 1620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.391536258406523e-06,
+      "loss": 2.621,
+      "step": 1621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.388830424533935e-06,
+      "loss": 2.7619,
+      "step": 1622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.386122753681983e-06,
+      "loss": 2.6438,
+      "step": 1623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.38341324731841e-06,
+      "loss": 2.6127,
+      "step": 1624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.380701906911956e-06,
+      "loss": 2.6537,
+      "step": 1625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.377988733932357e-06,
+      "loss": 2.6518,
+      "step": 1626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.375273729850336e-06,
+      "loss": 2.6776,
+      "step": 1627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.372556896137612e-06,
+      "loss": 2.593,
+      "step": 1628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.369838234266899e-06,
+      "loss": 2.7141,
+      "step": 1629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.367117745711894e-06,
+      "loss": 2.5916,
+      "step": 1630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.364395431947293e-06,
+      "loss": 2.6253,
+      "step": 1631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.361671294448777e-06,
+      "loss": 2.6652,
+      "step": 1632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.358945334693014e-06,
+      "loss": 2.7834,
+      "step": 1633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.356217554157665e-06,
+      "loss": 2.6706,
+      "step": 1634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.353487954321371e-06,
+      "loss": 2.7517,
+      "step": 1635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.350756536663766e-06,
+      "loss": 2.6595,
+      "step": 1636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.348023302665465e-06,
+      "loss": 2.635,
+      "step": 1637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.345288253808071e-06,
+      "loss": 2.654,
+      "step": 1638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.342551391574165e-06,
+      "loss": 2.6327,
+      "step": 1639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.339812717447317e-06,
+      "loss": 2.5993,
+      "step": 1640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.337072232912079e-06,
+      "loss": 2.6868,
+      "step": 1641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.334329939453978e-06,
+      "loss": 2.6698,
+      "step": 1642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.331585838559528e-06,
+      "loss": 2.6223,
+      "step": 1643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.328839931716219e-06,
+      "loss": 2.642,
+      "step": 1644
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.326092220412526e-06,
+      "loss": 2.6088,
+      "step": 1645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.323342706137894e-06,
+      "loss": 2.7032,
+      "step": 1646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.320591390382749e-06,
+      "loss": 2.6657,
+      "step": 1647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.317838274638493e-06,
+      "loss": 2.7056,
+      "step": 1648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.315083360397507e-06,
+      "loss": 2.5988,
+      "step": 1649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.312326649153141e-06,
+      "loss": 2.6406,
+      "step": 1650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.309568142399725e-06,
+      "loss": 2.5961,
+      "step": 1651
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.306807841632559e-06,
+      "loss": 2.6201,
+      "step": 1652
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.304045748347914e-06,
+      "loss": 2.6196,
+      "step": 1653
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.301281864043035e-06,
+      "loss": 2.6613,
+      "step": 1654
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.29851619021614e-06,
+      "loss": 2.6631,
+      "step": 1655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.295748728366414e-06,
+      "loss": 2.6293,
+      "step": 1656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.292979479994009e-06,
+      "loss": 2.6627,
+      "step": 1657
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.290208446600052e-06,
+      "loss": 2.6828,
+      "step": 1658
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.287435629686631e-06,
+      "loss": 2.6419,
+      "step": 1659
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.284661030756804e-06,
+      "loss": 2.6758,
+      "step": 1660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.281884651314598e-06,
+      "loss": 2.7242,
+      "step": 1661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.279106492864997e-06,
+      "loss": 2.6083,
+      "step": 1662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.276326556913958e-06,
+      "loss": 2.6836,
+      "step": 1663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.273544844968393e-06,
+      "loss": 2.6749,
+      "step": 1664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.270761358536187e-06,
+      "loss": 2.6919,
+      "step": 1665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.267976099126178e-06,
+      "loss": 2.6537,
+      "step": 1666
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.265189068248167e-06,
+      "loss": 2.7961,
+      "step": 1667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.26240026741292e-06,
+      "loss": 2.6711,
+      "step": 1668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.259609698132159e-06,
+      "loss": 2.6726,
+      "step": 1669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.256817361918559e-06,
+      "loss": 2.6767,
+      "step": 1670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.254023260285765e-06,
+      "loss": 2.5824,
+      "step": 1671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.25122739474837e-06,
+      "loss": 2.6784,
+      "step": 1672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.248429766821925e-06,
+      "loss": 2.6351,
+      "step": 1673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.245630378022937e-06,
+      "loss": 2.7051,
+      "step": 1674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.24282922986887e-06,
+      "loss": 2.6769,
+      "step": 1675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.240026323878135e-06,
+      "loss": 2.7062,
+      "step": 1676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.237221661570102e-06,
+      "loss": 2.6614,
+      "step": 1677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.23441524446509e-06,
+      "loss": 2.659,
+      "step": 1678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.231607074084373e-06,
+      "loss": 2.6858,
+      "step": 1679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.228797151950172e-06,
+      "loss": 2.7062,
+      "step": 1680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.225985479585655e-06,
+      "loss": 2.7473,
+      "step": 1681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.223172058514945e-06,
+      "loss": 2.6338,
+      "step": 1682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.220356890263108e-06,
+      "loss": 2.6745,
+      "step": 1683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.21753997635616e-06,
+      "loss": 2.6657,
+      "step": 1684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.214721318321064e-06,
+      "loss": 2.676,
+      "step": 1685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.211900917685722e-06,
+      "loss": 2.6527,
+      "step": 1686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.20907877597899e-06,
+      "loss": 2.6874,
+      "step": 1687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.20625489473066e-06,
+      "loss": 2.5794,
+      "step": 1688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.203429275471473e-06,
+      "loss": 2.6683,
+      "step": 1689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.200601919733106e-06,
+      "loss": 2.6924,
+      "step": 1690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.197772829048183e-06,
+      "loss": 2.5881,
+      "step": 1691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.194942004950264e-06,
+      "loss": 2.647,
+      "step": 1692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.192109448973851e-06,
+      "loss": 2.6932,
+      "step": 1693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.189275162654385e-06,
+      "loss": 2.7064,
+      "step": 1694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.186439147528244e-06,
+      "loss": 2.5796,
+      "step": 1695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.183601405132743e-06,
+      "loss": 2.6001,
+      "step": 1696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.180761937006133e-06,
+      "loss": 2.6172,
+      "step": 1697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.177920744687604e-06,
+      "loss": 2.645,
+      "step": 1698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.175077829717276e-06,
+      "loss": 2.5924,
+      "step": 1699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.172233193636204e-06,
+      "loss": 2.712,
+      "step": 1700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.169386837986375e-06,
+      "loss": 2.6239,
+      "step": 1701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.166538764310715e-06,
+      "loss": 2.6057,
+      "step": 1702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.16368897415307e-06,
+      "loss": 2.6946,
+      "step": 1703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.160837469058228e-06,
+      "loss": 2.6274,
+      "step": 1704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.157984250571895e-06,
+      "loss": 2.6925,
+      "step": 1705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.155129320240718e-06,
+      "loss": 2.6372,
+      "step": 1706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.15227267961226e-06,
+      "loss": 2.6124,
+      "step": 1707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.149414330235024e-06,
+      "loss": 2.6529,
+      "step": 1708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.146554273658425e-06,
+      "loss": 2.6108,
+      "step": 1709
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.143692511432814e-06,
+      "loss": 2.6934,
+      "step": 1710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.140829045109462e-06,
+      "loss": 2.6159,
+      "step": 1711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.137963876240567e-06,
+      "loss": 2.6609,
+      "step": 1712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.135097006379245e-06,
+      "loss": 2.6956,
+      "step": 1713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.132228437079538e-06,
+      "loss": 2.71,
+      "step": 1714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.129358169896409e-06,
+      "loss": 2.5948,
+      "step": 1715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.126486206385738e-06,
+      "loss": 2.7078,
+      "step": 1716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.123612548104328e-06,
+      "loss": 2.6423,
+      "step": 1717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.1207371966099e-06,
+      "loss": 2.6635,
+      "step": 1718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.117860153461094e-06,
+      "loss": 2.662,
+      "step": 1719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.114981420217462e-06,
+      "loss": 2.72,
+      "step": 1720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.112100998439478e-06,
+      "loss": 2.6918,
+      "step": 1721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.10921888968853e-06,
+      "loss": 2.7431,
+      "step": 1722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.106335095526915e-06,
+      "loss": 2.6987,
+      "step": 1723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.10344961751785e-06,
+      "loss": 2.6441,
+      "step": 1724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.100562457225465e-06,
+      "loss": 2.638,
+      "step": 1725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.097673616214799e-06,
+      "loss": 2.6833,
+      "step": 1726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.094783096051799e-06,
+      "loss": 2.6006,
+      "step": 1727
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.091890898303329e-06,
+      "loss": 2.6818,
+      "step": 1728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.088997024537157e-06,
+      "loss": 2.7014,
+      "step": 1729
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.086101476321968e-06,
+      "loss": 2.669,
+      "step": 1730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.08320425522734e-06,
+      "loss": 2.621,
+      "step": 1731
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.080305362823771e-06,
+      "loss": 2.7584,
+      "step": 1732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.07740480068266e-06,
+      "loss": 2.6331,
+      "step": 1733
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.07450257037631e-06,
+      "loss": 2.6262,
+      "step": 1734
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.071598673477931e-06,
+      "loss": 2.6734,
+      "step": 1735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.068693111561634e-06,
+      "loss": 2.6566,
+      "step": 1736
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.065785886202435e-06,
+      "loss": 2.5897,
+      "step": 1737
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.062876998976248e-06,
+      "loss": 2.613,
+      "step": 1738
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.05996645145989e-06,
+      "loss": 2.7281,
+      "step": 1739
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.057054245231082e-06,
+      "loss": 2.6871,
+      "step": 1740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.054140381868435e-06,
+      "loss": 2.6406,
+      "step": 1741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.05122486295147e-06,
+      "loss": 2.72,
+      "step": 1742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.048307690060593e-06,
+      "loss": 2.6778,
+      "step": 1743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.045388864777113e-06,
+      "loss": 2.6426,
+      "step": 1744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.042468388683239e-06,
+      "loss": 2.6802,
+      "step": 1745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.039546263362065e-06,
+      "loss": 2.7086,
+      "step": 1746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.036622490397586e-06,
+      "loss": 2.649,
+      "step": 1747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.033697071374688e-06,
+      "loss": 2.6198,
+      "step": 1748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.03077000787915e-06,
+      "loss": 2.6678,
+      "step": 1749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.027841301497642e-06,
+      "loss": 2.6505,
+      "step": 1750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.024910953817724e-06,
+      "loss": 2.6223,
+      "step": 1751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.021978966427845e-06,
+      "loss": 2.6251,
+      "step": 1752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.019045340917348e-06,
+      "loss": 2.6015,
+      "step": 1753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.016110078876455e-06,
+      "loss": 2.6208,
+      "step": 1754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.013173181896283e-06,
+      "loss": 2.6054,
+      "step": 1755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.01023465156883e-06,
+      "loss": 2.6896,
+      "step": 1756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.007294489486985e-06,
+      "loss": 2.6012,
+      "step": 1757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.004352697244516e-06,
+      "loss": 2.6448,
+      "step": 1758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.00140927643608e-06,
+      "loss": 2.6736,
+      "step": 1759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.998464228657206e-06,
+      "loss": 2.7102,
+      "step": 1760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.99551755550432e-06,
+      "loss": 2.7744,
+      "step": 1761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.99256925857472e-06,
+      "loss": 2.7509,
+      "step": 1762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.989619339466586e-06,
+      "loss": 2.6956,
+      "step": 1763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.986667799778975e-06,
+      "loss": 2.6979,
+      "step": 1764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.983714641111827e-06,
+      "loss": 2.6856,
+      "step": 1765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.980759865065956e-06,
+      "loss": 2.656,
+      "step": 1766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.977803473243056e-06,
+      "loss": 2.677,
+      "step": 1767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.974845467245691e-06,
+      "loss": 2.6516,
+      "step": 1768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.971885848677307e-06,
+      "loss": 2.5876,
+      "step": 1769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.968924619142222e-06,
+      "loss": 2.6851,
+      "step": 1770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.965961780245622e-06,
+      "loss": 2.6092,
+      "step": 1771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.962997333593574e-06,
+      "loss": 2.683,
+      "step": 1772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.96003128079301e-06,
+      "loss": 2.7008,
+      "step": 1773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.957063623451734e-06,
+      "loss": 2.6507,
+      "step": 1774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.954094363178421e-06,
+      "loss": 2.6636,
+      "step": 1775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.951123501582617e-06,
+      "loss": 2.637,
+      "step": 1776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.94815104027473e-06,
+      "loss": 2.6393,
+      "step": 1777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.945176980866041e-06,
+      "loss": 2.6263,
+      "step": 1778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.942201324968694e-06,
+      "loss": 2.6327,
+      "step": 1779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.9392240741957e-06,
+      "loss": 2.62,
+      "step": 1780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.936245230160932e-06,
+      "loss": 2.68,
+      "step": 1781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.933264794479133e-06,
+      "loss": 2.6615,
+      "step": 1782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.930282768765897e-06,
+      "loss": 2.654,
+      "step": 1783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.927299154637693e-06,
+      "loss": 2.581,
+      "step": 1784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.924313953711845e-06,
+      "loss": 2.718,
+      "step": 1785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.921327167606535e-06,
+      "loss": 2.7044,
+      "step": 1786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.918338797940806e-06,
+      "loss": 2.6006,
+      "step": 1787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.915348846334563e-06,
+      "loss": 2.5567,
+      "step": 1788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.912357314408564e-06,
+      "loss": 2.7048,
+      "step": 1789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.909364203784425e-06,
+      "loss": 2.6354,
+      "step": 1790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.90636951608462e-06,
+      "loss": 2.6352,
+      "step": 1791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.903373252932474e-06,
+      "loss": 2.7186,
+      "step": 1792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.900375415952168e-06,
+      "loss": 2.5979,
+      "step": 1793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.897376006768736e-06,
+      "loss": 2.6765,
+      "step": 1794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.894375027008064e-06,
+      "loss": 2.6503,
+      "step": 1795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.89137247829689e-06,
+      "loss": 2.6411,
+      "step": 1796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.888368362262805e-06,
+      "loss": 2.7184,
+      "step": 1797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.88536268053424e-06,
+      "loss": 2.6076,
+      "step": 1798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.882355434740488e-06,
+      "loss": 2.6397,
+      "step": 1799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.879346626511679e-06,
+      "loss": 2.6532,
+      "step": 1800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.876336257478796e-06,
+      "loss": 2.605,
+      "step": 1801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.873324329273665e-06,
+      "loss": 2.694,
+      "step": 1802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.870310843528962e-06,
+      "loss": 2.6863,
+      "step": 1803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.867295801878202e-06,
+      "loss": 2.6529,
+      "step": 1804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.864279205955739e-06,
+      "loss": 2.6881,
+      "step": 1805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.861261057396786e-06,
+      "loss": 2.6238,
+      "step": 1806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.858241357837377e-06,
+      "loss": 2.6004,
+      "step": 1807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.855220108914405e-06,
+      "loss": 2.6759,
+      "step": 1808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.852197312265592e-06,
+      "loss": 2.6248,
+      "step": 1809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.849172969529501e-06,
+      "loss": 2.6296,
+      "step": 1810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.846147082345536e-06,
+      "loss": 2.6458,
+      "step": 1811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.843119652353934e-06,
+      "loss": 2.6932,
+      "step": 1812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.840090681195772e-06,
+      "loss": 2.6091,
+      "step": 1813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.83706017051296e-06,
+      "loss": 2.6635,
+      "step": 1814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.834028121948243e-06,
+      "loss": 2.5826,
+      "step": 1815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.830994537145202e-06,
+      "loss": 2.6633,
+      "step": 1816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.827959417748247e-06,
+      "loss": 2.6492,
+      "step": 1817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.824922765402622e-06,
+      "loss": 2.6941,
+      "step": 1818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.821884581754402e-06,
+      "loss": 2.6516,
+      "step": 1819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.81884486845049e-06,
+      "loss": 2.6678,
+      "step": 1820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.815803627138622e-06,
+      "loss": 2.7208,
+      "step": 1821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.81276085946736e-06,
+      "loss": 2.7257,
+      "step": 1822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.809716567086093e-06,
+      "loss": 2.6074,
+      "step": 1823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.806670751645037e-06,
+      "loss": 2.5826,
+      "step": 1824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.803623414795232e-06,
+      "loss": 2.7395,
+      "step": 1825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.800574558188548e-06,
+      "loss": 2.6227,
+      "step": 1826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.797524183477672e-06,
+      "loss": 2.6734,
+      "step": 1827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.794472292316118e-06,
+      "loss": 2.668,
+      "step": 1828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.791418886358221e-06,
+      "loss": 2.5939,
+      "step": 1829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.788363967259139e-06,
+      "loss": 2.676,
+      "step": 1830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.785307536674845e-06,
+      "loss": 2.6078,
+      "step": 1831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.782249596262137e-06,
+      "loss": 2.6484,
+      "step": 1832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.779190147678628e-06,
+      "loss": 2.6949,
+      "step": 1833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.776129192582754e-06,
+      "loss": 2.7315,
+      "step": 1834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.773066732633755e-06,
+      "loss": 2.6648,
+      "step": 1835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.770002769491705e-06,
+      "loss": 2.6242,
+      "step": 1836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.766937304817476e-06,
+      "loss": 2.6293,
+      "step": 1837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.76387034027276e-06,
+      "loss": 2.6687,
+      "step": 1838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.76080187752007e-06,
+      "loss": 2.6618,
+      "step": 1839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.757731918222719e-06,
+      "loss": 2.7212,
+      "step": 1840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.754660464044836e-06,
+      "loss": 2.6433,
+      "step": 1841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.751587516651361e-06,
+      "loss": 2.6669,
+      "step": 1842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.748513077708044e-06,
+      "loss": 2.6281,
+      "step": 1843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.745437148881443e-06,
+      "loss": 2.6469,
+      "step": 1844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.742359731838921e-06,
+      "loss": 2.698,
+      "step": 1845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.739280828248653e-06,
+      "loss": 2.731,
+      "step": 1846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.736200439779615e-06,
+      "loss": 2.6812,
+      "step": 1847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.733118568101587e-06,
+      "loss": 2.657,
+      "step": 1848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.73003521488516e-06,
+      "loss": 2.6624,
+      "step": 1849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.726950381801723e-06,
+      "loss": 2.6686,
+      "step": 1850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.723864070523464e-06,
+      "loss": 2.706,
+      "step": 1851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.72077628272338e-06,
+      "loss": 2.6232,
+      "step": 1852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.717687020075264e-06,
+      "loss": 2.582,
+      "step": 1853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.714596284253708e-06,
+      "loss": 2.7062,
+      "step": 1854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.711504076934102e-06,
+      "loss": 2.6781,
+      "step": 1855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.70841039979264e-06,
+      "loss": 2.7162,
+      "step": 1856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.705315254506304e-06,
+      "loss": 2.6546,
+      "step": 1857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.702218642752876e-06,
+      "loss": 2.6792,
+      "step": 1858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.699120566210936e-06,
+      "loss": 2.6012,
+      "step": 1859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.69602102655985e-06,
+      "loss": 2.6899,
+      "step": 1860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.692920025479785e-06,
+      "loss": 2.5782,
+      "step": 1861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.689817564651696e-06,
+      "loss": 2.7143,
+      "step": 1862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.68671364575733e-06,
+      "loss": 2.7233,
+      "step": 1863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.683608270479229e-06,
+      "loss": 2.7083,
+      "step": 1864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.680501440500712e-06,
+      "loss": 2.5875,
+      "step": 1865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.677393157505903e-06,
+      "loss": 2.6226,
+      "step": 1866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.674283423179699e-06,
+      "loss": 2.6625,
+      "step": 1867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.671172239207793e-06,
+      "loss": 2.6573,
+      "step": 1868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.668059607276661e-06,
+      "loss": 2.6963,
+      "step": 1869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.664945529073564e-06,
+      "loss": 2.6304,
+      "step": 1870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.661830006286543e-06,
+      "loss": 2.6749,
+      "step": 1871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.658713040604432e-06,
+      "loss": 2.6517,
+      "step": 1872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.655594633716836e-06,
+      "loss": 2.7227,
+      "step": 1873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.65247478731415e-06,
+      "loss": 2.6873,
+      "step": 1874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.649353503087541e-06,
+      "loss": 2.6735,
+      "step": 1875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.646230782728964e-06,
+      "loss": 2.7038,
+      "step": 1876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.643106627931148e-06,
+      "loss": 2.6722,
+      "step": 1877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.639981040387596e-06,
+      "loss": 2.6873,
+      "step": 1878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.636854021792595e-06,
+      "loss": 2.6769,
+      "step": 1879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.633725573841204e-06,
+      "loss": 2.7439,
+      "step": 1880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.630595698229258e-06,
+      "loss": 2.6287,
+      "step": 1881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.627464396653362e-06,
+      "loss": 2.6841,
+      "step": 1882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.624331670810898e-06,
+      "loss": 2.6634,
+      "step": 1883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.6211975224000216e-06,
+      "loss": 2.5771,
+      "step": 1884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.618061953119654e-06,
+      "loss": 2.6117,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.614924964669492e-06,
+      "loss": 2.6723,
+      "step": 1886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.611786558749998e-06,
+      "loss": 2.6995,
+      "step": 1887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.608646737062406e-06,
+      "loss": 2.6913,
+      "step": 1888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.605505501308711e-06,
+      "loss": 2.6085,
+      "step": 1889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.602362853191683e-06,
+      "loss": 2.6719,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.599218794414851e-06,
+      "loss": 2.6832,
+      "step": 1891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.596073326682514e-06,
+      "loss": 2.6049,
+      "step": 1892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.592926451699729e-06,
+      "loss": 2.6564,
+      "step": 1893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.5897781711723215e-06,
+      "loss": 2.6405,
+      "step": 1894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.586628486806875e-06,
+      "loss": 2.6922,
+      "step": 1895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.583477400310737e-06,
+      "loss": 2.6558,
+      "step": 1896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.580324913392009e-06,
+      "loss": 2.6471,
+      "step": 1897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.57717102775956e-06,
+      "loss": 2.6513,
+      "step": 1898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.574015745123011e-06,
+      "loss": 2.7249,
+      "step": 1899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.570859067192744e-06,
+      "loss": 2.7373,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.5677009956798954e-06,
+      "loss": 2.5935,
+      "step": 1901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.564541532296355e-06,
+      "loss": 2.6203,
+      "step": 1902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.561380678754773e-06,
+      "loss": 2.6898,
+      "step": 1903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.558218436768547e-06,
+      "loss": 2.6242,
+      "step": 1904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.555054808051832e-06,
+      "loss": 2.7042,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.551889794319529e-06,
+      "loss": 2.6613,
+      "step": 1906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.548723397287299e-06,
+      "loss": 2.7029,
+      "step": 1907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.54555561867154e-06,
+      "loss": 2.6497,
+      "step": 1908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.5423864601894105e-06,
+      "loss": 2.6369,
+      "step": 1909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.539215923558814e-06,
+      "loss": 2.6295,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.536044010498396e-06,
+      "loss": 2.6424,
+      "step": 1911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.532870722727552e-06,
+      "loss": 2.6584,
+      "step": 1912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.529696061966425e-06,
+      "loss": 2.7021,
+      "step": 1913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.526520029935897e-06,
+      "loss": 2.6886,
+      "step": 1914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.523342628357599e-06,
+      "loss": 2.6267,
+      "step": 1915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.520163858953898e-06,
+      "loss": 2.6035,
+      "step": 1916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.516983723447908e-06,
+      "loss": 2.6851,
+      "step": 1917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.513802223563482e-06,
+      "loss": 2.5682,
+      "step": 1918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.5106193610252085e-06,
+      "loss": 2.6665,
+      "step": 1919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.507435137558422e-06,
+      "loss": 2.5941,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.504249554889188e-06,
+      "loss": 2.6645,
+      "step": 1921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.501062614744313e-06,
+      "loss": 2.6964,
+      "step": 1922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.497874318851337e-06,
+      "loss": 2.6294,
+      "step": 1923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.494684668938537e-06,
+      "loss": 2.66,
+      "step": 1924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.491493666734924e-06,
+      "loss": 2.6949,
+      "step": 1925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.488301313970236e-06,
+      "loss": 2.6831,
+      "step": 1926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.485107612374953e-06,
+      "loss": 2.6323,
+      "step": 1927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.48191256368028e-06,
+      "loss": 2.6523,
+      "step": 1928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.478716169618151e-06,
+      "loss": 2.7119,
+      "step": 1929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.475518431921233e-06,
+      "loss": 2.6386,
+      "step": 1930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.472319352322921e-06,
+      "loss": 2.6199,
+      "step": 1931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.469118932557336e-06,
+      "loss": 2.6657,
+      "step": 1932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.465917174359323e-06,
+      "loss": 2.646,
+      "step": 1933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.462714079464459e-06,
+      "loss": 2.7399,
+      "step": 1934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.4595096496090405e-06,
+      "loss": 2.6845,
+      "step": 1935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.456303886530088e-06,
+      "loss": 2.6438,
+      "step": 1936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.453096791965349e-06,
+      "loss": 2.716,
+      "step": 1937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.4498883676532855e-06,
+      "loss": 2.6695,
+      "step": 1938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.446678615333086e-06,
+      "loss": 2.5435,
+      "step": 1939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.443467536744659e-06,
+      "loss": 2.6431,
+      "step": 1940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.440255133628627e-06,
+      "loss": 2.6527,
+      "step": 1941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.437041407726335e-06,
+      "loss": 2.6001,
+      "step": 1942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.433826360779845e-06,
+      "loss": 2.6345,
+      "step": 1943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.430609994531932e-06,
+      "loss": 2.6339,
+      "step": 1944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.427392310726088e-06,
+      "loss": 2.6102,
+      "step": 1945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.42417331110652e-06,
+      "loss": 2.671,
+      "step": 1946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.420952997418147e-06,
+      "loss": 2.6978,
+      "step": 1947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.417731371406601e-06,
+      "loss": 2.7422,
+      "step": 1948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.414508434818224e-06,
+      "loss": 2.6862,
+      "step": 1949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.411284189400071e-06,
+      "loss": 2.7018,
+      "step": 1950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.4080586368999045e-06,
+      "loss": 2.6477,
+      "step": 1951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.404831779066194e-06,
+      "loss": 2.7387,
+      "step": 1952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.401603617648121e-06,
+      "loss": 2.6949,
+      "step": 1953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.398374154395571e-06,
+      "loss": 2.6481,
+      "step": 1954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.395143391059134e-06,
+      "loss": 2.6152,
+      "step": 1955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.391911329390107e-06,
+      "loss": 2.7155,
+      "step": 1956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.38867797114049e-06,
+      "loss": 2.6121,
+      "step": 1957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.385443318062985e-06,
+      "loss": 2.7134,
+      "step": 1958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.382207371910997e-06,
+      "loss": 2.7012,
+      "step": 1959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.378970134438633e-06,
+      "loss": 2.6704,
+      "step": 1960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.375731607400697e-06,
+      "loss": 2.7034,
+      "step": 1961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.372491792552694e-06,
+      "loss": 2.6533,
+      "step": 1962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.369250691650825e-06,
+      "loss": 2.6812,
+      "step": 1963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3660083064519934e-06,
+      "loss": 2.6846,
+      "step": 1964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.362764638713793e-06,
+      "loss": 2.7218,
+      "step": 1965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.359519690194514e-06,
+      "loss": 2.5868,
+      "step": 1966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.356273462653143e-06,
+      "loss": 2.656,
+      "step": 1967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.35302595784936e-06,
+      "loss": 2.6581,
+      "step": 1968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3497771775435334e-06,
+      "loss": 2.6754,
+      "step": 1969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.346527123496727e-06,
+      "loss": 2.6621,
+      "step": 1970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.343275797470696e-06,
+      "loss": 2.6569,
+      "step": 1971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.34002320122788e-06,
+      "loss": 2.6671,
+      "step": 1972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.336769336531411e-06,
+      "loss": 2.6859,
+      "step": 1973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.333514205145107e-06,
+      "loss": 2.641,
+      "step": 1974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3302578088334755e-06,
+      "loss": 2.6524,
+      "step": 1975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.327000149361705e-06,
+      "loss": 2.6552,
+      "step": 1976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.323741228495672e-06,
+      "loss": 2.7321,
+      "step": 1977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3204810480019375e-06,
+      "loss": 2.6562,
+      "step": 1978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.31721960964774e-06,
+      "loss": 2.6284,
+      "step": 1979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.313956915201007e-06,
+      "loss": 2.6708,
+      "step": 1980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.310692966430341e-06,
+      "loss": 2.6233,
+      "step": 1981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.307427765105031e-06,
+      "loss": 2.7387,
+      "step": 1982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.304161312995037e-06,
+      "loss": 2.6641,
+      "step": 1983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.300893611871002e-06,
+      "loss": 2.7388,
+      "step": 1984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.297624663504246e-06,
+      "loss": 2.6561,
+      "step": 1985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.294354469666761e-06,
+      "loss": 2.7115,
+      "step": 1986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.291083032131221e-06,
+      "loss": 2.6942,
+      "step": 1987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.287810352670967e-06,
+      "loss": 2.6535,
+      "step": 1988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.284536433060019e-06,
+      "loss": 2.6396,
+      "step": 1989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.281261275073066e-06,
+      "loss": 2.6622,
+      "step": 1990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.277984880485469e-06,
+      "loss": 2.6431,
+      "step": 1991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.27470725107326e-06,
+      "loss": 2.6784,
+      "step": 1992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.271428388613138e-06,
+      "loss": 2.6935,
+      "step": 1993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.268148294882472e-06,
+      "loss": 2.6522,
+      "step": 1994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.264866971659303e-06,
+      "loss": 2.6588,
+      "step": 1995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.261584420722328e-06,
+      "loss": 2.5726,
+      "step": 1996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.25830064385092e-06,
+      "loss": 2.6375,
+      "step": 1997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2550156428251115e-06,
+      "loss": 2.6533,
+      "step": 1998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.251729419425597e-06,
+      "loss": 2.6267,
+      "step": 1999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.248441975433738e-06,
+      "loss": 2.7149,
+      "step": 2000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.245153312631554e-06,
+      "loss": 2.6383,
+      "step": 2001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.241863432801726e-06,
+      "loss": 2.6659,
+      "step": 2002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.238572337727598e-06,
+      "loss": 2.5886,
+      "step": 2003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.235280029193168e-06,
+      "loss": 2.657,
+      "step": 2004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2319865089830945e-06,
+      "loss": 2.6674,
+      "step": 2005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2286917788826926e-06,
+      "loss": 2.6483,
+      "step": 2006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.225395840677931e-06,
+      "loss": 2.6821,
+      "step": 2007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2220986961554375e-06,
+      "loss": 2.6107,
+      "step": 2008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.21880034710249e-06,
+      "loss": 2.6164,
+      "step": 2009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.21550079530702e-06,
+      "loss": 2.6794,
+      "step": 2010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.212200042557615e-06,
+      "loss": 2.7263,
+      "step": 2011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2088980906435055e-06,
+      "loss": 2.6384,
+      "step": 2012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.20559494135458e-06,
+      "loss": 2.6273,
+      "step": 2013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.202290596481372e-06,
+      "loss": 2.6932,
+      "step": 2014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.1989850578150646e-06,
+      "loss": 2.6641,
+      "step": 2015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.195678327147485e-06,
+      "loss": 2.7335,
+      "step": 2016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.192370406271112e-06,
+      "loss": 2.6202,
+      "step": 2017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.189061296979063e-06,
+      "loss": 2.7109,
+      "step": 2018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.185751001065106e-06,
+      "loss": 2.6939,
+      "step": 2019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.1824395203236465e-06,
+      "loss": 2.6725,
+      "step": 2020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.1791268565497375e-06,
+      "loss": 2.7043,
+      "step": 2021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.17581301153907e-06,
+      "loss": 2.6356,
+      "step": 2022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.172497987087976e-06,
+      "loss": 2.6149,
+      "step": 2023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.169181784993426e-06,
+      "loss": 2.7201,
+      "step": 2024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.165864407053032e-06,
+      "loss": 2.7348,
+      "step": 2025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.16254585506504e-06,
+      "loss": 2.6267,
+      "step": 2026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.159226130828333e-06,
+      "loss": 2.6369,
+      "step": 2027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.155905236142431e-06,
+      "loss": 2.676,
+      "step": 2028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.152583172807486e-06,
+      "loss": 2.6592,
+      "step": 2029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.149259942624287e-06,
+      "loss": 2.6929,
+      "step": 2030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.145935547394251e-06,
+      "loss": 2.6634,
+      "step": 2031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.1426099889194325e-06,
+      "loss": 2.6434,
+      "step": 2032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.13928326900251e-06,
+      "loss": 2.5939,
+      "step": 2033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.135955389446797e-06,
+      "loss": 2.7276,
+      "step": 2034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.132626352056232e-06,
+      "loss": 2.6581,
+      "step": 2035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.129296158635382e-06,
+      "loss": 2.6581,
+      "step": 2036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.125964810989443e-06,
+      "loss": 2.6372,
+      "step": 2037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.122632310924233e-06,
+      "loss": 2.6453,
+      "step": 2038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.1192986602461965e-06,
+      "loss": 2.663,
+      "step": 2039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.115963860762402e-06,
+      "loss": 2.6691,
+      "step": 2040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.112627914280541e-06,
+      "loss": 2.6749,
+      "step": 2041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.109290822608925e-06,
+      "loss": 2.6284,
+      "step": 2042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.105952587556488e-06,
+      "loss": 2.6296,
+      "step": 2043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.102613210932783e-06,
+      "loss": 2.6791,
+      "step": 2044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.099272694547983e-06,
+      "loss": 2.641,
+      "step": 2045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0959310402128755e-06,
+      "loss": 2.678,
+      "step": 2046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.092588249738871e-06,
+      "loss": 2.6695,
+      "step": 2047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.089244324937988e-06,
+      "loss": 2.7401,
+      "step": 2048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.085899267622867e-06,
+      "loss": 2.5922,
+      "step": 2049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.082553079606756e-06,
+      "loss": 2.6884,
+      "step": 2050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.079205762703523e-06,
+      "loss": 2.6745,
+      "step": 2051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.075857318727641e-06,
+      "loss": 2.6069,
+      "step": 2052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.072507749494199e-06,
+      "loss": 2.6158,
+      "step": 2053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.069157056818894e-06,
+      "loss": 2.6937,
+      "step": 2054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.065805242518031e-06,
+      "loss": 2.7328,
+      "step": 2055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.062452308408525e-06,
+      "loss": 2.6956,
+      "step": 2056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.059098256307897e-06,
+      "loss": 2.6338,
+      "step": 2057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.055743088034275e-06,
+      "loss": 2.5955,
+      "step": 2058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0523868054063885e-06,
+      "loss": 2.6106,
+      "step": 2059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.049029410243575e-06,
+      "loss": 2.6566,
+      "step": 2060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0456709043657756e-06,
+      "loss": 2.6138,
+      "step": 2061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.04231128959353e-06,
+      "loss": 2.7321,
+      "step": 2062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.03895056774798e-06,
+      "loss": 2.6864,
+      "step": 2063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.03558874065087e-06,
+      "loss": 2.6286,
+      "step": 2064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.03222581012454e-06,
+      "loss": 2.6111,
+      "step": 2065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.028861777991931e-06,
+      "loss": 2.6821,
+      "step": 2066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.025496646076579e-06,
+      "loss": 2.7185,
+      "step": 2067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.022130416202617e-06,
+      "loss": 2.6799,
+      "step": 2068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.018763090194775e-06,
+      "loss": 2.6479,
+      "step": 2069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.015394669878375e-06,
+      "loss": 2.621,
+      "step": 2070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.012025157079332e-06,
+      "loss": 2.666,
+      "step": 2071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.008654553624155e-06,
+      "loss": 2.5922,
+      "step": 2072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.005282861339939e-06,
+      "loss": 2.5893,
+      "step": 2073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.001910082054379e-06,
+      "loss": 2.6349,
+      "step": 2074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.998536217595752e-06,
+      "loss": 2.645,
+      "step": 2075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9951612697929215e-06,
+      "loss": 2.6657,
+      "step": 2076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.991785240475344e-06,
+      "loss": 2.6851,
+      "step": 2077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.988408131473061e-06,
+      "loss": 2.6584,
+      "step": 2078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9850299446166944e-06,
+      "loss": 2.6507,
+      "step": 2079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.981650681737457e-06,
+      "loss": 2.6118,
+      "step": 2080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.978270344667143e-06,
+      "loss": 2.6328,
+      "step": 2081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.974888935238126e-06,
+      "loss": 2.6844,
+      "step": 2082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.97150645528336e-06,
+      "loss": 2.6174,
+      "step": 2083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.968122906636386e-06,
+      "loss": 2.7061,
+      "step": 2084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.96473829113132e-06,
+      "loss": 2.7102,
+      "step": 2085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.961352610602854e-06,
+      "loss": 2.6913,
+      "step": 2086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.957965866886261e-06,
+      "loss": 2.633,
+      "step": 2087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.95457806181739e-06,
+      "loss": 2.6832,
+      "step": 2088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9511891972326665e-06,
+      "loss": 2.6589,
+      "step": 2089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.947799274969082e-06,
+      "loss": 2.6559,
+      "step": 2090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9444082968642146e-06,
+      "loss": 2.6971,
+      "step": 2091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.941016264756203e-06,
+      "loss": 2.6488,
+      "step": 2092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.937623180483764e-06,
+      "loss": 2.6468,
+      "step": 2093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.934229045886185e-06,
+      "loss": 2.6436,
+      "step": 2094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.930833862803318e-06,
+      "loss": 2.6045,
+      "step": 2095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.927437633075585e-06,
+      "loss": 2.6139,
+      "step": 2096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.924040358543978e-06,
+      "loss": 2.6462,
+      "step": 2097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.920642041050055e-06,
+      "loss": 2.6976,
+      "step": 2098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.917242682435937e-06,
+      "loss": 2.6934,
+      "step": 2099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.913842284544306e-06,
+      "loss": 2.6733,
+      "step": 2100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.910440849218419e-06,
+      "loss": 2.6655,
+      "step": 2101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9070383783020835e-06,
+      "loss": 2.7103,
+      "step": 2102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.903634873639673e-06,
+      "loss": 2.6257,
+      "step": 2103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.900230337076123e-06,
+      "loss": 2.6417,
+      "step": 2104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.896824770456926e-06,
+      "loss": 2.6461,
+      "step": 2105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.8934181756281345e-06,
+      "loss": 2.6562,
+      "step": 2106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.890010554436355e-06,
+      "loss": 2.7379,
+      "step": 2107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.886601908728755e-06,
+      "loss": 2.6968,
+      "step": 2108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.883192240353055e-06,
+      "loss": 2.6604,
+      "step": 2109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.879781551157527e-06,
+      "loss": 2.7208,
+      "step": 2110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.876369842991004e-06,
+      "loss": 2.564,
+      "step": 2111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.872957117702863e-06,
+      "loss": 2.7158,
+      "step": 2112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.869543377143039e-06,
+      "loss": 2.6613,
+      "step": 2113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.866128623162011e-06,
+      "loss": 2.6753,
+      "step": 2114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.862712857610812e-06,
+      "loss": 2.659,
+      "step": 2115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.859296082341023e-06,
+      "loss": 2.6723,
+      "step": 2116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.85587829920477e-06,
+      "loss": 2.589,
+      "step": 2117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.852459510054728e-06,
+      "loss": 2.6739,
+      "step": 2118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.849039716744117e-06,
+      "loss": 2.6916,
+      "step": 2119
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.845618921126698e-06,
+      "loss": 2.6627,
+      "step": 2120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.84219712505678e-06,
+      "loss": 2.6592,
+      "step": 2121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.838774330389209e-06,
+      "loss": 2.6186,
+      "step": 2122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.835350538979378e-06,
+      "loss": 2.6255,
+      "step": 2123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.831925752683215e-06,
+      "loss": 2.603,
+      "step": 2124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.828499973357194e-06,
+      "loss": 2.7174,
+      "step": 2125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.82507320285832e-06,
+      "loss": 2.6758,
+      "step": 2126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.821645443044138e-06,
+      "loss": 2.7024,
+      "step": 2127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.8182166957727315e-06,
+      "loss": 2.6239,
+      "step": 2128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.814786962902717e-06,
+      "loss": 2.6452,
+      "step": 2129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.811356246293246e-06,
+      "loss": 2.6782,
+      "step": 2130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.8079245478040035e-06,
+      "loss": 2.6748,
+      "step": 2131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.804491869295207e-06,
+      "loss": 2.7632,
+      "step": 2132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.801058212627602e-06,
+      "loss": 2.6223,
+      "step": 2133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.797623579662467e-06,
+      "loss": 2.6908,
+      "step": 2134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.794187972261612e-06,
+      "loss": 2.6903,
+      "step": 2135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.790751392287371e-06,
+      "loss": 2.7248,
+      "step": 2136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.787313841602607e-06,
+      "loss": 2.6355,
+      "step": 2137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.78387532207071e-06,
+      "loss": 2.6205,
+      "step": 2138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.780435835555592e-06,
+      "loss": 2.6875,
+      "step": 2139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.776995383921694e-06,
+      "loss": 2.727,
+      "step": 2140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.773553969033975e-06,
+      "loss": 2.6588,
+      "step": 2141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.770111592757923e-06,
+      "loss": 2.5947,
+      "step": 2142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.76666825695954e-06,
+      "loss": 2.6337,
+      "step": 2143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.76322396350535e-06,
+      "loss": 2.6863,
+      "step": 2144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.7597787142623995e-06,
+      "loss": 2.619,
+      "step": 2145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.756332511098249e-06,
+      "loss": 2.6504,
+      "step": 2146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.752885355880978e-06,
+      "loss": 2.6399,
+      "step": 2147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.749437250479184e-06,
+      "loss": 2.6415,
+      "step": 2148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.745988196761976e-06,
+      "loss": 2.6324,
+      "step": 2149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.742538196598978e-06,
+      "loss": 2.598,
+      "step": 2150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.739087251860327e-06,
+      "loss": 2.641,
+      "step": 2151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.735635364416674e-06,
+      "loss": 2.6891,
+      "step": 2152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.73218253613918e-06,
+      "loss": 2.6004,
+      "step": 2153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.728728768899513e-06,
+      "loss": 2.6119,
+      "step": 2154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.725274064569856e-06,
+      "loss": 2.6917,
+      "step": 2155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.721818425022897e-06,
+      "loss": 2.7065,
+      "step": 2156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.718361852131825e-06,
+      "loss": 2.7026,
+      "step": 2157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.714904347770346e-06,
+      "loss": 2.6372,
+      "step": 2158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.711445913812663e-06,
+      "loss": 2.6499,
+      "step": 2159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.707986552133485e-06,
+      "loss": 2.7222,
+      "step": 2160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.704526264608026e-06,
+      "loss": 2.6491,
+      "step": 2161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.701065053111997e-06,
+      "loss": 2.6262,
+      "step": 2162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.697602919521617e-06,
+      "loss": 2.69,
+      "step": 2163
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.694139865713596e-06,
+      "loss": 2.7199,
+      "step": 2164
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.69067589356515e-06,
+      "loss": 2.6462,
+      "step": 2165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.687211004953992e-06,
+      "loss": 2.5932,
+      "step": 2166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.6837452017583274e-06,
+      "loss": 2.6422,
+      "step": 2167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.680278485856861e-06,
+      "loss": 2.6831,
+      "step": 2168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.676810859128793e-06,
+      "loss": 2.6163,
+      "step": 2169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.673342323453813e-06,
+      "loss": 2.6077,
+      "step": 2170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.6698728807121075e-06,
+      "loss": 2.6895,
+      "step": 2171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.666402532784354e-06,
+      "loss": 2.6931,
+      "step": 2172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.662931281551719e-06,
+      "loss": 2.7252,
+      "step": 2173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.659459128895859e-06,
+      "loss": 2.6353,
+      "step": 2174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.655986076698922e-06,
+      "loss": 2.6958,
+      "step": 2175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.652512126843538e-06,
+      "loss": 2.6665,
+      "step": 2176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.64903728121283e-06,
+      "loss": 2.6537,
+      "step": 2177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.645561541690402e-06,
+      "loss": 2.6111,
+      "step": 2178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.642084910160346e-06,
+      "loss": 2.6802,
+      "step": 2179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.6386073885072325e-06,
+      "loss": 2.6797,
+      "step": 2180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.635128978616119e-06,
+      "loss": 2.6137,
+      "step": 2181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.631649682372545e-06,
+      "loss": 2.635,
+      "step": 2182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.628169501662527e-06,
+      "loss": 2.5967,
+      "step": 2183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.62468843837256e-06,
+      "loss": 2.6551,
+      "step": 2184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.6212064943896244e-06,
+      "loss": 2.6702,
+      "step": 2185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.617723671601171e-06,
+      "loss": 2.625,
+      "step": 2186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.614239971895129e-06,
+      "loss": 2.6636,
+      "step": 2187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.610755397159905e-06,
+      "loss": 2.6584,
+      "step": 2188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.607269949284377e-06,
+      "loss": 2.62,
+      "step": 2189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.603783630157901e-06,
+      "loss": 2.6503,
+      "step": 2190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.600296441670296e-06,
+      "loss": 2.7038,
+      "step": 2191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.596808385711864e-06,
+      "loss": 2.6608,
+      "step": 2192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5933194641733685e-06,
+      "loss": 2.6939,
+      "step": 2193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.589829678946045e-06,
+      "loss": 2.6359,
+      "step": 2194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.586339031921598e-06,
+      "loss": 2.6563,
+      "step": 2195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5828475249922e-06,
+      "loss": 2.6107,
+      "step": 2196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.579355160050486e-06,
+      "loss": 2.6401,
+      "step": 2197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.575861938989556e-06,
+      "loss": 2.6106,
+      "step": 2198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.572367863702982e-06,
+      "loss": 2.632,
+      "step": 2199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.568872936084789e-06,
+      "loss": 2.6863,
+      "step": 2200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.56537715802947e-06,
+      "loss": 2.6323,
+      "step": 2201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.561880531431977e-06,
+      "loss": 2.6323,
+      "step": 2202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.558383058187723e-06,
+      "loss": 2.6985,
+      "step": 2203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.554884740192578e-06,
+      "loss": 2.6375,
+      "step": 2204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5513855793428704e-06,
+      "loss": 2.6678,
+      "step": 2205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.54788557753539e-06,
+      "loss": 2.7157,
+      "step": 2206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.544384736667375e-06,
+      "loss": 2.6997,
+      "step": 2207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.540883058636524e-06,
+      "loss": 2.6676,
+      "step": 2208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.537380545340987e-06,
+      "loss": 2.7182,
+      "step": 2209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.533877198679367e-06,
+      "loss": 2.5994,
+      "step": 2210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.530373020550719e-06,
+      "loss": 2.6906,
+      "step": 2211
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.526868012854549e-06,
+      "loss": 2.5979,
+      "step": 2212
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.523362177490813e-06,
+      "loss": 2.5426,
+      "step": 2213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5198555163599145e-06,
+      "loss": 2.6567,
+      "step": 2214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.516348031362705e-06,
+      "loss": 2.6364,
+      "step": 2215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.512839724400483e-06,
+      "loss": 2.7129,
+      "step": 2216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.509330597374993e-06,
+      "loss": 2.6228,
+      "step": 2217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.505820652188423e-06,
+      "loss": 2.6523,
+      "step": 2218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.502309890743405e-06,
+      "loss": 2.6646,
+      "step": 2219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.498798314943014e-06,
+      "loss": 2.6894,
+      "step": 2220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.495285926690762e-06,
+      "loss": 2.677,
+      "step": 2221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.49177272789061e-06,
+      "loss": 2.629,
+      "step": 2222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.4882587204469526e-06,
+      "loss": 2.6288,
+      "step": 2223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.484743906264621e-06,
+      "loss": 2.6987,
+      "step": 2224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.481228287248891e-06,
+      "loss": 2.7451,
+      "step": 2225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.477711865305465e-06,
+      "loss": 2.597,
+      "step": 2226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.47419464234049e-06,
+      "loss": 2.608,
+      "step": 2227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.470676620260539e-06,
+      "loss": 2.6789,
+      "step": 2228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.467157800972626e-06,
+      "loss": 2.7145,
+      "step": 2229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.463638186384191e-06,
+      "loss": 2.6441,
+      "step": 2230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.460117778403107e-06,
+      "loss": 2.6588,
+      "step": 2231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.456596578937682e-06,
+      "loss": 2.6432,
+      "step": 2232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.453074589896642e-06,
+      "loss": 2.6297,
+      "step": 2233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.44955181318915e-06,
+      "loss": 2.6008,
+      "step": 2234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.4460282507247955e-06,
+      "loss": 2.6748,
+      "step": 2235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.44250390441359e-06,
+      "loss": 2.669,
+      "step": 2236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.438978776165971e-06,
+      "loss": 2.6262,
+      "step": 2237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.435452867892803e-06,
+      "loss": 2.6532,
+      "step": 2238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.431926181505368e-06,
+      "loss": 2.6584,
+      "step": 2239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.428398718915377e-06,
+      "loss": 2.6827,
+      "step": 2240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.424870482034953e-06,
+      "loss": 2.7095,
+      "step": 2241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.421341472776646e-06,
+      "loss": 2.6855,
+      "step": 2242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.417811693053422e-06,
+      "loss": 2.6938,
+      "step": 2243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.414281144778662e-06,
+      "loss": 2.6846,
+      "step": 2244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.41074982986617e-06,
+      "loss": 2.7143,
+      "step": 2245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.407217750230159e-06,
+      "loss": 2.6086,
+      "step": 2246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.40368490778526e-06,
+      "loss": 2.6131,
+      "step": 2247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.4001513044465165e-06,
+      "loss": 2.6665,
+      "step": 2248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3966169421293865e-06,
+      "loss": 2.6659,
+      "step": 2249
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.393081822749735e-06,
+      "loss": 2.6735,
+      "step": 2250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.389545948223841e-06,
+      "loss": 2.6734,
+      "step": 2251
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.386009320468393e-06,
+      "loss": 2.7476,
+      "step": 2252
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.382471941400487e-06,
+      "loss": 2.761,
+      "step": 2253
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.378933812937624e-06,
+      "loss": 2.6509,
+      "step": 2254
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3753949369977144e-06,
+      "loss": 2.7778,
+      "step": 2255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3718553154990715e-06,
+      "loss": 2.6281,
+      "step": 2256
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.368314950360416e-06,
+      "loss": 2.6756,
+      "step": 2257
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.364773843500866e-06,
+      "loss": 2.691,
+      "step": 2258
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.361231996839948e-06,
+      "loss": 2.6471,
+      "step": 2259
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3576894122975875e-06,
+      "loss": 2.6601,
+      "step": 2260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.354146091794105e-06,
+      "loss": 2.6104,
+      "step": 2261
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.350602037250228e-06,
+      "loss": 2.6084,
+      "step": 2262
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.347057250587076e-06,
+      "loss": 2.6456,
+      "step": 2263
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.343511733726169e-06,
+      "loss": 2.6334,
+      "step": 2264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.33996548858942e-06,
+      "loss": 2.6882,
+      "step": 2265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.336418517099139e-06,
+      "loss": 2.6631,
+      "step": 2266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.332870821178028e-06,
+      "loss": 2.6559,
+      "step": 2267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.329322402749181e-06,
+      "loss": 2.7163,
+      "step": 2268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.325773263736088e-06,
+      "loss": 2.6896,
+      "step": 2269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.322223406062624e-06,
+      "loss": 2.7365,
+      "step": 2270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.318672831653058e-06,
+      "loss": 2.6482,
+      "step": 2271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.3151215424320456e-06,
+      "loss": 2.6986,
+      "step": 2272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.3115695403246295e-06,
+      "loss": 2.6829,
+      "step": 2273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.308016827256237e-06,
+      "loss": 2.6327,
+      "step": 2274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.304463405152687e-06,
+      "loss": 2.7324,
+      "step": 2275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.300909275940177e-06,
+      "loss": 2.6246,
+      "step": 2276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2973544415452904e-06,
+      "loss": 2.6727,
+      "step": 2277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.293798903894991e-06,
+      "loss": 2.6373,
+      "step": 2278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.290242664916627e-06,
+      "loss": 2.5985,
+      "step": 2279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.286685726537923e-06,
+      "loss": 2.6756,
+      "step": 2280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2831280906869845e-06,
+      "loss": 2.6671,
+      "step": 2281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.279569759292297e-06,
+      "loss": 2.656,
+      "step": 2282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.276010734282718e-06,
+      "loss": 2.6263,
+      "step": 2283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.272451017587485e-06,
+      "loss": 2.6064,
+      "step": 2284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2688906111362115e-06,
+      "loss": 2.6916,
+      "step": 2285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.26532951685888e-06,
+      "loss": 2.5983,
+      "step": 2286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.261767736685851e-06,
+      "loss": 2.7233,
+      "step": 2287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2582052725478526e-06,
+      "loss": 2.6235,
+      "step": 2288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.254642126375987e-06,
+      "loss": 2.6657,
+      "step": 2289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.251078300101725e-06,
+      "loss": 2.7661,
+      "step": 2290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2475137956569055e-06,
+      "loss": 2.6466,
+      "step": 2291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.243948614973734e-06,
+      "loss": 2.6415,
+      "step": 2292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2403827599847845e-06,
+      "loss": 2.6358,
+      "step": 2293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2368162326229976e-06,
+      "loss": 2.6638,
+      "step": 2294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.233249034821673e-06,
+      "loss": 2.6445,
+      "step": 2295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.229681168514481e-06,
+      "loss": 2.6037,
+      "step": 2296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.22611263563545e-06,
+      "loss": 2.6929,
+      "step": 2297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.222543438118968e-06,
+      "loss": 2.6239,
+      "step": 2298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.218973577899787e-06,
+      "loss": 2.633,
+      "step": 2299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.215403056913018e-06,
+      "loss": 2.5738,
+      "step": 2300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.211831877094131e-06,
+      "loss": 2.7019,
+      "step": 2301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.208260040378946e-06,
+      "loss": 2.5788,
+      "step": 2302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.204687548703649e-06,
+      "loss": 2.7252,
+      "step": 2303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2011144040047755e-06,
+      "loss": 2.6874,
+      "step": 2304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.1975406082192124e-06,
+      "loss": 2.6715,
+      "step": 2305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.193966163284207e-06,
+      "loss": 2.6905,
+      "step": 2306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.190391071137355e-06,
+      "loss": 2.7034,
+      "step": 2307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.186815333716598e-06,
+      "loss": 2.6329,
+      "step": 2308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.183238952960236e-06,
+      "loss": 2.6543,
+      "step": 2309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.179661930806912e-06,
+      "loss": 2.7285,
+      "step": 2310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.176084269195617e-06,
+      "loss": 2.671,
+      "step": 2311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.172505970065692e-06,
+      "loss": 2.6806,
+      "step": 2312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.1689270353568205e-06,
+      "loss": 2.6669,
+      "step": 2313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.1653474670090315e-06,
+      "loss": 2.7325,
+      "step": 2314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.1617672669626975e-06,
+      "loss": 2.5708,
+      "step": 2315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.1581864371585345e-06,
+      "loss": 2.7672,
+      "step": 2316
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.154604979537597e-06,
+      "loss": 2.6032,
+      "step": 2317
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.151022896041281e-06,
+      "loss": 2.6126,
+      "step": 2318
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.147440188611324e-06,
+      "loss": 2.643,
+      "step": 2319
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.143856859189801e-06,
+      "loss": 2.6268,
+      "step": 2320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.14027290971912e-06,
+      "loss": 2.6439,
+      "step": 2321
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.136688342142031e-06,
+      "loss": 2.6419,
+      "step": 2322
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.133103158401614e-06,
+      "loss": 2.732,
+      "step": 2323
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.1295173604412896e-06,
+      "loss": 2.6553,
+      "step": 2324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.1259309502048015e-06,
+      "loss": 2.5888,
+      "step": 2325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.122343929636236e-06,
+      "loss": 2.6877,
+      "step": 2326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.118756300680002e-06,
+      "loss": 2.7013,
+      "step": 2327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.115168065280842e-06,
+      "loss": 2.7469,
+      "step": 2328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.111579225383828e-06,
+      "loss": 2.6236,
+      "step": 2329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.107989782934356e-06,
+      "loss": 2.6571,
+      "step": 2330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.104399739878153e-06,
+      "loss": 2.6487,
+      "step": 2331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.100809098161268e-06,
+      "loss": 2.6545,
+      "step": 2332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.097217859730077e-06,
+      "loss": 2.6378,
+      "step": 2333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.0936260265312774e-06,
+      "loss": 2.6689,
+      "step": 2334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.09003360051189e-06,
+      "loss": 2.6799,
+      "step": 2335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.0864405836192575e-06,
+      "loss": 2.5981,
+      "step": 2336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.082846977801043e-06,
+      "loss": 2.7446,
+      "step": 2337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.0792527850052284e-06,
+      "loss": 2.6462,
+      "step": 2338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.075658007180112e-06,
+      "loss": 2.6056,
+      "step": 2339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.072062646274312e-06,
+      "loss": 2.6472,
+      "step": 2340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.068466704236761e-06,
+      "loss": 2.6964,
+      "step": 2341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.064870183016707e-06,
+      "loss": 2.6976,
+      "step": 2342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.061273084563712e-06,
+      "loss": 2.6996,
+      "step": 2343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.057675410827654e-06,
+      "loss": 2.591,
+      "step": 2344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.054077163758713e-06,
+      "loss": 2.612,
+      "step": 2345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.050478345307395e-06,
+      "loss": 2.7057,
+      "step": 2346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.0468789574245e-06,
+      "loss": 2.5794,
+      "step": 2347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.043279002061149e-06,
+      "loss": 2.608,
+      "step": 2348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.039678481168761e-06,
+      "loss": 2.7182,
+      "step": 2349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.036077396699072e-06,
+      "loss": 2.6355,
+      "step": 2350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.032475750604115e-06,
+      "loss": 2.684,
+      "step": 2351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.028873544836225e-06,
+      "loss": 2.6642,
+      "step": 2352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.025270781348055e-06,
+      "loss": 2.6382,
+      "step": 2353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.0216674620925445e-06,
+      "loss": 2.6262,
+      "step": 2354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.01806358902294e-06,
+      "loss": 2.7556,
+      "step": 2355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.014459164092793e-06,
+      "loss": 2.7119,
+      "step": 2356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.010854189255948e-06,
+      "loss": 2.6821,
+      "step": 2357
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.007248666466548e-06,
+      "loss": 2.6781,
+      "step": 2358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.003642597679036e-06,
+      "loss": 2.7182,
+      "step": 2359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.00003598484815e-06,
+      "loss": 2.7197,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.996428829928921e-06,
+      "loss": 2.5993,
+      "step": 2361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.992821134876675e-06,
+      "loss": 2.6254,
+      "step": 2362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.989212901647034e-06,
+      "loss": 2.6326,
+      "step": 2363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.985604132195907e-06,
+      "loss": 2.6796,
+      "step": 2364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9819948284794934e-06,
+      "loss": 2.6547,
+      "step": 2365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.978384992454287e-06,
+      "loss": 2.6236,
+      "step": 2366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.974774626077068e-06,
+      "loss": 2.6473,
+      "step": 2367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.971163731304901e-06,
+      "loss": 2.6648,
+      "step": 2368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.967552310095139e-06,
+      "loss": 2.7312,
+      "step": 2369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.963940364405425e-06,
+      "loss": 2.669,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.960327896193677e-06,
+      "loss": 2.6204,
+      "step": 2371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9567149074181065e-06,
+      "loss": 2.6389,
+      "step": 2372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.953101400037198e-06,
+      "loss": 2.6826,
+      "step": 2373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.949487376009723e-06,
+      "loss": 2.6967,
+      "step": 2374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9458728372947325e-06,
+      "loss": 2.6348,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.942257785851554e-06,
+      "loss": 2.7018,
+      "step": 2376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.938642223639794e-06,
+      "loss": 2.566,
+      "step": 2377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.935026152619338e-06,
+      "loss": 2.6348,
+      "step": 2378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.931409574750341e-06,
+      "loss": 2.6891,
+      "step": 2379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.927792491993242e-06,
+      "loss": 2.6055,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.924174906308746e-06,
+      "loss": 2.7235,
+      "step": 2381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.920556819657834e-06,
+      "loss": 2.6771,
+      "step": 2382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.916938234001759e-06,
+      "loss": 2.6831,
+      "step": 2383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.91331915130204e-06,
+      "loss": 2.6476,
+      "step": 2384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.909699573520475e-06,
+      "loss": 2.6575,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.90607950261912e-06,
+      "loss": 2.6737,
+      "step": 2386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.902458940560304e-06,
+      "loss": 2.718,
+      "step": 2387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8988378893066216e-06,
+      "loss": 2.676,
+      "step": 2388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.895216350820928e-06,
+      "loss": 2.6771,
+      "step": 2389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.891594327066351e-06,
+      "loss": 2.6621,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.887971820006275e-06,
+      "loss": 2.5823,
+      "step": 2391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8843488316043465e-06,
+      "loss": 2.6023,
+      "step": 2392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.880725363824477e-06,
+      "loss": 2.649,
+      "step": 2393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.877101418630836e-06,
+      "loss": 2.6288,
+      "step": 2394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.873476997987848e-06,
+      "loss": 2.5761,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.869852103860203e-06,
+      "loss": 2.5876,
+      "step": 2396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.866226738212838e-06,
+      "loss": 2.6608,
+      "step": 2397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.862600903010955e-06,
+      "loss": 2.7348,
+      "step": 2398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.858974600220002e-06,
+      "loss": 2.5687,
+      "step": 2399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.85534783180569e-06,
+      "loss": 2.6701,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8517205997339745e-06,
+      "loss": 2.665,
+      "step": 2401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.848092905971062e-06,
+      "loss": 2.6485,
+      "step": 2402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8444647524834175e-06,
+      "loss": 2.6918,
+      "step": 2403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8408361412377475e-06,
+      "loss": 2.6414,
+      "step": 2404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.837207074201009e-06,
+      "loss": 2.5748,
+      "step": 2405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.833577553340406e-06,
+      "loss": 2.7079,
+      "step": 2406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.829947580623389e-06,
+      "loss": 2.6633,
+      "step": 2407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.826317158017652e-06,
+      "loss": 2.5891,
+      "step": 2408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.822686287491135e-06,
+      "loss": 2.7186,
+      "step": 2409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.819054971012021e-06,
+      "loss": 2.6201,
+      "step": 2410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.815423210548733e-06,
+      "loss": 2.7069,
+      "step": 2411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.811791008069932e-06,
+      "loss": 2.6388,
+      "step": 2412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.808158365544526e-06,
+      "loss": 2.5819,
+      "step": 2413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.804525284941656e-06,
+      "loss": 2.757,
+      "step": 2414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.8008917682307e-06,
+      "loss": 2.6321,
+      "step": 2415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.797257817381275e-06,
+      "loss": 2.6303,
+      "step": 2416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.793623434363234e-06,
+      "loss": 2.7753,
+      "step": 2417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.789988621146662e-06,
+      "loss": 2.6921,
+      "step": 2418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.786353379701876e-06,
+      "loss": 2.6491,
+      "step": 2419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7827177119994295e-06,
+      "loss": 2.6384,
+      "step": 2420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.779081620010104e-06,
+      "loss": 2.7265,
+      "step": 2421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.775445105704911e-06,
+      "loss": 2.6475,
+      "step": 2422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.771808171055093e-06,
+      "loss": 2.6921,
+      "step": 2423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7681708180321186e-06,
+      "loss": 2.6509,
+      "step": 2424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764533048607683e-06,
+      "loss": 2.6544,
+      "step": 2425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.760894864753705e-06,
+      "loss": 2.5806,
+      "step": 2426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.757256268442336e-06,
+      "loss": 2.7059,
+      "step": 2427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.753617261645943e-06,
+      "loss": 2.6141,
+      "step": 2428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.749977846337117e-06,
+      "loss": 2.6742,
+      "step": 2429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.746338024488674e-06,
+      "loss": 2.5911,
+      "step": 2430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.742697798073645e-06,
+      "loss": 2.7477,
+      "step": 2431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7390571690652865e-06,
+      "loss": 2.6306,
+      "step": 2432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.735416139437067e-06,
+      "loss": 2.5899,
+      "step": 2433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.731774711162679e-06,
+      "loss": 2.7124,
+      "step": 2434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.728132886216024e-06,
+      "loss": 2.589,
+      "step": 2435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.724490666571223e-06,
+      "loss": 2.6137,
+      "step": 2436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.72084805420261e-06,
+      "loss": 2.6335,
+      "step": 2437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.717205051084731e-06,
+      "loss": 2.6438,
+      "step": 2438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.713561659192344e-06,
+      "loss": 2.6924,
+      "step": 2439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.709917880500421e-06,
+      "loss": 2.6231,
+      "step": 2440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.706273716984138e-06,
+      "loss": 2.6835,
+      "step": 2441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.702629170618883e-06,
+      "loss": 2.6564,
+      "step": 2442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.69898424338025e-06,
+      "loss": 2.6796,
+      "step": 2443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.695338937244043e-06,
+      "loss": 2.6499,
+      "step": 2444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.691693254186268e-06,
+      "loss": 2.6737,
+      "step": 2445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.688047196183135e-06,
+      "loss": 2.5916,
+      "step": 2446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.68440076521106e-06,
+      "loss": 2.6264,
+      "step": 2447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.680753963246659e-06,
+      "loss": 2.6874,
+      "step": 2448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.677106792266748e-06,
+      "loss": 2.6396,
+      "step": 2449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.673459254248348e-06,
+      "loss": 2.6445,
+      "step": 2450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.669811351168674e-06,
+      "loss": 2.5818,
+      "step": 2451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.66616308500514e-06,
+      "loss": 2.677,
+      "step": 2452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.662514457735357e-06,
+      "loss": 2.6552,
+      "step": 2453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6588654713371335e-06,
+      "loss": 2.6647,
+      "step": 2454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.655216127788472e-06,
+      "loss": 2.5916,
+      "step": 2455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.651566429067566e-06,
+      "loss": 2.6386,
+      "step": 2456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.647916377152803e-06,
+      "loss": 2.6703,
+      "step": 2457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.644265974022766e-06,
+      "loss": 2.6478,
+      "step": 2458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.64061522165622e-06,
+      "loss": 2.7191,
+      "step": 2459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.6369641220321285e-06,
+      "loss": 2.6172,
+      "step": 2460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.633312677129635e-06,
+      "loss": 2.656,
+      "step": 2461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.629660888928078e-06,
+      "loss": 2.6756,
+      "step": 2462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.626008759406973e-06,
+      "loss": 2.685,
+      "step": 2463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.622356290546031e-06,
+      "loss": 2.6683,
+      "step": 2464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.618703484325138e-06,
+      "loss": 2.7485,
+      "step": 2465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.615050342724368e-06,
+      "loss": 2.6274,
+      "step": 2466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.611396867723974e-06,
+      "loss": 2.7015,
+      "step": 2467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.607743061304391e-06,
+      "loss": 2.7141,
+      "step": 2468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.604088925446234e-06,
+      "loss": 2.6667,
+      "step": 2469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.600434462130296e-06,
+      "loss": 2.6072,
+      "step": 2470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.59677967333755e-06,
+      "loss": 2.6509,
+      "step": 2471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.593124561049141e-06,
+      "loss": 2.6117,
+      "step": 2472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.58946912724639e-06,
+      "loss": 2.5953,
+      "step": 2473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.585813373910797e-06,
+      "loss": 2.6499,
+      "step": 2474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.58215730302403e-06,
+      "loss": 2.5894,
+      "step": 2475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.578500916567929e-06,
+      "loss": 2.6359,
+      "step": 2476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.574844216524512e-06,
+      "loss": 2.708,
+      "step": 2477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.57118720487596e-06,
+      "loss": 2.6469,
+      "step": 2478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.567529883604623e-06,
+      "loss": 2.7058,
+      "step": 2479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.563872254693023e-06,
+      "loss": 2.6455,
+      "step": 2480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5602143201238455e-06,
+      "loss": 2.6648,
+      "step": 2481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.556556081879945e-06,
+      "loss": 2.6647,
+      "step": 2482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.552897541944335e-06,
+      "loss": 2.7041,
+      "step": 2483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.549238702300201e-06,
+      "loss": 2.6327,
+      "step": 2484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.545579564930882e-06,
+      "loss": 2.6958,
+      "step": 2485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.541920131819883e-06,
+      "loss": 2.6781,
+      "step": 2486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.538260404950868e-06,
+      "loss": 2.6084,
+      "step": 2487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5346003863076625e-06,
+      "loss": 2.6837,
+      "step": 2488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.530940077874248e-06,
+      "loss": 2.6415,
+      "step": 2489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.527279481634763e-06,
+      "loss": 2.6302,
+      "step": 2490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.523618599573505e-06,
+      "loss": 2.6959,
+      "step": 2491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.519957433674921e-06,
+      "loss": 2.6286,
+      "step": 2492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.516295985923614e-06,
+      "loss": 2.6848,
+      "step": 2493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.512634258304345e-06,
+      "loss": 2.6512,
+      "step": 2494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.508972252802021e-06,
+      "loss": 2.6381,
+      "step": 2495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.505309971401698e-06,
+      "loss": 2.6623,
+      "step": 2496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5016474160885895e-06,
+      "loss": 2.664,
+      "step": 2497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.497984588848049e-06,
+      "loss": 2.6386,
+      "step": 2498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.494321491665582e-06,
+      "loss": 2.7303,
+      "step": 2499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.490658126526839e-06,
+      "loss": 2.6077,
+      "step": 2500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.486994495417617e-06,
+      "loss": 2.661,
+      "step": 2501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.483330600323856e-06,
+      "loss": 2.6634,
+      "step": 2502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.479666443231638e-06,
+      "loss": 2.7236,
+      "step": 2503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.476002026127191e-06,
+      "loss": 2.6422,
+      "step": 2504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4723373509968784e-06,
+      "loss": 2.5998,
+      "step": 2505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.468672419827208e-06,
+      "loss": 2.6579,
+      "step": 2506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.465007234604823e-06,
+      "loss": 2.6539,
+      "step": 2507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.46134179731651e-06,
+      "loss": 2.6315,
+      "step": 2508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.457676109949186e-06,
+      "loss": 2.6266,
+      "step": 2509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.454010174489904e-06,
+      "loss": 2.6705,
+      "step": 2510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4503439929258575e-06,
+      "loss": 2.6654,
+      "step": 2511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.446677567244366e-06,
+      "loss": 2.6761,
+      "step": 2512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.443010899432885e-06,
+      "loss": 2.6101,
+      "step": 2513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.439343991479001e-06,
+      "loss": 2.7103,
+      "step": 2514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.435676845370432e-06,
+      "loss": 2.7172,
+      "step": 2515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.432009463095022e-06,
+      "loss": 2.6422,
+      "step": 2516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4283418466407454e-06,
+      "loss": 2.6411,
+      "step": 2517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.424673997995702e-06,
+      "loss": 2.6572,
+      "step": 2518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.421005919148119e-06,
+      "loss": 2.7137,
+      "step": 2519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.417337612086348e-06,
+      "loss": 2.5854,
+      "step": 2520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.413669078798863e-06,
+      "loss": 2.6196,
+      "step": 2521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.410000321274262e-06,
+      "loss": 2.6426,
+      "step": 2522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.406331341501264e-06,
+      "loss": 2.6591,
+      "step": 2523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.402662141468709e-06,
+      "loss": 2.6216,
+      "step": 2524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.398992723165556e-06,
+      "loss": 2.6204,
+      "step": 2525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.395323088580881e-06,
+      "loss": 2.6852,
+      "step": 2526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.39165323970388e-06,
+      "loss": 2.6667,
+      "step": 2527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.387983178523863e-06,
+      "loss": 2.6476,
+      "step": 2528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.384312907030257e-06,
+      "loss": 2.6053,
+      "step": 2529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3806424272126e-06,
+      "loss": 2.6773,
+      "step": 2530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3769717410605445e-06,
+      "loss": 2.6699,
+      "step": 2531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.373300850563858e-06,
+      "loss": 2.6606,
+      "step": 2532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.369629757712412e-06,
+      "loss": 2.6871,
+      "step": 2533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.365958464496196e-06,
+      "loss": 2.5946,
+      "step": 2534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3622869729052996e-06,
+      "loss": 2.6361,
+      "step": 2535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.358615284929927e-06,
+      "loss": 2.6942,
+      "step": 2536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.354943402560382e-06,
+      "loss": 2.6008,
+      "step": 2537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3512713277870806e-06,
+      "loss": 2.6507,
+      "step": 2538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.34759906260054e-06,
+      "loss": 2.6595,
+      "step": 2539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.34392660899138e-06,
+      "loss": 2.6514,
+      "step": 2540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.340253968950323e-06,
+      "loss": 2.6078,
+      "step": 2541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3365811444681925e-06,
+      "loss": 2.6501,
+      "step": 2542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.332908137535915e-06,
+      "loss": 2.6146,
+      "step": 2543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3292349501445086e-06,
+      "loss": 2.6602,
+      "step": 2544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.325561584285097e-06,
+      "loss": 2.6768,
+      "step": 2545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.321888041948897e-06,
+      "loss": 2.7354,
+      "step": 2546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.318214325127219e-06,
+      "loss": 2.6446,
+      "step": 2547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.314540435811474e-06,
+      "loss": 2.6017,
+      "step": 2548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.31086637599316e-06,
+      "loss": 2.68,
+      "step": 2549
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.307192147663871e-06,
+      "loss": 2.7121,
+      "step": 2550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.303517752815294e-06,
+      "loss": 2.6497,
+      "step": 2551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.299843193439202e-06,
+      "loss": 2.6255,
+      "step": 2552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.296168471527457e-06,
+      "loss": 2.6539,
+      "step": 2553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.292493589072014e-06,
+      "loss": 2.6317,
+      "step": 2554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2888185480649115e-06,
+      "loss": 2.6419,
+      "step": 2555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.285143350498275e-06,
+      "loss": 2.6579,
+      "step": 2556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.281467998364314e-06,
+      "loss": 2.6572,
+      "step": 2557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.277792493655325e-06,
+      "loss": 2.6325,
+      "step": 2558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.274116838363681e-06,
+      "loss": 2.637,
+      "step": 2559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.270441034481841e-06,
+      "loss": 2.7018,
+      "step": 2560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.266765084002345e-06,
+      "loss": 2.6837,
+      "step": 2561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.263088988917809e-06,
+      "loss": 2.706,
+      "step": 2562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.259412751220932e-06,
+      "loss": 2.6659,
+      "step": 2563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.255736372904486e-06,
+      "loss": 2.7235,
+      "step": 2564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.252059855961325e-06,
+      "loss": 2.5623,
+      "step": 2565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.248383202384367e-06,
+      "loss": 2.6232,
+      "step": 2566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.244706414166618e-06,
+      "loss": 2.6269,
+      "step": 2567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.241029493301148e-06,
+      "loss": 2.6381,
+      "step": 2568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2373524417811005e-06,
+      "loss": 2.7007,
+      "step": 2569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.233675261599691e-06,
+      "loss": 2.6554,
+      "step": 2570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.229997954750206e-06,
+      "loss": 2.5916,
+      "step": 2571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.226320523225996e-06,
+      "loss": 2.7395,
+      "step": 2572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.222642969020484e-06,
+      "loss": 2.5754,
+      "step": 2573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.218965294127155e-06,
+      "loss": 2.6858,
+      "step": 2574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.215287500539567e-06,
+      "loss": 2.6783,
+      "step": 2575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.211609590251332e-06,
+      "loss": 2.638,
+      "step": 2576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.207931565256133e-06,
+      "loss": 2.6539,
+      "step": 2577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.204253427547713e-06,
+      "loss": 2.6863,
+      "step": 2578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.200575179119874e-06,
+      "loss": 2.6633,
+      "step": 2579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.196896821966482e-06,
+      "loss": 2.6165,
+      "step": 2580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1932183580814596e-06,
+      "loss": 2.6133,
+      "step": 2581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.189539789458786e-06,
+      "loss": 2.6816,
+      "step": 2582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1858611180925e-06,
+      "loss": 2.5606,
+      "step": 2583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.182182345976694e-06,
+      "loss": 2.6753,
+      "step": 2584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.178503475105515e-06,
+      "loss": 2.5845,
+      "step": 2585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.174824507473166e-06,
+      "loss": 2.6456,
+      "step": 2586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1711454450738985e-06,
+      "loss": 2.6267,
+      "step": 2587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.16746628990202e-06,
+      "loss": 2.629,
+      "step": 2588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.163787043951885e-06,
+      "loss": 2.7306,
+      "step": 2589
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.160107709217896e-06,
+      "loss": 2.7194,
+      "step": 2590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.156428287694508e-06,
+      "loss": 2.6611,
+      "step": 2591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.15274878137622e-06,
+      "loss": 2.6495,
+      "step": 2592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.149069192257579e-06,
+      "loss": 2.6636,
+      "step": 2593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.145389522333173e-06,
+      "loss": 2.7155,
+      "step": 2594
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1417097735976385e-06,
+      "loss": 2.6599,
+      "step": 2595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.138029948045653e-06,
+      "loss": 2.6632,
+      "step": 2596
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.134350047671932e-06,
+      "loss": 2.7116,
+      "step": 2597
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.130670074471239e-06,
+      "loss": 2.7048,
+      "step": 2598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.1269900304383716e-06,
+      "loss": 2.5873,
+      "step": 2599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.1233099175681635e-06,
+      "loss": 2.6369,
+      "step": 2600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.119629737855494e-06,
+      "loss": 2.617,
+      "step": 2601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.115949493295272e-06,
+      "loss": 2.6502,
+      "step": 2602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.11226918588244e-06,
+      "loss": 2.6235,
+      "step": 2603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.108588817611982e-06,
+      "loss": 2.7326,
+      "step": 2604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.104908390478912e-06,
+      "loss": 2.6299,
+      "step": 2605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.1012279064782735e-06,
+      "loss": 2.6549,
+      "step": 2606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.097547367605139e-06,
+      "loss": 2.5896,
+      "step": 2607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.093866775854618e-06,
+      "loss": 2.6411,
+      "step": 2608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0901861332218426e-06,
+      "loss": 2.7401,
+      "step": 2609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.086505441701975e-06,
+      "loss": 2.5355,
+      "step": 2610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.082824703290203e-06,
+      "loss": 2.6943,
+      "step": 2611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.079143919981742e-06,
+      "loss": 2.6469,
+      "step": 2612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.075463093771827e-06,
+      "loss": 2.6192,
+      "step": 2613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.071782226655721e-06,
+      "loss": 2.6045,
+      "step": 2614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.068101320628708e-06,
+      "loss": 2.6476,
+      "step": 2615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.064420377686089e-06,
+      "loss": 2.6541,
+      "step": 2616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0607393998231915e-06,
+      "loss": 2.7291,
+      "step": 2617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0570583890353585e-06,
+      "loss": 2.6682,
+      "step": 2618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.053377347317952e-06,
+      "loss": 2.7015,
+      "step": 2619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0496962766663495e-06,
+      "loss": 2.6009,
+      "step": 2620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.046015179075941e-06,
+      "loss": 2.6367,
+      "step": 2621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.042334056542141e-06,
+      "loss": 2.6655,
+      "step": 2622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.038652911060368e-06,
+      "loss": 2.6123,
+      "step": 2623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.034971744626055e-06,
+      "loss": 2.6229,
+      "step": 2624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.03129055923465e-06,
+      "loss": 2.7157,
+      "step": 2625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0276093568816076e-06,
+      "loss": 2.6586,
+      "step": 2626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.023928139562393e-06,
+      "loss": 2.6053,
+      "step": 2627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.020246909272479e-06,
+      "loss": 2.6287,
+      "step": 2628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.016565668007348e-06,
+      "loss": 2.6305,
+      "step": 2629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.012884417762485e-06,
+      "loss": 2.6296,
+      "step": 2630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.009203160533378e-06,
+      "loss": 2.6361,
+      "step": 2631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.005521898315526e-06,
+      "loss": 2.6595,
+      "step": 2632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.001840633104427e-06,
+      "loss": 2.6568,
+      "step": 2633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.998159366895576e-06,
+      "loss": 2.6456,
+      "step": 2634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.994478101684476e-06,
+      "loss": 2.6475,
+      "step": 2635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9907968394666235e-06,
+      "loss": 2.6431,
+      "step": 2636
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.987115582237517e-06,
+      "loss": 2.6308,
+      "step": 2637
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.983434331992654e-06,
+      "loss": 2.6435,
+      "step": 2638
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9797530907275226e-06,
+      "loss": 2.6304,
+      "step": 2639
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.976071860437609e-06,
+      "loss": 2.7513,
+      "step": 2640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.972390643118395e-06,
+      "loss": 2.6713,
+      "step": 2641
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.968709440765352e-06,
+      "loss": 2.5709,
+      "step": 2642
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.965028255373946e-06,
+      "loss": 2.6788,
+      "step": 2643
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.961347088939636e-06,
+      "loss": 2.6325,
+      "step": 2644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.957665943457861e-06,
+      "loss": 2.6324,
+      "step": 2645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.953984820924059e-06,
+      "loss": 2.668,
+      "step": 2646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.950303723333654e-06,
+      "loss": 2.6199,
+      "step": 2647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.946622652682049e-06,
+      "loss": 2.6896,
+      "step": 2648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9429416109646415e-06,
+      "loss": 2.6048,
+      "step": 2649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.93926060017681e-06,
+      "loss": 2.6696,
+      "step": 2650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.935579622313913e-06,
+      "loss": 2.6426,
+      "step": 2651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9318986793712945e-06,
+      "loss": 2.6982,
+      "step": 2652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9282177733442804e-06,
+      "loss": 2.7121,
+      "step": 2653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.924536906228174e-06,
+      "loss": 2.6679,
+      "step": 2654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.92085608001826e-06,
+      "loss": 2.6969,
+      "step": 2655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.917175296709798e-06,
+      "loss": 2.6429,
+      "step": 2656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913494558298026e-06,
+      "loss": 2.6598,
+      "step": 2657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.909813866778159e-06,
+      "loss": 2.5899,
+      "step": 2658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906133224145384e-06,
+      "loss": 2.6275,
+      "step": 2659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9024526323948615e-06,
+      "loss": 2.6471,
+      "step": 2660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.898772093521729e-06,
+      "loss": 2.6574,
+      "step": 2661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.895091609521089e-06,
+      "loss": 2.6883,
+      "step": 2662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.891411182388017e-06,
+      "loss": 2.6651,
+      "step": 2663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.887730814117561e-06,
+      "loss": 2.7436,
+      "step": 2664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.884050506704731e-06,
+      "loss": 2.6419,
+      "step": 2665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.880370262144508e-06,
+      "loss": 2.651,
+      "step": 2666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.876690082431838e-06,
+      "loss": 2.6608,
+      "step": 2667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.873009969561631e-06,
+      "loss": 2.6285,
+      "step": 2668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.869329925528761e-06,
+      "loss": 2.7146,
+      "step": 2669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.865649952328069e-06,
+      "loss": 2.69,
+      "step": 2670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.861970051954349e-06,
+      "loss": 2.6843,
+      "step": 2671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.858290226402363e-06,
+      "loss": 2.6725,
+      "step": 2672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.854610477666828e-06,
+      "loss": 2.6652,
+      "step": 2673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.850930807742423e-06,
+      "loss": 2.6462,
+      "step": 2674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.847251218623781e-06,
+      "loss": 2.6496,
+      "step": 2675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.843571712305493e-06,
+      "loss": 2.5744,
+      "step": 2676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.839892290782106e-06,
+      "loss": 2.6276,
+      "step": 2677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.836212956048118e-06,
+      "loss": 2.6695,
+      "step": 2678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.832533710097981e-06,
+      "loss": 2.6719,
+      "step": 2679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8288545549261015e-06,
+      "loss": 2.6805,
+      "step": 2680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.825175492526836e-06,
+      "loss": 2.6126,
+      "step": 2681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.821496524894486e-06,
+      "loss": 2.6576,
+      "step": 2682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.817817654023307e-06,
+      "loss": 2.5895,
+      "step": 2683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.814138881907502e-06,
+      "loss": 2.6072,
+      "step": 2684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8104602105412145e-06,
+      "loss": 2.5893,
+      "step": 2685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.806781641918541e-06,
+      "loss": 2.7113,
+      "step": 2686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.803103178033519e-06,
+      "loss": 2.6346,
+      "step": 2687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.799424820880127e-06,
+      "loss": 2.6388,
+      "step": 2688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.795746572452288e-06,
+      "loss": 2.6287,
+      "step": 2689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7920684347438684e-06,
+      "loss": 2.6859,
+      "step": 2690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.788390409748669e-06,
+      "loss": 2.609,
+      "step": 2691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.784712499460436e-06,
+      "loss": 2.6476,
+      "step": 2692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.781034705872846e-06,
+      "loss": 2.5426,
+      "step": 2693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.777357030979518e-06,
+      "loss": 2.6719,
+      "step": 2694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.773679476774007e-06,
+      "loss": 2.6589,
+      "step": 2695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7700020452497966e-06,
+      "loss": 2.6338,
+      "step": 2696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.76632473840031e-06,
+      "loss": 2.6465,
+      "step": 2697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.762647558218902e-06,
+      "loss": 2.6392,
+      "step": 2698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.758970506698854e-06,
+      "loss": 2.642,
+      "step": 2699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.755293585833383e-06,
+      "loss": 2.7156,
+      "step": 2700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.751616797615634e-06,
+      "loss": 2.6581,
+      "step": 2701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.747940144038677e-06,
+      "loss": 2.6454,
+      "step": 2702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.744263627095513e-06,
+      "loss": 2.6659,
+      "step": 2703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.74058724877907e-06,
+      "loss": 2.6447,
+      "step": 2704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7369110110821916e-06,
+      "loss": 2.6568,
+      "step": 2705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.733234915997656e-06,
+      "loss": 2.6994,
+      "step": 2706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.729558965518161e-06,
+      "loss": 2.6792,
+      "step": 2707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.725883161636321e-06,
+      "loss": 2.6697,
+      "step": 2708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.722207506344678e-06,
+      "loss": 2.6328,
+      "step": 2709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7185320016356865e-06,
+      "loss": 2.6109,
+      "step": 2710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7148566495017266e-06,
+      "loss": 2.7365,
+      "step": 2711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.71118145193509e-06,
+      "loss": 2.6449,
+      "step": 2712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.707506410927987e-06,
+      "loss": 2.7096,
+      "step": 2713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.703831528472545e-06,
+      "loss": 2.6897,
+      "step": 2714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.700156806560802e-06,
+      "loss": 2.7108,
+      "step": 2715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.696482247184708e-06,
+      "loss": 2.595,
+      "step": 2716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.692807852336128e-06,
+      "loss": 2.6842,
+      "step": 2717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.689133624006842e-06,
+      "loss": 2.6532,
+      "step": 2718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.685459564188527e-06,
+      "loss": 2.6963,
+      "step": 2719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.681785674872781e-06,
+      "loss": 2.6756,
+      "step": 2720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6781119580511054e-06,
+      "loss": 2.6272,
+      "step": 2721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.674438415714904e-06,
+      "loss": 2.5714,
+      "step": 2722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.670765049855492e-06,
+      "loss": 2.6903,
+      "step": 2723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.667091862464087e-06,
+      "loss": 2.7184,
+      "step": 2724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.663418855531808e-06,
+      "loss": 2.6602,
+      "step": 2725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.659746031049679e-06,
+      "loss": 2.7038,
+      "step": 2726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.656073391008622e-06,
+      "loss": 2.7132,
+      "step": 2727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.652400937399461e-06,
+      "loss": 2.632,
+      "step": 2728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.648728672212922e-06,
+      "loss": 2.542,
+      "step": 2729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.64505659743962e-06,
+      "loss": 2.6601,
+      "step": 2730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.641384715070075e-06,
+      "loss": 2.7607,
+      "step": 2731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.637713027094703e-06,
+      "loss": 2.6137,
+      "step": 2732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.634041535503806e-06,
+      "loss": 2.6352,
+      "step": 2733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.630370242287588e-06,
+      "loss": 2.6447,
+      "step": 2734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.626699149436144e-06,
+      "loss": 2.6718,
+      "step": 2735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.623028258939456e-06,
+      "loss": 2.5931,
+      "step": 2736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.619357572787402e-06,
+      "loss": 2.6813,
+      "step": 2737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.615687092969745e-06,
+      "loss": 2.6646,
+      "step": 2738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6120168214761375e-06,
+      "loss": 2.6439,
+      "step": 2739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.60834676029612e-06,
+      "loss": 2.6731,
+      "step": 2740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.60467691141912e-06,
+      "loss": 2.6288,
+      "step": 2741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.601007276834445e-06,
+      "loss": 2.5862,
+      "step": 2742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.597337858531291e-06,
+      "loss": 2.6987,
+      "step": 2743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.593668658498737e-06,
+      "loss": 2.6573,
+      "step": 2744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.589999678725739e-06,
+      "loss": 2.538,
+      "step": 2745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5863309212011376e-06,
+      "loss": 2.6364,
+      "step": 2746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.582662387913654e-06,
+      "loss": 2.6445,
+      "step": 2747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.578994080851882e-06,
+      "loss": 2.6315,
+      "step": 2748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5753260020043e-06,
+      "loss": 2.682,
+      "step": 2749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.571658153359256e-06,
+      "loss": 2.6893,
+      "step": 2750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.567990536904979e-06,
+      "loss": 2.6365,
+      "step": 2751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.56432315462957e-06,
+      "loss": 2.663,
+      "step": 2752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.560656008521e-06,
+      "loss": 2.6727,
+      "step": 2753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.556989100567116e-06,
+      "loss": 2.6254,
+      "step": 2754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5533224327556365e-06,
+      "loss": 2.7143,
+      "step": 2755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.549656007074144e-06,
+      "loss": 2.6576,
+      "step": 2756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.545989825510096e-06,
+      "loss": 2.6654,
+      "step": 2757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.542323890050817e-06,
+      "loss": 2.6936,
+      "step": 2758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.53865820268349e-06,
+      "loss": 2.6241,
+      "step": 2759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.534992765395176e-06,
+      "loss": 2.6479,
+      "step": 2760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.531327580172794e-06,
+      "loss": 2.7168,
+      "step": 2761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.527662649003124e-06,
+      "loss": 2.7257,
+      "step": 2762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.523997973872811e-06,
+      "loss": 2.6712,
+      "step": 2763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.520333556768364e-06,
+      "loss": 2.6463,
+      "step": 2764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.516669399676146e-06,
+      "loss": 2.6049,
+      "step": 2765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5130055045823854e-06,
+      "loss": 2.7109,
+      "step": 2766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.509341873473163e-06,
+      "loss": 2.6654,
+      "step": 2767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.50567850833442e-06,
+      "loss": 2.6184,
+      "step": 2768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.502015411151955e-06,
+      "loss": 2.6817,
+      "step": 2769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.498352583911413e-06,
+      "loss": 2.6387,
+      "step": 2770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.494690028598303e-06,
+      "loss": 2.6517,
+      "step": 2771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.491027747197981e-06,
+      "loss": 2.5928,
+      "step": 2772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.487365741695657e-06,
+      "loss": 2.639,
+      "step": 2773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.483704014076387e-06,
+      "loss": 2.6485,
+      "step": 2774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.480042566325083e-06,
+      "loss": 2.6277,
+      "step": 2775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4763814004264975e-06,
+      "loss": 2.6591,
+      "step": 2776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.472720518365237e-06,
+      "loss": 2.6442,
+      "step": 2777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.469059922125753e-06,
+      "loss": 2.7325,
+      "step": 2778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.465399613692338e-06,
+      "loss": 2.6149,
+      "step": 2779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.461739595049132e-06,
+      "loss": 2.6063,
+      "step": 2780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.45807986818012e-06,
+      "loss": 2.6832,
+      "step": 2781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.454420435069119e-06,
+      "loss": 2.6907,
+      "step": 2782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4507612976997995e-06,
+      "loss": 2.6816,
+      "step": 2783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.447102458055666e-06,
+      "loss": 2.6151,
+      "step": 2784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.443443918120056e-06,
+      "loss": 2.7189,
+      "step": 2785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.439785679876156e-06,
+      "loss": 2.6381,
+      "step": 2786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.436127745306978e-06,
+      "loss": 2.6872,
+      "step": 2787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.432470116395378e-06,
+      "loss": 2.7392,
+      "step": 2788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4288127951240435e-06,
+      "loss": 2.7333,
+      "step": 2789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.425155783475489e-06,
+      "loss": 2.6514,
+      "step": 2790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.421499083432071e-06,
+      "loss": 2.686,
+      "step": 2791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4178426969759735e-06,
+      "loss": 2.6539,
+      "step": 2792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.414186626089205e-06,
+      "loss": 2.7679,
+      "step": 2793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.410530872753611e-06,
+      "loss": 2.7303,
+      "step": 2794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4068754389508616e-06,
+      "loss": 2.6204,
+      "step": 2795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.403220326662452e-06,
+      "loss": 2.7422,
+      "step": 2796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.399565537869703e-06,
+      "loss": 2.6077,
+      "step": 2797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.395911074553767e-06,
+      "loss": 2.6091,
+      "step": 2798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.39225693869561e-06,
+      "loss": 2.6848,
+      "step": 2799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.388603132276028e-06,
+      "loss": 2.6819,
+      "step": 2800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.384949657275635e-06,
+      "loss": 2.6121,
+      "step": 2801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.381296515674863e-06,
+      "loss": 2.65,
+      "step": 2802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.377643709453972e-06,
+      "loss": 2.7185,
+      "step": 2803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.373991240593028e-06,
+      "loss": 2.6532,
+      "step": 2804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.370339111071924e-06,
+      "loss": 2.6991,
+      "step": 2805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.366687322870367e-06,
+      "loss": 2.7279,
+      "step": 2806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.363035877967874e-06,
+      "loss": 2.6513,
+      "step": 2807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.359384778343781e-06,
+      "loss": 2.6675,
+      "step": 2808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.355734025977236e-06,
+      "loss": 2.6525,
+      "step": 2809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3520836228471984e-06,
+      "loss": 2.6791,
+      "step": 2810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.348433570932435e-06,
+      "loss": 2.6881,
+      "step": 2811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.34478387221153e-06,
+      "loss": 2.6875,
+      "step": 2812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.341134528662867e-06,
+      "loss": 2.5748,
+      "step": 2813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.337485542264643e-06,
+      "loss": 2.6744,
+      "step": 2814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.333836914994862e-06,
+      "loss": 2.6094,
+      "step": 2815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.330188648831328e-06,
+      "loss": 2.6684,
+      "step": 2816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.326540745751653e-06,
+      "loss": 2.6109,
+      "step": 2817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.322893207733253e-06,
+      "loss": 2.6934,
+      "step": 2818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.319246036753343e-06,
+      "loss": 2.6643,
+      "step": 2819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.315599234788941e-06,
+      "loss": 2.6827,
+      "step": 2820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.311952803816867e-06,
+      "loss": 2.6841,
+      "step": 2821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.308306745813734e-06,
+      "loss": 2.6124,
+      "step": 2822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.304661062755958e-06,
+      "loss": 2.6302,
+      "step": 2823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.301015756619751e-06,
+      "loss": 2.6232,
+      "step": 2824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.297370829381119e-06,
+      "loss": 2.6296,
+      "step": 2825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.293726283015866e-06,
+      "loss": 2.6566,
+      "step": 2826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2900821194995816e-06,
+      "loss": 2.5653,
+      "step": 2827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.286438340807656e-06,
+      "loss": 2.6404,
+      "step": 2828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.282794948915271e-06,
+      "loss": 2.6901,
+      "step": 2829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.279151945797392e-06,
+      "loss": 2.6951,
+      "step": 2830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.275509333428778e-06,
+      "loss": 2.668,
+      "step": 2831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.271867113783978e-06,
+      "loss": 2.6277,
+      "step": 2832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.268225288837323e-06,
+      "loss": 2.6172,
+      "step": 2833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.264583860562933e-06,
+      "loss": 2.6269,
+      "step": 2834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.260942830934716e-06,
+      "loss": 2.6494,
+      "step": 2835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.257302201926356e-06,
+      "loss": 2.7063,
+      "step": 2836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.253661975511327e-06,
+      "loss": 2.6011,
+      "step": 2837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.250022153662885e-06,
+      "loss": 2.6948,
+      "step": 2838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2463827383540585e-06,
+      "loss": 2.6208,
+      "step": 2839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.242743731557664e-06,
+      "loss": 2.6425,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2391051352462955e-06,
+      "loss": 2.617,
+      "step": 2841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.235466951392319e-06,
+      "loss": 2.6801,
+      "step": 2842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.231829181967885e-06,
+      "loss": 2.6224,
+      "step": 2843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2281918289449085e-06,
+      "loss": 2.7156,
+      "step": 2844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.22455489429509e-06,
+      "loss": 2.6448,
+      "step": 2845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.220918379989898e-06,
+      "loss": 2.6914,
+      "step": 2846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.217282288000571e-06,
+      "loss": 2.7017,
+      "step": 2847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.213646620298126e-06,
+      "loss": 2.6849,
+      "step": 2848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.210011378853341e-06,
+      "loss": 2.6176,
+      "step": 2849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.206376565636767e-06,
+      "loss": 2.6665,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.202742182618726e-06,
+      "loss": 2.652,
+      "step": 2851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1991082317693025e-06,
+      "loss": 2.7521,
+      "step": 2852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.195474715058346e-06,
+      "loss": 2.686,
+      "step": 2853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1918416344554745e-06,
+      "loss": 2.7268,
+      "step": 2854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1882089919300685e-06,
+      "loss": 2.6579,
+      "step": 2855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.184576789451269e-06,
+      "loss": 2.6946,
+      "step": 2856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.180945028987979e-06,
+      "loss": 2.5797,
+      "step": 2857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.177313712508866e-06,
+      "loss": 2.6554,
+      "step": 2858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.173682841982349e-06,
+      "loss": 2.5895,
+      "step": 2859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.170052419376613e-06,
+      "loss": 2.6936,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.166422446659596e-06,
+      "loss": 2.6218,
+      "step": 2861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.162792925798992e-06,
+      "loss": 2.6657,
+      "step": 2862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.159163858762255e-06,
+      "loss": 2.6374,
+      "step": 2863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.155535247516583e-06,
+      "loss": 2.6809,
+      "step": 2864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.151907094028938e-06,
+      "loss": 2.5906,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.148279400266029e-06,
+      "loss": 2.6567,
+      "step": 2866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.144652168194312e-06,
+      "loss": 2.6529,
+      "step": 2867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.141025399779997e-06,
+      "loss": 2.6523,
+      "step": 2868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.137399096989048e-06,
+      "loss": 2.5649,
+      "step": 2869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.133773261787164e-06,
+      "loss": 2.6384,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.130147896139799e-06,
+      "loss": 2.6136,
+      "step": 2871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.126523002012153e-06,
+      "loss": 2.6306,
+      "step": 2872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1228985813691655e-06,
+      "loss": 2.6449,
+      "step": 2873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.119274636175523e-06,
+      "loss": 2.6157,
+      "step": 2874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.115651168395654e-06,
+      "loss": 2.6717,
+      "step": 2875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.112028179993727e-06,
+      "loss": 2.6079,
+      "step": 2876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1084056729336494e-06,
+      "loss": 2.6685,
+      "step": 2877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.104783649179074e-06,
+      "loss": 2.667,
+      "step": 2878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.101162110693381e-06,
+      "loss": 2.6576,
+      "step": 2879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.097541059439698e-06,
+      "loss": 2.6894,
+      "step": 2880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.093920497380882e-06,
+      "loss": 2.6466,
+      "step": 2881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.090300426479526e-06,
+      "loss": 2.5433,
+      "step": 2882
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.08668084869796e-06,
+      "loss": 2.6327,
+      "step": 2883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.083061765998243e-06,
+      "loss": 2.6971,
+      "step": 2884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.079443180342167e-06,
+      "loss": 2.6611,
+      "step": 2885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.075825093691256e-06,
+      "loss": 2.6221,
+      "step": 2886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.07220750800676e-06,
+      "loss": 2.6151,
+      "step": 2887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.068590425249659e-06,
+      "loss": 2.6131,
+      "step": 2888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.064973847380665e-06,
+      "loss": 2.665,
+      "step": 2889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.061357776360207e-06,
+      "loss": 2.6432,
+      "step": 2890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.057742214148446e-06,
+      "loss": 2.6678,
+      "step": 2891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.054127162705268e-06,
+      "loss": 2.6281,
+      "step": 2892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.050512623990277e-06,
+      "loss": 2.6213,
+      "step": 2893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.046898599962802e-06,
+      "loss": 2.664,
+      "step": 2894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.043285092581895e-06,
+      "loss": 2.717,
+      "step": 2895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.039672103806324e-06,
+      "loss": 2.672,
+      "step": 2896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.036059635594578e-06,
+      "loss": 2.6475,
+      "step": 2897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.032447689904862e-06,
+      "loss": 2.6989,
+      "step": 2898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.028836268695101e-06,
+      "loss": 2.7171,
+      "step": 2899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.025225373922935e-06,
+      "loss": 2.576,
+      "step": 2900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.021615007545714e-06,
+      "loss": 2.6322,
+      "step": 2901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.018005171520507e-06,
+      "loss": 2.6902,
+      "step": 2902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.014395867804096e-06,
+      "loss": 2.6051,
+      "step": 2903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.010787098352967e-06,
+      "loss": 2.6447,
+      "step": 2904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.007178865123325e-06,
+      "loss": 2.5791,
+      "step": 2905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.003571170071082e-06,
+      "loss": 2.7057,
+      "step": 2906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.999964015151852e-06,
+      "loss": 2.6201,
+      "step": 2907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.996357402320965e-06,
+      "loss": 2.6172,
+      "step": 2908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992751333533453e-06,
+      "loss": 2.6178,
+      "step": 2909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989145810744054e-06,
+      "loss": 2.6533,
+      "step": 2910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985540835907206e-06,
+      "loss": 2.6538,
+      "step": 2911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981936410977061e-06,
+      "loss": 2.669,
+      "step": 2912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.978332537907457e-06,
+      "loss": 2.6458,
+      "step": 2913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.974729218651946e-06,
+      "loss": 2.662,
+      "step": 2914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9711264551637754e-06,
+      "loss": 2.5632,
+      "step": 2915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.967524249395888e-06,
+      "loss": 2.6414,
+      "step": 2916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.963922603300931e-06,
+      "loss": 2.6295,
+      "step": 2917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9603215188312395e-06,
+      "loss": 2.6381,
+      "step": 2918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9567209979388534e-06,
+      "loss": 2.6013,
+      "step": 2919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9531210425755015e-06,
+      "loss": 2.5983,
+      "step": 2920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.949521654692607e-06,
+      "loss": 2.6168,
+      "step": 2921
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.945922836241287e-06,
+      "loss": 2.6182,
+      "step": 2922
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.94232458917235e-06,
+      "loss": 2.619,
+      "step": 2923
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9387269154362885e-06,
+      "loss": 2.674,
+      "step": 2924
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.935129816983294e-06,
+      "loss": 2.628,
+      "step": 2925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.931533295763242e-06,
+      "loss": 2.6354,
+      "step": 2926
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.92793735372569e-06,
+      "loss": 2.6959,
+      "step": 2927
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.924341992819889e-06,
+      "loss": 2.6668,
+      "step": 2928
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.920747214994774e-06,
+      "loss": 2.621,
+      "step": 2929
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.917153022198958e-06,
+      "loss": 2.6059,
+      "step": 2930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.913559416380743e-06,
+      "loss": 2.6156,
+      "step": 2931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.909966399488112e-06,
+      "loss": 2.6575,
+      "step": 2932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.906373973468724e-06,
+      "loss": 2.6449,
+      "step": 2933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.902782140269925e-06,
+      "loss": 2.6265,
+      "step": 2934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8991909018387335e-06,
+      "loss": 2.6542,
+      "step": 2935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8956002601218475e-06,
+      "loss": 2.6499,
+      "step": 2936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.892010217065646e-06,
+      "loss": 2.7068,
+      "step": 2937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.888420774616174e-06,
+      "loss": 2.6732,
+      "step": 2938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8848319347191586e-06,
+      "loss": 2.5922,
+      "step": 2939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.88124369932e-06,
+      "loss": 2.6991,
+      "step": 2940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.877656070363765e-06,
+      "loss": 2.6329,
+      "step": 2941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8740690497951985e-06,
+      "loss": 2.5863,
+      "step": 2942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.870482639558712e-06,
+      "loss": 2.6688,
+      "step": 2943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8668968415983865e-06,
+      "loss": 2.6206,
+      "step": 2944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.86331165785797e-06,
+      "loss": 2.7185,
+      "step": 2945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8597270902808815e-06,
+      "loss": 2.6477,
+      "step": 2946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8561431408102e-06,
+      "loss": 2.7119,
+      "step": 2947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.852559811388676e-06,
+      "loss": 2.6264,
+      "step": 2948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8489771039587206e-06,
+      "loss": 2.7197,
+      "step": 2949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.845395020462404e-06,
+      "loss": 2.7418,
+      "step": 2950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.841813562841466e-06,
+      "loss": 2.5998,
+      "step": 2951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.838232733037303e-06,
+      "loss": 2.5869,
+      "step": 2952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.834652532990969e-06,
+      "loss": 2.6349,
+      "step": 2953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.831072964643182e-06,
+      "loss": 2.701,
+      "step": 2954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.82749402993431e-06,
+      "loss": 2.6664,
+      "step": 2955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.823915730804384e-06,
+      "loss": 2.6516,
+      "step": 2956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.82033806919309e-06,
+      "loss": 2.6611,
+      "step": 2957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.816761047039765e-06,
+      "loss": 2.6717,
+      "step": 2958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8131846662834022e-06,
+      "loss": 2.6486,
+      "step": 2959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.809608928862648e-06,
+      "loss": 2.6584,
+      "step": 2960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8060338367157932e-06,
+      "loss": 2.6436,
+      "step": 2961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.802459391780788e-06,
+      "loss": 2.5881,
+      "step": 2962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7988855959952274e-06,
+      "loss": 2.6525,
+      "step": 2963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7953124512963525e-06,
+      "loss": 2.5943,
+      "step": 2964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.791739959621054e-06,
+      "loss": 2.7263,
+      "step": 2965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.788168122905872e-06,
+      "loss": 2.7343,
+      "step": 2966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7845969430869824e-06,
+      "loss": 2.6264,
+      "step": 2967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7810264221002135e-06,
+      "loss": 2.6328,
+      "step": 2968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7774565618810342e-06,
+      "loss": 2.7423,
+      "step": 2969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7738873643645523e-06,
+      "loss": 2.6964,
+      "step": 2970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7703188314855205e-06,
+      "loss": 2.7102,
+      "step": 2971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7667509651783284e-06,
+      "loss": 2.663,
+      "step": 2972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7631837673770045e-06,
+      "loss": 2.6206,
+      "step": 2973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7596172400152176e-06,
+      "loss": 2.6462,
+      "step": 2974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7560513850262683e-06,
+      "loss": 2.6434,
+      "step": 2975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7524862043430958e-06,
+      "loss": 2.6681,
+      "step": 2976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.748921699898277e-06,
+      "loss": 2.6275,
+      "step": 2977
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7453578736240144e-06,
+      "loss": 2.6895,
+      "step": 2978
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7417947274521482e-06,
+      "loss": 2.5784,
+      "step": 2979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.738232263314151e-06,
+      "loss": 2.6758,
+      "step": 2980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7346704831411214e-06,
+      "loss": 2.5907,
+      "step": 2981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7311093888637906e-06,
+      "loss": 2.5856,
+      "step": 2982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.727548982412517e-06,
+      "loss": 2.5965,
+      "step": 2983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7239892657172838e-06,
+      "loss": 2.6087,
+      "step": 2984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.720430240707704e-06,
+      "loss": 2.6108,
+      "step": 2985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.716871909313017e-06,
+      "loss": 2.6258,
+      "step": 2986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.713314273462078e-06,
+      "loss": 2.6892,
+      "step": 2987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7097573350833734e-06,
+      "loss": 2.6252,
+      "step": 2988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.70620109610501e-06,
+      "loss": 2.6542,
+      "step": 2989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7026455584547104e-06,
+      "loss": 2.6905,
+      "step": 2990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6990907240598243e-06,
+      "loss": 2.6939,
+      "step": 2991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6955365948473154e-06,
+      "loss": 2.6377,
+      "step": 2992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6919831727437643e-06,
+      "loss": 2.711,
+      "step": 2993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.688430459675374e-06,
+      "loss": 2.5972,
+      "step": 2994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6848784575679565e-06,
+      "loss": 2.6626,
+      "step": 2995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.681327168346942e-06,
+      "loss": 2.6117,
+      "step": 2996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6777765939373778e-06,
+      "loss": 2.5554,
+      "step": 2997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6742267362639133e-06,
+      "loss": 2.6821,
+      "step": 2998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.670677597250819e-06,
+      "loss": 2.6185,
+      "step": 2999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.667129178821974e-06,
+      "loss": 2.6461,
+      "step": 3000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6635814829008622e-06,
+      "loss": 2.5958,
+      "step": 3001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.66003451141058e-06,
+      "loss": 2.7365,
+      "step": 3002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6564882662738327e-06,
+      "loss": 2.7088,
+      "step": 3003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.652942749412925e-06,
+      "loss": 2.6585,
+      "step": 3004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6493979627497726e-06,
+      "loss": 2.7167,
+      "step": 3005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6458539082058965e-06,
+      "loss": 2.707,
+      "step": 3006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6423105877024146e-06,
+      "loss": 2.6567,
+      "step": 3007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6387680031600514e-06,
+      "loss": 2.6402,
+      "step": 3008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6352261564991353e-06,
+      "loss": 2.6495,
+      "step": 3009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6316850496395863e-06,
+      "loss": 2.6496,
+      "step": 3010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6281446845009306e-06,
+      "loss": 2.663,
+      "step": 3011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6246050630022877e-06,
+      "loss": 2.6547,
+      "step": 3012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.621066187062377e-06,
+      "loss": 2.6502,
+      "step": 3013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6175280585995155e-06,
+      "loss": 2.6378,
+      "step": 3014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6139906795316083e-06,
+      "loss": 2.6579,
+      "step": 3015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6104540517761594e-06,
+      "loss": 2.719,
+      "step": 3016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6069181772502664e-06,
+      "loss": 2.6423,
+      "step": 3017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.603383057870615e-06,
+      "loss": 2.6259,
+      "step": 3018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5998486955534843e-06,
+      "loss": 2.6473,
+      "step": 3019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.596315092214742e-06,
+      "loss": 2.5485,
+      "step": 3020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.592782249769843e-06,
+      "loss": 2.6435,
+      "step": 3021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.589250170133831e-06,
+      "loss": 2.6423,
+      "step": 3022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5857188552213396e-06,
+      "loss": 2.5983,
+      "step": 3023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.58218830694658e-06,
+      "loss": 2.6115,
+      "step": 3024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5786585272233543e-06,
+      "loss": 2.6675,
+      "step": 3025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5751295179650493e-06,
+      "loss": 2.5747,
+      "step": 3026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.571601281084625e-06,
+      "loss": 2.7295,
+      "step": 3027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5680738184946324e-06,
+      "loss": 2.6217,
+      "step": 3028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.564547132107199e-06,
+      "loss": 2.6671,
+      "step": 3029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5610212238340305e-06,
+      "loss": 2.6717,
+      "step": 3030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5574960955864125e-06,
+      "loss": 2.5909,
+      "step": 3031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5539717492752058e-06,
+      "loss": 2.65,
+      "step": 3032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.55044818681085e-06,
+      "loss": 2.6487,
+      "step": 3033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.54692541010336e-06,
+      "loss": 2.6478,
+      "step": 3034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5434034210623203e-06,
+      "loss": 2.6256,
+      "step": 3035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.539882221596892e-06,
+      "loss": 2.5995,
+      "step": 3036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5363618136158107e-06,
+      "loss": 2.6576,
+      "step": 3037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5328421990273754e-06,
+      "loss": 2.6563,
+      "step": 3038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.529323379739461e-06,
+      "loss": 2.6623,
+      "step": 3039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.525805357659513e-06,
+      "loss": 2.6689,
+      "step": 3040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5222881346945366e-06,
+      "loss": 2.674,
+      "step": 3041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.518771712751111e-06,
+      "loss": 2.6774,
+      "step": 3042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5152560937353795e-06,
+      "loss": 2.6502,
+      "step": 3043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5117412795530487e-06,
+      "loss": 2.6734,
+      "step": 3044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5082272721093896e-06,
+      "loss": 2.7093,
+      "step": 3045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.504714073309239e-06,
+      "loss": 2.6537,
+      "step": 3046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5012016850569885e-06,
+      "loss": 2.6707,
+      "step": 3047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.497690109256597e-06,
+      "loss": 2.6849,
+      "step": 3048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4941793478115788e-06,
+      "loss": 2.526,
+      "step": 3049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4906694026250075e-06,
+      "loss": 2.7049,
+      "step": 3050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.487160275599519e-06,
+      "loss": 2.6761,
+      "step": 3051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.483651968637297e-06,
+      "loss": 2.6752,
+      "step": 3052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.480144483640087e-06,
+      "loss": 2.6686,
+      "step": 3053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.476637822509189e-06,
+      "loss": 2.6384,
+      "step": 3054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.473131987145452e-06,
+      "loss": 2.6115,
+      "step": 3055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.469626979449282e-06,
+      "loss": 2.6746,
+      "step": 3056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4661228013206356e-06,
+      "loss": 2.7466,
+      "step": 3057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4626194546590147e-06,
+      "loss": 2.611,
+      "step": 3058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4591169413634766e-06,
+      "loss": 2.6562,
+      "step": 3059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.455615263332627e-06,
+      "loss": 2.6048,
+      "step": 3060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4521144224646115e-06,
+      "loss": 2.6268,
+      "step": 3061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.448614420657129e-06,
+      "loss": 2.6602,
+      "step": 3062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4451152598074242e-06,
+      "loss": 2.6399,
+      "step": 3063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.441616941812279e-06,
+      "loss": 2.7166,
+      "step": 3064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4381194685680237e-06,
+      "loss": 2.6095,
+      "step": 3065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4346228419705306e-06,
+      "loss": 2.609,
+      "step": 3066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.431127063915213e-06,
+      "loss": 2.6888,
+      "step": 3067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.42763213629702e-06,
+      "loss": 2.6157,
+      "step": 3068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4241380610104446e-06,
+      "loss": 2.6061,
+      "step": 3069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4206448399495164e-06,
+      "loss": 2.6116,
+      "step": 3070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4171524750078026e-06,
+      "loss": 2.6437,
+      "step": 3071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.413660968078403e-06,
+      "loss": 2.6768,
+      "step": 3072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4101703210539554e-06,
+      "loss": 2.7413,
+      "step": 3073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.406680535826634e-06,
+      "loss": 2.686,
+      "step": 3074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.403191614288137e-06,
+      "loss": 2.6404,
+      "step": 3075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3997035583297044e-06,
+      "loss": 2.6581,
+      "step": 3076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3962163698421013e-06,
+      "loss": 2.5855,
+      "step": 3077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3927300507156235e-06,
+      "loss": 2.6772,
+      "step": 3078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.389244602840096e-06,
+      "loss": 2.6938,
+      "step": 3079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3857600281048725e-06,
+      "loss": 2.5572,
+      "step": 3080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3822763283988304e-06,
+      "loss": 2.7293,
+      "step": 3081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.378793505610376e-06,
+      "loss": 2.7345,
+      "step": 3082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3753115616274414e-06,
+      "loss": 2.6772,
+      "step": 3083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.371830498337475e-06,
+      "loss": 2.6541,
+      "step": 3084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3683503176274573e-06,
+      "loss": 2.6551,
+      "step": 3085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.364871021383882e-06,
+      "loss": 2.6293,
+      "step": 3086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3613926114927687e-06,
+      "loss": 2.5975,
+      "step": 3087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3579150898396575e-06,
+      "loss": 2.7173,
+      "step": 3088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.354438458309599e-06,
+      "loss": 2.6713,
+      "step": 3089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.350962718787171e-06,
+      "loss": 2.654,
+      "step": 3090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3474878731564636e-06,
+      "loss": 2.6575,
+      "step": 3091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.34401392330108e-06,
+      "loss": 2.6649,
+      "step": 3092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3405408711041416e-06,
+      "loss": 2.6611,
+      "step": 3093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3370687184482837e-06,
+      "loss": 2.6129,
+      "step": 3094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3335974672156478e-06,
+      "loss": 2.6362,
+      "step": 3095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3301271192878933e-06,
+      "loss": 2.6746,
+      "step": 3096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3266576765461893e-06,
+      "loss": 2.6236,
+      "step": 3097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.323189140871209e-06,
+      "loss": 2.6446,
+      "step": 3098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3197215141431393e-06,
+      "loss": 2.6818,
+      "step": 3099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3162547982416746e-06,
+      "loss": 2.694,
+      "step": 3100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3127889950460094e-06,
+      "loss": 2.6314,
+      "step": 3101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3093241064348504e-06,
+      "loss": 2.6314,
+      "step": 3102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3058601342864055e-06,
+      "loss": 2.7166,
+      "step": 3103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3023970804783846e-06,
+      "loss": 2.6209,
+      "step": 3104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.298934946888005e-06,
+      "loss": 2.667,
+      "step": 3105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2954737353919765e-06,
+      "loss": 2.6198,
+      "step": 3106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2920134478665157e-06,
+      "loss": 2.6396,
+      "step": 3107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2885540861873394e-06,
+      "loss": 2.6841,
+      "step": 3108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.285095652229655e-06,
+      "loss": 2.6016,
+      "step": 3109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2816381478681754e-06,
+      "loss": 2.7135,
+      "step": 3110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2781815749771064e-06,
+      "loss": 2.654,
+      "step": 3111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.274725935430144e-06,
+      "loss": 2.7401,
+      "step": 3112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2712712311004868e-06,
+      "loss": 2.6259,
+      "step": 3113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2678174638608216e-06,
+      "loss": 2.6957,
+      "step": 3114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.264364635583328e-06,
+      "loss": 2.6822,
+      "step": 3115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2609127481396745e-06,
+      "loss": 2.6515,
+      "step": 3116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2574618034010245e-06,
+      "loss": 2.6432,
+      "step": 3117
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.254011803238026e-06,
+      "loss": 2.6261,
+      "step": 3118
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2505627495208157e-06,
+      "loss": 2.6521,
+      "step": 3119
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2471146441190225e-06,
+      "loss": 2.64,
+      "step": 3120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2436674889017515e-06,
+      "loss": 2.6527,
+      "step": 3121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2402212857376013e-06,
+      "loss": 2.779,
+      "step": 3122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2367760364946515e-06,
+      "loss": 2.6442,
+      "step": 3123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2333317430404616e-06,
+      "loss": 2.7384,
+      "step": 3124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.229888407242078e-06,
+      "loss": 2.6448,
+      "step": 3125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.226446030966025e-06,
+      "loss": 2.6685,
+      "step": 3126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2230046160783075e-06,
+      "loss": 2.7439,
+      "step": 3127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.21956416444441e-06,
+      "loss": 2.6095,
+      "step": 3128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.216124677929292e-06,
+      "loss": 2.615,
+      "step": 3129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2126861583973933e-06,
+      "loss": 2.6194,
+      "step": 3130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.209248607712631e-06,
+      "loss": 2.6774,
+      "step": 3131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.205812027738389e-06,
+      "loss": 2.7047,
+      "step": 3132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.202376420337533e-06,
+      "loss": 2.6342,
+      "step": 3133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1989417873724006e-06,
+      "loss": 2.628,
+      "step": 3134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.195508130704795e-06,
+      "loss": 2.6562,
+      "step": 3135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1920754521959964e-06,
+      "loss": 2.7168,
+      "step": 3136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.188643753706755e-06,
+      "loss": 2.6174,
+      "step": 3137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.185213037097284e-06,
+      "loss": 2.8094,
+      "step": 3138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1817833042272693e-06,
+      "loss": 2.6018,
+      "step": 3139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.178354556955864e-06,
+      "loss": 2.6915,
+      "step": 3140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.174926797141682e-06,
+      "loss": 2.6288,
+      "step": 3141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1715000266428086e-06,
+      "loss": 2.6472,
+      "step": 3142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1680742473167857e-06,
+      "loss": 2.7575,
+      "step": 3143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1646494610206235e-06,
+      "loss": 2.6665,
+      "step": 3144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1612256696107935e-06,
+      "loss": 2.6755,
+      "step": 3145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.157802874943222e-06,
+      "loss": 2.6504,
+      "step": 3146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.154381078873302e-06,
+      "loss": 2.6824,
+      "step": 3147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.150960283255885e-06,
+      "loss": 2.6909,
+      "step": 3148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.147540489945272e-06,
+      "loss": 2.6337,
+      "step": 3149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1441217007952306e-06,
+      "loss": 2.5962,
+      "step": 3150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.140703917658979e-06,
+      "loss": 2.6722,
+      "step": 3151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1372871423891894e-06,
+      "loss": 2.6387,
+      "step": 3152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1338713768379914e-06,
+      "loss": 2.6061,
+      "step": 3153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.130456622856964e-06,
+      "loss": 2.634,
+      "step": 3154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1270428822971377e-06,
+      "loss": 2.6421,
+      "step": 3155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.123630157008997e-06,
+      "loss": 2.6871,
+      "step": 3156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1202184488424736e-06,
+      "loss": 2.6308,
+      "step": 3157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.116807759646947e-06,
+      "loss": 2.6123,
+      "step": 3158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1133980912712454e-06,
+      "loss": 2.7216,
+      "step": 3159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.109989445563647e-06,
+      "loss": 2.6297,
+      "step": 3160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.106581824371867e-06,
+      "loss": 2.6199,
+      "step": 3161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.103175229543075e-06,
+      "loss": 2.6822,
+      "step": 3162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.099769662923879e-06,
+      "loss": 2.6134,
+      "step": 3163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0963651263603283e-06,
+      "loss": 2.5942,
+      "step": 3164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.092961621697919e-06,
+      "loss": 2.5549,
+      "step": 3165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0895591507815826e-06,
+      "loss": 2.671,
+      "step": 3166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.086157715455694e-06,
+      "loss": 2.644,
+      "step": 3167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0827573175640664e-06,
+      "loss": 2.6921,
+      "step": 3168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.079357958949946e-06,
+      "loss": 2.6248,
+      "step": 3169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0759596414560217e-06,
+      "loss": 2.5752,
+      "step": 3170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0725623669244163e-06,
+      "loss": 2.5986,
+      "step": 3171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.069166137196684e-06,
+      "loss": 2.6883,
+      "step": 3172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.065770954113816e-06,
+      "loss": 2.6521,
+      "step": 3173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0623768195162373e-06,
+      "loss": 2.6025,
+      "step": 3174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0589837352437984e-06,
+      "loss": 2.6119,
+      "step": 3175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0555917031357875e-06,
+      "loss": 2.5326,
+      "step": 3176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0522007250309194e-06,
+      "loss": 2.6804,
+      "step": 3177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0488108027673356e-06,
+      "loss": 2.5973,
+      "step": 3178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0454219381826117e-06,
+      "loss": 2.6258,
+      "step": 3179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0420341331137403e-06,
+      "loss": 2.6671,
+      "step": 3180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.038647389397147e-06,
+      "loss": 2.6192,
+      "step": 3181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.035261708868683e-06,
+      "loss": 2.6357,
+      "step": 3182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.031877093363615e-06,
+      "loss": 2.6022,
+      "step": 3183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0284935447166405e-06,
+      "loss": 2.592,
+      "step": 3184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0251110647618777e-06,
+      "loss": 2.6195,
+      "step": 3185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.021729655332858e-06,
+      "loss": 2.6267,
+      "step": 3186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0183493182625425e-06,
+      "loss": 2.7265,
+      "step": 3187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.014970055383306e-06,
+      "loss": 2.6186,
+      "step": 3188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0115918685269407e-06,
+      "loss": 2.7246,
+      "step": 3189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.008214759524657e-06,
+      "loss": 2.6508,
+      "step": 3190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.00483873020708e-06,
+      "loss": 2.6751,
+      "step": 3191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0014637824042504e-06,
+      "loss": 2.7048,
+      "step": 3192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9980899179456207e-06,
+      "loss": 2.6583,
+      "step": 3193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9947171386600614e-06,
+      "loss": 2.6959,
+      "step": 3194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9913454463758475e-06,
+      "loss": 2.625,
+      "step": 3195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9879748429206677e-06,
+      "loss": 2.6527,
+      "step": 3196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9846053301216266e-06,
+      "loss": 2.6026,
+      "step": 3197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.981236909805225e-06,
+      "loss": 2.6655,
+      "step": 3198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9778695837973835e-06,
+      "loss": 2.6412,
+      "step": 3199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9745033539234225e-06,
+      "loss": 2.6703,
+      "step": 3200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9711382220080707e-06,
+      "loss": 2.6641,
+      "step": 3201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.967774189875462e-06,
+      "loss": 2.5928,
+      "step": 3202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9644112593491315e-06,
+      "loss": 2.6651,
+      "step": 3203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9610494322520205e-06,
+      "loss": 2.6588,
+      "step": 3204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.957688710406472e-06,
+      "loss": 2.6506,
+      "step": 3205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9543290956342257e-06,
+      "loss": 2.6105,
+      "step": 3206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.950970589756425e-06,
+      "loss": 2.6382,
+      "step": 3207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9476131945936136e-06,
+      "loss": 2.6222,
+      "step": 3208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9442569119657272e-06,
+      "loss": 2.668,
+      "step": 3209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.940901743692103e-06,
+      "loss": 2.6281,
+      "step": 3210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.937547691591477e-06,
+      "loss": 2.6558,
+      "step": 3211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.93419475748197e-06,
+      "loss": 2.6126,
+      "step": 3212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.930842943181107e-06,
+      "loss": 2.6823,
+      "step": 3213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9274922505058023e-06,
+      "loss": 2.6244,
+      "step": 3214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9241426812723594e-06,
+      "loss": 2.6772,
+      "step": 3215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.920794237296477e-06,
+      "loss": 2.6715,
+      "step": 3216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.917446920393246e-06,
+      "loss": 2.6698,
+      "step": 3217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9141007323771354e-06,
+      "loss": 2.6468,
+      "step": 3218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9107556750620125e-06,
+      "loss": 2.6555,
+      "step": 3219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.90741175026113e-06,
+      "loss": 2.6863,
+      "step": 3220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.904068959787123e-06,
+      "loss": 2.6622,
+      "step": 3221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.900727305452018e-06,
+      "loss": 2.6133,
+      "step": 3222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.897386789067217e-06,
+      "loss": 2.6701,
+      "step": 3223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8940474124435113e-06,
+      "loss": 2.5804,
+      "step": 3224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8907091773910767e-06,
+      "loss": 2.6238,
+      "step": 3225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.88737208571946e-06,
+      "loss": 2.661,
+      "step": 3226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.884036139237598e-06,
+      "loss": 2.7094,
+      "step": 3227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8807013397538052e-06,
+      "loss": 2.6378,
+      "step": 3228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8773676890757686e-06,
+      "loss": 2.6439,
+      "step": 3229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.874035189010558e-06,
+      "loss": 2.6084,
+      "step": 3230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8707038413646195e-06,
+      "loss": 2.5824,
+      "step": 3231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8673736479437693e-06,
+      "loss": 2.6603,
+      "step": 3232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.864044610553204e-06,
+      "loss": 2.7048,
+      "step": 3233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8607167309974913e-06,
+      "loss": 2.6921,
+      "step": 3234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8573900110805687e-06,
+      "loss": 2.6561,
+      "step": 3235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8540644526057515e-06,
+      "loss": 2.6524,
+      "step": 3236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.850740057375716e-06,
+      "loss": 2.6112,
+      "step": 3237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8474168271925164e-06,
+      "loss": 2.6567,
+      "step": 3238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8440947638575728e-06,
+      "loss": 2.6931,
+      "step": 3239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.84077386917167e-06,
+      "loss": 2.6583,
+      "step": 3240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8374541449349618e-06,
+      "loss": 2.6866,
+      "step": 3241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8341355929469685e-06,
+      "loss": 2.6575,
+      "step": 3242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.830818215006573e-06,
+      "loss": 2.5843,
+      "step": 3243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8275020129120255e-06,
+      "loss": 2.6367,
+      "step": 3244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8241869884609306e-06,
+      "loss": 2.684,
+      "step": 3245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8208731434502624e-06,
+      "loss": 2.6369,
+      "step": 3246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8175604796763535e-06,
+      "loss": 2.6747,
+      "step": 3247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8142489989348963e-06,
+      "loss": 2.6182,
+      "step": 3248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8109387030209378e-06,
+      "loss": 2.6057,
+      "step": 3249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.807629593728889e-06,
+      "loss": 2.703,
+      "step": 3250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.804321672852517e-06,
+      "loss": 2.636,
+      "step": 3251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.801014942184937e-06,
+      "loss": 2.7136,
+      "step": 3252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.797709403518628e-06,
+      "loss": 2.6241,
+      "step": 3253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7944050586454215e-06,
+      "loss": 2.6385,
+      "step": 3254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.791101909356495e-06,
+      "loss": 2.6506,
+      "step": 3255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7877999574423888e-06,
+      "loss": 2.6524,
+      "step": 3256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.784499204692981e-06,
+      "loss": 2.6686,
+      "step": 3257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7811996528975116e-06,
+      "loss": 2.6946,
+      "step": 3258
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.777901303844565e-06,
+      "loss": 2.6535,
+      "step": 3259
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7746041593220708e-06,
+      "loss": 2.6229,
+      "step": 3260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.771308221117309e-06,
+      "loss": 2.6513,
+      "step": 3261
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7680134910169076e-06,
+      "loss": 2.6046,
+      "step": 3262
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.764719970806834e-06,
+      "loss": 2.6608,
+      "step": 3263
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7614276622724034e-06,
+      "loss": 2.7327,
+      "step": 3264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.758136567198276e-06,
+      "loss": 2.6414,
+      "step": 3265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7548466873684487e-06,
+      "loss": 2.6444,
+      "step": 3266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7515580245662642e-06,
+      "loss": 2.622,
+      "step": 3267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7482705805744036e-06,
+      "loss": 2.6529,
+      "step": 3268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.744984357174889e-06,
+      "loss": 2.6172,
+      "step": 3269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7416993561490792e-06,
+      "loss": 2.6538,
+      "step": 3270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7384155792776724e-06,
+      "loss": 2.7075,
+      "step": 3271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.735133028340698e-06,
+      "loss": 2.7104,
+      "step": 3272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7318517051175285e-06,
+      "loss": 2.6058,
+      "step": 3273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7285716113868637e-06,
+      "loss": 2.6399,
+      "step": 3274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7252927489267412e-06,
+      "loss": 2.6675,
+      "step": 3275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.722015119514533e-06,
+      "loss": 2.5875,
+      "step": 3276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7187387249269346e-06,
+      "loss": 2.5922,
+      "step": 3277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7154635669399815e-06,
+      "loss": 2.6909,
+      "step": 3278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7121896473290345e-06,
+      "loss": 2.6686,
+      "step": 3279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.708916967868781e-06,
+      "loss": 2.6689,
+      "step": 3280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7056455303332395e-06,
+      "loss": 2.7007,
+      "step": 3281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7023753364957574e-06,
+      "loss": 2.6925,
+      "step": 3282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6991063881289993e-06,
+      "loss": 2.6068,
+      "step": 3283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6958386870049635e-06,
+      "loss": 2.6524,
+      "step": 3284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6925722348949713e-06,
+      "loss": 2.7001,
+      "step": 3285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.689307033569659e-06,
+      "loss": 2.6027,
+      "step": 3286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6860430847989937e-06,
+      "loss": 2.7431,
+      "step": 3287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.682780390352262e-06,
+      "loss": 2.6573,
+      "step": 3288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.679518951998066e-06,
+      "loss": 2.6407,
+      "step": 3289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.676258771504329e-06,
+      "loss": 2.691,
+      "step": 3290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.672999850638296e-06,
+      "loss": 2.6638,
+      "step": 3291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6697421911665266e-06,
+      "loss": 2.6658,
+      "step": 3292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.666485794854894e-06,
+      "loss": 2.6818,
+      "step": 3293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.66323066346859e-06,
+      "loss": 2.6306,
+      "step": 3294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6599767987721204e-06,
+      "loss": 2.6003,
+      "step": 3295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.656724202529306e-06,
+      "loss": 2.6275,
+      "step": 3296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6534728765032735e-06,
+      "loss": 2.6191,
+      "step": 3297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6502228224564674e-06,
+      "loss": 2.6425,
+      "step": 3298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.646974042150642e-06,
+      "loss": 2.7057,
+      "step": 3299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.643726537346858e-06,
+      "loss": 2.6791,
+      "step": 3300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6404803098054864e-06,
+      "loss": 2.6741,
+      "step": 3301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6372353612862097e-06,
+      "loss": 2.6667,
+      "step": 3302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6339916935480074e-06,
+      "loss": 2.5933,
+      "step": 3303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6307493083491743e-06,
+      "loss": 2.5993,
+      "step": 3304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.627508207447308e-06,
+      "loss": 2.6687,
+      "step": 3305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.624268392599304e-06,
+      "loss": 2.6504,
+      "step": 3306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6210298655613675e-06,
+      "loss": 2.8095,
+      "step": 3307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.617792628089004e-06,
+      "loss": 2.6407,
+      "step": 3308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.614556681937016e-06,
+      "loss": 2.5914,
+      "step": 3309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6113220288595108e-06,
+      "loss": 2.6431,
+      "step": 3310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6080886706098953e-06,
+      "loss": 2.5787,
+      "step": 3311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6048566089408677e-06,
+      "loss": 2.6024,
+      "step": 3312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6016258456044324e-06,
+      "loss": 2.6328,
+      "step": 3313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.598396382351881e-06,
+      "loss": 2.648,
+      "step": 3314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.595168220933808e-06,
+      "loss": 2.6744,
+      "step": 3315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.591941363100097e-06,
+      "loss": 2.6276,
+      "step": 3316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5887158105999296e-06,
+      "loss": 2.6204,
+      "step": 3317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5854915651817752e-06,
+      "loss": 2.6349,
+      "step": 3318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5822686285934003e-06,
+      "loss": 2.6618,
+      "step": 3319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.579047002581853e-06,
+      "loss": 2.6182,
+      "step": 3320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.575826688893479e-06,
+      "loss": 2.6529,
+      "step": 3321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5726076892739127e-06,
+      "loss": 2.6714,
+      "step": 3322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5693900054680687e-06,
+      "loss": 2.6813,
+      "step": 3323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.566173639220155e-06,
+      "loss": 2.5702,
+      "step": 3324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.562958592273666e-06,
+      "loss": 2.5918,
+      "step": 3325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.559744866371374e-06,
+      "loss": 2.6733,
+      "step": 3326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5565324632553424e-06,
+      "loss": 2.6715,
+      "step": 3327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.553321384666916e-06,
+      "loss": 2.6722,
+      "step": 3328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5501116323467158e-06,
+      "loss": 2.6869,
+      "step": 3329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5469032080346546e-06,
+      "loss": 2.6226,
+      "step": 3330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.543696113469913e-06,
+      "loss": 2.6881,
+      "step": 3331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.540490350390961e-06,
+      "loss": 2.6106,
+      "step": 3332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5372859205355438e-06,
+      "loss": 2.635,
+      "step": 3333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.534082825640679e-06,
+      "loss": 2.6057,
+      "step": 3334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5308810674426664e-06,
+      "loss": 2.5823,
+      "step": 3335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5276806476770823e-06,
+      "loss": 2.5575,
+      "step": 3336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5244815680787688e-06,
+      "loss": 2.6762,
+      "step": 3337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5212838303818504e-06,
+      "loss": 2.6,
+      "step": 3338
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5180874363197217e-06,
+      "loss": 2.6776,
+      "step": 3339
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5148923876250488e-06,
+      "loss": 2.6363,
+      "step": 3340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5116986860297644e-06,
+      "loss": 2.6612,
+      "step": 3341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.508506333265077e-06,
+      "loss": 2.6565,
+      "step": 3342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5053153310614618e-06,
+      "loss": 2.7092,
+      "step": 3343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5021256811486615e-06,
+      "loss": 2.6535,
+      "step": 3344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4989373852556882e-06,
+      "loss": 2.6255,
+      "step": 3345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4957504451108123e-06,
+      "loss": 2.646,
+      "step": 3346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.492564862441578e-06,
+      "loss": 2.6091,
+      "step": 3347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.489380638974792e-06,
+      "loss": 2.6949,
+      "step": 3348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4861977764365193e-06,
+      "loss": 2.6436,
+      "step": 3349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.483016276552093e-06,
+      "loss": 2.6458,
+      "step": 3350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4798361410461032e-06,
+      "loss": 2.5949,
+      "step": 3351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.476657371642402e-06,
+      "loss": 2.6374,
+      "step": 3352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4734799700641044e-06,
+      "loss": 2.6388,
+      "step": 3353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4703039380335764e-06,
+      "loss": 2.6556,
+      "step": 3354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.467129277272449e-06,
+      "loss": 2.678,
+      "step": 3355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.463955989501607e-06,
+      "loss": 2.6213,
+      "step": 3356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.460784076441188e-06,
+      "loss": 2.6325,
+      "step": 3357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.45761353981059e-06,
+      "loss": 2.697,
+      "step": 3358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4544443813284626e-06,
+      "loss": 2.6433,
+      "step": 3359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4512766027127043e-06,
+      "loss": 2.6948,
+      "step": 3360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.448110205680472e-06,
+      "loss": 2.6405,
+      "step": 3361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.444945191948171e-06,
+      "loss": 2.6321,
+      "step": 3362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4417815632314545e-06,
+      "loss": 2.617,
+      "step": 3363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4386193212452282e-06,
+      "loss": 2.7497,
+      "step": 3364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4354584677036447e-06,
+      "loss": 2.6276,
+      "step": 3365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.432299004320105e-06,
+      "loss": 2.6757,
+      "step": 3366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4291409328072567e-06,
+      "loss": 2.6937,
+      "step": 3367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.425984254876989e-06,
+      "loss": 2.6213,
+      "step": 3368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.42282897224044e-06,
+      "loss": 2.6516,
+      "step": 3369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4196750866079926e-06,
+      "loss": 2.6153,
+      "step": 3370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.416522599689265e-06,
+      "loss": 2.5948,
+      "step": 3371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.413371513193125e-06,
+      "loss": 2.6158,
+      "step": 3372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.41022182882768e-06,
+      "loss": 2.6297,
+      "step": 3373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4070735483002715e-06,
+      "loss": 2.6733,
+      "step": 3374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.403926673317487e-06,
+      "loss": 2.6192,
+      "step": 3375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4007812055851505e-06,
+      "loss": 2.7399,
+      "step": 3376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.397637146808319e-06,
+      "loss": 2.6543,
+      "step": 3377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.39449449869129e-06,
+      "loss": 2.6259,
+      "step": 3378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.391353262937597e-06,
+      "loss": 2.6749,
+      "step": 3379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3882134412500024e-06,
+      "loss": 2.6198,
+      "step": 3380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3850750353305076e-06,
+      "loss": 2.6897,
+      "step": 3381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3819380468803465e-06,
+      "loss": 2.726,
+      "step": 3382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3788024775999797e-06,
+      "loss": 2.6685,
+      "step": 3383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3756683291891024e-06,
+      "loss": 2.6826,
+      "step": 3384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3725356033466413e-06,
+      "loss": 2.6532,
+      "step": 3385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3694043017707453e-06,
+      "loss": 2.6585,
+      "step": 3386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3662744261587974e-06,
+      "loss": 2.7155,
+      "step": 3387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.363145978207408e-06,
+      "loss": 2.6602,
+      "step": 3388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3600189596124062e-06,
+      "loss": 2.5648,
+      "step": 3389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.356893372068855e-06,
+      "loss": 2.5983,
+      "step": 3390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3537692172710364e-06,
+      "loss": 2.7076,
+      "step": 3391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3506464969124583e-06,
+      "loss": 2.5825,
+      "step": 3392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.347525212685852e-06,
+      "loss": 2.6187,
+      "step": 3393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.344405366283164e-06,
+      "loss": 2.7154,
+      "step": 3394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.341286959395568e-06,
+      "loss": 2.6793,
+      "step": 3395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3381699937134573e-06,
+      "loss": 2.7009,
+      "step": 3396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3350544709264377e-06,
+      "loss": 2.6818,
+      "step": 3397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.331940392723339e-06,
+      "loss": 2.5899,
+      "step": 3398
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3288277607922083e-06,
+      "loss": 2.6467,
+      "step": 3399
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3257165768203023e-06,
+      "loss": 2.6486,
+      "step": 3400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3226068424940977e-06,
+      "loss": 2.6688,
+      "step": 3401
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3194985594992886e-06,
+      "loss": 2.7263,
+      "step": 3402
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3163917295207727e-06,
+      "loss": 2.6499,
+      "step": 3403
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.313286354242669e-06,
+      "loss": 2.6823,
+      "step": 3404
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.310182435348305e-06,
+      "loss": 2.6996,
+      "step": 3405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.307079974520216e-06,
+      "loss": 2.6513,
+      "step": 3406
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3039789734401524e-06,
+      "loss": 2.6777,
+      "step": 3407
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3008794337890668e-06,
+      "loss": 2.628,
+      "step": 3408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2977813572471247e-06,
+      "loss": 2.6652,
+      "step": 3409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.294684745493699e-06,
+      "loss": 2.6418,
+      "step": 3410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2915896002073623e-06,
+      "loss": 2.6945,
+      "step": 3411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2884959230658982e-06,
+      "loss": 2.6566,
+      "step": 3412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2854037157462934e-06,
+      "loss": 2.6553,
+      "step": 3413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2823129799247367e-06,
+      "loss": 2.6584,
+      "step": 3414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.279223717276621e-06,
+      "loss": 2.6969,
+      "step": 3415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.276135929476537e-06,
+      "loss": 2.652,
+      "step": 3416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.273049618198278e-06,
+      "loss": 2.6993,
+      "step": 3417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2699647851148394e-06,
+      "loss": 2.6675,
+      "step": 3418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2668814318984135e-06,
+      "loss": 2.6863,
+      "step": 3419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.263799560220386e-06,
+      "loss": 2.6843,
+      "step": 3420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2607191717513466e-06,
+      "loss": 2.6775,
+      "step": 3421
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.257640268161079e-06,
+      "loss": 2.6673,
+      "step": 3422
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2545628511185575e-06,
+      "loss": 2.608,
+      "step": 3423
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.251486922291957e-06,
+      "loss": 2.6463,
+      "step": 3424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2484124833486405e-06,
+      "loss": 2.677,
+      "step": 3425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2453395359551655e-06,
+      "loss": 2.5861,
+      "step": 3426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.242268081777284e-06,
+      "loss": 2.689,
+      "step": 3427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2391981224799313e-06,
+      "loss": 2.6624,
+      "step": 3428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.236129659727239e-06,
+      "loss": 2.6186,
+      "step": 3429
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2330626951825273e-06,
+      "loss": 2.6339,
+      "step": 3430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2299972305082974e-06,
+      "loss": 2.6637,
+      "step": 3431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2269332673662446e-06,
+      "loss": 2.6399,
+      "step": 3432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2238708074172494e-06,
+      "loss": 2.6729,
+      "step": 3433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.220809852321373e-06,
+      "loss": 2.7001,
+      "step": 3434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2177504037378643e-06,
+      "loss": 2.638,
+      "step": 3435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.214692463325158e-06,
+      "loss": 2.7031,
+      "step": 3436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.211636032740864e-06,
+      "loss": 2.7039,
+      "step": 3437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2085811136417807e-06,
+      "loss": 2.6452,
+      "step": 3438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.205527707683883e-06,
+      "loss": 2.5616,
+      "step": 3439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.202475816522329e-06,
+      "loss": 2.6819,
+      "step": 3440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1994254418114524e-06,
+      "loss": 2.6206,
+      "step": 3441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1963765852047684e-06,
+      "loss": 2.705,
+      "step": 3442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.193329248354964e-06,
+      "loss": 2.6622,
+      "step": 3443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.190283432913909e-06,
+      "loss": 2.7014,
+      "step": 3444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.187239140532641e-06,
+      "loss": 2.5978,
+      "step": 3445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.184196372861378e-06,
+      "loss": 2.7076,
+      "step": 3446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1811551315495117e-06,
+      "loss": 2.634,
+      "step": 3447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1781154182456e-06,
+      "loss": 2.644,
+      "step": 3448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1750772345973792e-06,
+      "loss": 2.7215,
+      "step": 3449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1720405822517553e-06,
+      "loss": 2.6766,
+      "step": 3450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1690054628548e-06,
+      "loss": 2.6465,
+      "step": 3451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.165971878051757e-06,
+      "loss": 2.6375,
+      "step": 3452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.162939829487042e-06,
+      "loss": 2.5823,
+      "step": 3453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.159909318804229e-06,
+      "loss": 2.6816,
+      "step": 3454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1568803476460663e-06,
+      "loss": 2.646,
+      "step": 3455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.153852917654466e-06,
+      "loss": 2.6226,
+      "step": 3456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1508270304705e-06,
+      "loss": 2.6239,
+      "step": 3457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.147802687734409e-06,
+      "loss": 2.6156,
+      "step": 3458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1447798910855967e-06,
+      "loss": 2.6762,
+      "step": 3459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.141758642162624e-06,
+      "loss": 2.6727,
+      "step": 3460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1387389426032166e-06,
+      "loss": 2.6309,
+      "step": 3461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.135720794044261e-06,
+      "loss": 2.6472,
+      "step": 3462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.132704198121802e-06,
+      "loss": 2.6596,
+      "step": 3463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1296891564710394e-06,
+      "loss": 2.608,
+      "step": 3464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.126675670726334e-06,
+      "loss": 2.6529,
+      "step": 3465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1236637425212036e-06,
+      "loss": 2.6605,
+      "step": 3466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.120653373488322e-06,
+      "loss": 2.5814,
+      "step": 3467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1176445652595136e-06,
+      "loss": 2.6051,
+      "step": 3468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1146373194657597e-06,
+      "loss": 2.6105,
+      "step": 3469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1116316377371983e-06,
+      "loss": 2.6385,
+      "step": 3470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1086275217031104e-06,
+      "loss": 2.599,
+      "step": 3471
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1056249729919363e-06,
+      "loss": 2.6093,
+      "step": 3472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1026239932312664e-06,
+      "loss": 2.6781,
+      "step": 3473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.099624584047834e-06,
+      "loss": 2.5979,
+      "step": 3474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0966267470675273e-06,
+      "loss": 2.7162,
+      "step": 3475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.093630483915382e-06,
+      "loss": 2.6451,
+      "step": 3476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.090635796215576e-06,
+      "loss": 2.6497,
+      "step": 3477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0876426855914372e-06,
+      "loss": 2.6684,
+      "step": 3478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0846511536654395e-06,
+      "loss": 2.6506,
+      "step": 3479
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0816612020591957e-06,
+      "loss": 2.5886,
+      "step": 3480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.078672832393469e-06,
+      "loss": 2.6391,
+      "step": 3481
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0756860462881584e-06,
+      "loss": 2.6647,
+      "step": 3482
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0727008453623083e-06,
+      "loss": 2.6163,
+      "step": 3483
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0697172312341057e-06,
+      "loss": 2.6709,
+      "step": 3484
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0667352055208712e-06,
+      "loss": 2.6637,
+      "step": 3485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0637547698390693e-06,
+      "loss": 2.7299,
+      "step": 3486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0607759258043015e-06,
+      "loss": 2.7425,
+      "step": 3487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0577986750313063e-06,
+      "loss": 2.6344,
+      "step": 3488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0548230191339586e-06,
+      "loss": 2.6592,
+      "step": 3489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.051848959725271e-06,
+      "loss": 2.6275,
+      "step": 3490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.048876498417384e-06,
+      "loss": 2.6389,
+      "step": 3491
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0459056368215786e-06,
+      "loss": 2.6054,
+      "step": 3492
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.042936376548268e-06,
+      "loss": 2.6942,
+      "step": 3493
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0399687192069916e-06,
+      "loss": 2.6033,
+      "step": 3494
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.037002666406427e-06,
+      "loss": 2.6206,
+      "step": 3495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.034038219754379e-06,
+      "loss": 2.6497,
+      "step": 3496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.03107538085778e-06,
+      "loss": 2.6147,
+      "step": 3497
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0281141513226928e-06,
+      "loss": 2.67,
+      "step": 3498
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0251545327543103e-06,
+      "loss": 2.583,
+      "step": 3499
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.022196526756946e-06,
+      "loss": 2.5972,
+      "step": 3500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.019240134934046e-06,
+      "loss": 2.6103,
+      "step": 3501
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0162853588881747e-06,
+      "loss": 2.6865,
+      "step": 3502
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.013332200221026e-06,
+      "loss": 2.6527,
+      "step": 3503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.0103806605334164e-06,
+      "loss": 2.6688,
+      "step": 3504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.0074307414252813e-06,
+      "loss": 2.6101,
+      "step": 3505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.00448244449568e-06,
+      "loss": 2.6441,
+      "step": 3506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.0015357713427956e-06,
+      "loss": 2.6877,
+      "step": 3507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.998590723563924e-06,
+      "loss": 2.6248,
+      "step": 3508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9956473027554846e-06,
+      "loss": 2.6189,
+      "step": 3509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9927055105130155e-06,
+      "loss": 2.7095,
+      "step": 3510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.989765348431171e-06,
+      "loss": 2.6014,
+      "step": 3511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9868268181037186e-06,
+      "loss": 2.5836,
+      "step": 3512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9838899211235456e-06,
+      "loss": 2.6471,
+      "step": 3513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9809546590826528e-06,
+      "loss": 2.6729,
+      "step": 3514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9780210335721533e-06,
+      "loss": 2.6876,
+      "step": 3515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975089046182277e-06,
+      "loss": 2.6129,
+      "step": 3516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.972158698502358e-06,
+      "loss": 2.6423,
+      "step": 3517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9692299921208514e-06,
+      "loss": 2.677,
+      "step": 3518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9663029286253125e-06,
+      "loss": 2.6435,
+      "step": 3519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.963377509602415e-06,
+      "loss": 2.608,
+      "step": 3520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9604537366379374e-06,
+      "loss": 2.7004,
+      "step": 3521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.957531611316763e-06,
+      "loss": 2.6802,
+      "step": 3522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.954611135222887e-06,
+      "loss": 2.6623,
+      "step": 3523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9516923099394098e-06,
+      "loss": 2.6511,
+      "step": 3524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.948775137048532e-06,
+      "loss": 2.6506,
+      "step": 3525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9458596181315643e-06,
+      "loss": 2.6643,
+      "step": 3526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9429457547689202e-06,
+      "loss": 2.5896,
+      "step": 3527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9400335485401108e-06,
+      "loss": 2.6084,
+      "step": 3528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9371230010237533e-06,
+      "loss": 2.6805,
+      "step": 3529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.934214113797568e-06,
+      "loss": 2.5856,
+      "step": 3530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9313068884383675e-06,
+      "loss": 2.6195,
+      "step": 3531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9284013265220704e-06,
+      "loss": 2.6662,
+      "step": 3532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9254974296236923e-06,
+      "loss": 2.6359,
+      "step": 3533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.922595199317342e-06,
+      "loss": 2.6846,
+      "step": 3534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9196946371762302e-06,
+      "loss": 2.6942,
+      "step": 3535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.916795744772661e-06,
+      "loss": 2.6985,
+      "step": 3536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.913898523678034e-06,
+      "loss": 2.6153,
+      "step": 3537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.911002975462843e-06,
+      "loss": 2.6085,
+      "step": 3538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9081091016966726e-06,
+      "loss": 2.6645,
+      "step": 3539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9052169039482021e-06,
+      "loss": 2.6544,
+      "step": 3540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.902326383785204e-06,
+      "loss": 2.6124,
+      "step": 3541
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8994375427745354e-06,
+      "loss": 2.6142,
+      "step": 3542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8965503824821496e-06,
+      "loss": 2.7093,
+      "step": 3543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.893664904473087e-06,
+      "loss": 2.5922,
+      "step": 3544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.890781110311472e-06,
+      "loss": 2.6567,
+      "step": 3545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.887899001560522e-06,
+      "loss": 2.5766,
+      "step": 3546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8850185797825387e-06,
+      "loss": 2.6745,
+      "step": 3547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8821398465389074e-06,
+      "loss": 2.6071,
+      "step": 3548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8792628033900996e-06,
+      "loss": 2.6608,
+      "step": 3549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8763874518956732e-06,
+      "loss": 2.6754,
+      "step": 3550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8735137936142633e-06,
+      "loss": 2.6622,
+      "step": 3551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8706418301035922e-06,
+      "loss": 2.5588,
+      "step": 3552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8677715629204634e-06,
+      "loss": 2.712,
+      "step": 3553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8649029936207562e-06,
+      "loss": 2.5905,
+      "step": 3554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8620361237594353e-06,
+      "loss": 2.6531,
+      "step": 3555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.859170954890539e-06,
+      "loss": 2.6179,
+      "step": 3556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8563074885671872e-06,
+      "loss": 2.6481,
+      "step": 3557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8534457263415755e-06,
+      "loss": 2.6328,
+      "step": 3558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8505856697649782e-06,
+      "loss": 2.6305,
+      "step": 3559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.84772732038774e-06,
+      "loss": 2.6124,
+      "step": 3560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.844870679759283e-06,
+      "loss": 2.7188,
+      "step": 3561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8420157494281048e-06,
+      "loss": 2.6347,
+      "step": 3562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8391625309417733e-06,
+      "loss": 2.5938,
+      "step": 3563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8363110258469313e-06,
+      "loss": 2.6279,
+      "step": 3564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8334612356892872e-06,
+      "loss": 2.6082,
+      "step": 3565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8306131620136254e-06,
+      "loss": 2.6212,
+      "step": 3566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8277668063637993e-06,
+      "loss": 2.6493,
+      "step": 3567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8249221702827263e-06,
+      "loss": 2.6484,
+      "step": 3568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8220792553123967e-06,
+      "loss": 2.6328,
+      "step": 3569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8192380629938683e-06,
+      "loss": 2.6656,
+      "step": 3570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8163985948672585e-06,
+      "loss": 2.6258,
+      "step": 3571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8135608524717569e-06,
+      "loss": 2.6008,
+      "step": 3572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8107248373456166e-06,
+      "loss": 2.6724,
+      "step": 3573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8078905510261502e-06,
+      "loss": 2.613,
+      "step": 3574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8050579950497383e-06,
+      "loss": 2.5896,
+      "step": 3575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8022271709518191e-06,
+      "loss": 2.6528,
+      "step": 3576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7993980802668947e-06,
+      "loss": 2.6364,
+      "step": 3577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7965707245285297e-06,
+      "loss": 2.6629,
+      "step": 3578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7937451052693411e-06,
+      "loss": 2.5258,
+      "step": 3579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.790921224021011e-06,
+      "loss": 2.6257,
+      "step": 3580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7880990823142803e-06,
+      "loss": 2.6245,
+      "step": 3581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7852786816789392e-06,
+      "loss": 2.6424,
+      "step": 3582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.782460023643841e-06,
+      "loss": 2.6385,
+      "step": 3583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7796431097368933e-06,
+      "loss": 2.5901,
+      "step": 3584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7768279414850565e-06,
+      "loss": 2.7313,
+      "step": 3585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7740145204143472e-06,
+      "loss": 2.6225,
+      "step": 3586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.77120284804983e-06,
+      "loss": 2.609,
+      "step": 3587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7683929259156269e-06,
+      "loss": 2.5658,
+      "step": 3588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7655847555349087e-06,
+      "loss": 2.6332,
+      "step": 3589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7627783384298996e-06,
+      "loss": 2.6455,
+      "step": 3590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7599736761218661e-06,
+      "loss": 2.6532,
+      "step": 3591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7571707701311308e-06,
+      "loss": 2.6609,
+      "step": 3592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7543696219770634e-06,
+      "loss": 2.6611,
+      "step": 3593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7515702331780753e-06,
+      "loss": 2.6353,
+      "step": 3594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7487726052516318e-06,
+      "loss": 2.6062,
+      "step": 3595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7459767397142359e-06,
+      "loss": 2.615,
+      "step": 3596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.743182638081441e-06,
+      "loss": 2.697,
+      "step": 3597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7403903018678447e-06,
+      "loss": 2.669,
+      "step": 3598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7375997325870813e-06,
+      "loss": 2.7607,
+      "step": 3599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7348109317518335e-06,
+      "loss": 2.7445,
+      "step": 3600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7320239008738248e-06,
+      "loss": 2.6954,
+      "step": 3601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7292386414638147e-06,
+      "loss": 2.6019,
+      "step": 3602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7264551550316072e-06,
+      "loss": 2.6176,
+      "step": 3603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7236734430860452e-06,
+      "loss": 2.6843,
+      "step": 3604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7208935071350047e-06,
+      "loss": 2.6367,
+      "step": 3605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7181153486854035e-06,
+      "loss": 2.5948,
+      "step": 3606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7153389692431977e-06,
+      "loss": 2.6563,
+      "step": 3607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.712564370313371e-06,
+      "loss": 2.6637,
+      "step": 3608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7097915533999498e-06,
+      "loss": 2.6502,
+      "step": 3609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.707020520005992e-06,
+      "loss": 2.6314,
+      "step": 3610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7042512716335873e-06,
+      "loss": 2.6642,
+      "step": 3611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7014838097838616e-06,
+      "loss": 2.6708,
+      "step": 3612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.698718135956966e-06,
+      "loss": 2.6321,
+      "step": 3613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6959542516520878e-06,
+      "loss": 2.6801,
+      "step": 3614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6931921583674442e-06,
+      "loss": 2.6916,
+      "step": 3615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6904318576002766e-06,
+      "loss": 2.7053,
+      "step": 3616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6876733508468596e-06,
+      "loss": 2.6202,
+      "step": 3617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6849166396024957e-06,
+      "loss": 2.6825,
+      "step": 3618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6821617253615085e-06,
+      "loss": 2.6674,
+      "step": 3619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.679408609617253e-06,
+      "loss": 2.6349,
+      "step": 3620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6766572938621084e-06,
+      "loss": 2.5911,
+      "step": 3621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6739077795874753e-06,
+      "loss": 2.6582,
+      "step": 3622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6711600682837804e-06,
+      "loss": 2.6284,
+      "step": 3623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6684141614404743e-06,
+      "loss": 2.6347,
+      "step": 3624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6656700605460241e-06,
+      "loss": 2.5685,
+      "step": 3625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.662927767087923e-06,
+      "loss": 2.6138,
+      "step": 3626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6601872825526844e-06,
+      "loss": 2.6048,
+      "step": 3627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6574486084258369e-06,
+      "loss": 2.5482,
+      "step": 3628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.654711746191931e-06,
+      "loss": 2.6903,
+      "step": 3629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.651976697334537e-06,
+      "loss": 2.7237,
+      "step": 3630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6492434633362353e-06,
+      "loss": 2.6154,
+      "step": 3631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6465120456786299e-06,
+      "loss": 2.6503,
+      "step": 3632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6437824458423364e-06,
+      "loss": 2.5702,
+      "step": 3633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6410546653069876e-06,
+      "loss": 2.668,
+      "step": 3634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6383287055512248e-06,
+      "loss": 2.6632,
+      "step": 3635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6356045680527078e-06,
+      "loss": 2.6327,
+      "step": 3636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6328822542881063e-06,
+      "loss": 2.6444,
+      "step": 3637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6301617657331036e-06,
+      "loss": 2.6169,
+      "step": 3638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6274431038623889e-06,
+      "loss": 2.7123,
+      "step": 3639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6247262701496647e-06,
+      "loss": 2.6729,
+      "step": 3640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6220112660676446e-06,
+      "loss": 2.6921,
+      "step": 3641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6192980930880436e-06,
+      "loss": 2.6717,
+      "step": 3642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6165867526815904e-06,
+      "loss": 2.6762,
+      "step": 3643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6138772463180197e-06,
+      "loss": 2.6258,
+      "step": 3644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6111695754660667e-06,
+      "loss": 2.6422,
+      "step": 3645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6084637415934772e-06,
+      "loss": 2.6206,
+      "step": 3646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6057597461670021e-06,
+      "loss": 2.6465,
+      "step": 3647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6030575906523882e-06,
+      "loss": 2.634,
+      "step": 3648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6003572765143938e-06,
+      "loss": 2.6626,
+      "step": 3649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5976588052167724e-06,
+      "loss": 2.6713,
+      "step": 3650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5949621782222818e-06,
+      "loss": 2.6567,
+      "step": 3651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.592267396992681e-06,
+      "loss": 2.6939,
+      "step": 3652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.589574462988725e-06,
+      "loss": 2.6471,
+      "step": 3653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5868833776701697e-06,
+      "loss": 2.6824,
+      "step": 3654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5841941424957707e-06,
+      "loss": 2.6933,
+      "step": 3655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5815067589232752e-06,
+      "loss": 2.7013,
+      "step": 3656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5788212284094318e-06,
+      "loss": 2.6304,
+      "step": 3657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.576137552409983e-06,
+      "loss": 2.6306,
+      "step": 3658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5734557323796662e-06,
+      "loss": 2.6083,
+      "step": 3659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.570775769772212e-06,
+      "loss": 2.6218,
+      "step": 3660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5680976660403457e-06,
+      "loss": 2.6054,
+      "step": 3661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5654214226357822e-06,
+      "loss": 2.6408,
+      "step": 3662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5627470410092299e-06,
+      "loss": 2.5708,
+      "step": 3663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5600745226103897e-06,
+      "loss": 2.6608,
+      "step": 3664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5574038688879474e-06,
+      "loss": 2.6498,
+      "step": 3665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5547350812895828e-06,
+      "loss": 2.5546,
+      "step": 3666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5520681612619642e-06,
+      "loss": 2.6358,
+      "step": 3667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5494031102507423e-06,
+      "loss": 2.7091,
+      "step": 3668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.546739929700562e-06,
+      "loss": 2.6239,
+      "step": 3669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5440786210550462e-06,
+      "loss": 2.6243,
+      "step": 3670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.54141918575681e-06,
+      "loss": 2.6059,
+      "step": 3671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5387616252474518e-06,
+      "loss": 2.6445,
+      "step": 3672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5361059409675487e-06,
+      "loss": 2.5956,
+      "step": 3673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5334521343566665e-06,
+      "loss": 2.6545,
+      "step": 3674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5308002068533524e-06,
+      "loss": 2.7324,
+      "step": 3675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5281501598951304e-06,
+      "loss": 2.5989,
+      "step": 3676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5255019949185101e-06,
+      "loss": 2.6392,
+      "step": 3677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.522855713358981e-06,
+      "loss": 2.6765,
+      "step": 3678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5202113166510058e-06,
+      "loss": 2.5957,
+      "step": 3679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5175688062280325e-06,
+      "loss": 2.67,
+      "step": 3680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5149281835224823e-06,
+      "loss": 2.72,
+      "step": 3681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.512289449965757e-06,
+      "loss": 2.6175,
+      "step": 3682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5096526069882284e-06,
+      "loss": 2.707,
+      "step": 3683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5070176560192478e-06,
+      "loss": 2.6159,
+      "step": 3684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5043845984871413e-06,
+      "loss": 2.6281,
+      "step": 3685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5017534358192066e-06,
+      "loss": 2.6792,
+      "step": 3686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4991241694417163e-06,
+      "loss": 2.7462,
+      "step": 3687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4964968007799113e-06,
+      "loss": 2.6426,
+      "step": 3688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4938713312580089e-06,
+      "loss": 2.6421,
+      "step": 3689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4912477622991916e-06,
+      "loss": 2.6067,
+      "step": 3690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4886260953256155e-06,
+      "loss": 2.6122,
+      "step": 3691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4860063317584056e-06,
+      "loss": 2.6184,
+      "step": 3692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4833884730176522e-06,
+      "loss": 2.661,
+      "step": 3693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.480772520522416e-06,
+      "loss": 2.6725,
+      "step": 3694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4781584756907252e-06,
+      "loss": 2.6505,
+      "step": 3695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.475546339939568e-06,
+      "loss": 2.5426,
+      "step": 3696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4729361146849041e-06,
+      "loss": 2.6471,
+      "step": 3697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4703278013416567e-06,
+      "loss": 2.6158,
+      "step": 3698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4677214013237078e-06,
+      "loss": 2.7239,
+      "step": 3699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4651169160439078e-06,
+      "loss": 2.6459,
+      "step": 3700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4625143469140673e-06,
+      "loss": 2.6151,
+      "step": 3701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.459913695344956e-06,
+      "loss": 2.684,
+      "step": 3702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4573149627463079e-06,
+      "loss": 2.6105,
+      "step": 3703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4547181505268153e-06,
+      "loss": 2.6624,
+      "step": 3704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4521232600941265e-06,
+      "loss": 2.6577,
+      "step": 3705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4495302928548528e-06,
+      "loss": 2.659,
+      "step": 3706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4469392502145602e-06,
+      "loss": 2.5918,
+      "step": 3707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4443501335777732e-06,
+      "loss": 2.6259,
+      "step": 3708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4417629443479715e-06,
+      "loss": 2.6958,
+      "step": 3709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4391776839275872e-06,
+      "loss": 2.5639,
+      "step": 3710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4365943537180115e-06,
+      "loss": 2.6178,
+      "step": 3711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.434012955119588e-06,
+      "loss": 2.6114,
+      "step": 3712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4314334895316095e-06,
+      "loss": 2.6445,
+      "step": 3713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.428855958352326e-06,
+      "loss": 2.6532,
+      "step": 3714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.426280362978938e-06,
+      "loss": 2.6261,
+      "step": 3715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.423706704807592e-06,
+      "loss": 2.6352,
+      "step": 3716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.42113498523339e-06,
+      "loss": 2.6547,
+      "step": 3717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4185652056503817e-06,
+      "loss": 2.6637,
+      "step": 3718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4159973674515632e-06,
+      "loss": 2.6418,
+      "step": 3719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4134314720288795e-06,
+      "loss": 2.6188,
+      "step": 3720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4108675207732242e-06,
+      "loss": 2.513,
+      "step": 3721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.408305515074433e-06,
+      "loss": 2.6474,
+      "step": 3722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4057454563212907e-06,
+      "loss": 2.6265,
+      "step": 3723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4031873459015265e-06,
+      "loss": 2.6365,
+      "step": 3724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4006311852018095e-06,
+      "loss": 2.7005,
+      "step": 3725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3980769756077584e-06,
+      "loss": 2.6864,
+      "step": 3726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3955247185039267e-06,
+      "loss": 2.6698,
+      "step": 3727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3929744152738151e-06,
+      "loss": 2.5824,
+      "step": 3728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3904260672998637e-06,
+      "loss": 2.6167,
+      "step": 3729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3878796759634544e-06,
+      "loss": 2.612,
+      "step": 3730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.385335242644903e-06,
+      "loss": 2.6981,
+      "step": 3731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3827927687234694e-06,
+      "loss": 2.5605,
+      "step": 3732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.380252255577349e-06,
+      "loss": 2.6662,
+      "step": 3733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3777137045836753e-06,
+      "loss": 2.6809,
+      "step": 3734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.375177117118519e-06,
+      "loss": 2.6186,
+      "step": 3735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3726424945568818e-06,
+      "loss": 2.6248,
+      "step": 3736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3701098382727058e-06,
+      "loss": 2.6537,
+      "step": 3737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3675791496388658e-06,
+      "loss": 2.6943,
+      "step": 3738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3650504300271667e-06,
+      "loss": 2.739,
+      "step": 3739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3625236808083498e-06,
+      "loss": 2.657,
+      "step": 3740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3599989033520888e-06,
+      "loss": 2.5858,
+      "step": 3741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3574760990269843e-06,
+      "loss": 2.6964,
+      "step": 3742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3549552692005719e-06,
+      "loss": 2.684,
+      "step": 3743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3524364152393127e-06,
+      "loss": 2.7241,
+      "step": 3744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3499195385086006e-06,
+      "loss": 2.6208,
+      "step": 3745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3474046403727564e-06,
+      "loss": 2.6347,
+      "step": 3746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3448917221950264e-06,
+      "loss": 2.6736,
+      "step": 3747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.342380785337586e-06,
+      "loss": 2.7145,
+      "step": 3748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3398718311615367e-06,
+      "loss": 2.6835,
+      "step": 3749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3373648610269024e-06,
+      "loss": 2.6778,
+      "step": 3750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3348598762926345e-06,
+      "loss": 2.6814,
+      "step": 3751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3323568783166097e-06,
+      "loss": 2.6522,
+      "step": 3752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.32985586845562e-06,
+      "loss": 2.6313,
+      "step": 3753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3273568480653888e-06,
+      "loss": 2.618,
+      "step": 3754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3248598185005558e-06,
+      "loss": 2.6493,
+      "step": 3755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3223647811146828e-06,
+      "loss": 2.5133,
+      "step": 3756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3198717372602538e-06,
+      "loss": 2.6642,
+      "step": 3757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.317380688288667e-06,
+      "loss": 2.6682,
+      "step": 3758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.314891635550244e-06,
+      "loss": 2.6168,
+      "step": 3759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3124045803942231e-06,
+      "loss": 2.6701,
+      "step": 3760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3099195241687601e-06,
+      "loss": 2.6456,
+      "step": 3761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3074364682209239e-06,
+      "loss": 2.6762,
+      "step": 3762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3049554138967052e-06,
+      "loss": 2.625,
+      "step": 3763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3024763625410025e-06,
+      "loss": 2.6409,
+      "step": 3764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2999993154976342e-06,
+      "loss": 2.6153,
+      "step": 3765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2975242741093319e-06,
+      "loss": 2.6774,
+      "step": 3766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.295051239717735e-06,
+      "loss": 2.6651,
+      "step": 3767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2925802136634003e-06,
+      "loss": 2.7141,
+      "step": 3768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2901111972857948e-06,
+      "loss": 2.6631,
+      "step": 3769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2876441919232935e-06,
+      "loss": 2.6027,
+      "step": 3770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2851791989131834e-06,
+      "loss": 2.6291,
+      "step": 3771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2827162195916632e-06,
+      "loss": 2.6595,
+      "step": 3772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.280255255293833e-06,
+      "loss": 2.7016,
+      "step": 3773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2777963073537065e-06,
+      "loss": 2.5785,
+      "step": 3774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.275339377104205e-06,
+      "loss": 2.6995,
+      "step": 3775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2728844658771505e-06,
+      "loss": 2.6404,
+      "step": 3776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2704315750032747e-06,
+      "loss": 2.6854,
+      "step": 3777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2679807058122163e-06,
+      "loss": 2.5633,
+      "step": 3778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2655318596325105e-06,
+      "loss": 2.6775,
+      "step": 3779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2630850377916025e-06,
+      "loss": 2.6606,
+      "step": 3780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2606402416158391e-06,
+      "loss": 2.6764,
+      "step": 3781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2581974724304668e-06,
+      "loss": 2.6326,
+      "step": 3782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.255756731559637e-06,
+      "loss": 2.6422,
+      "step": 3783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2533180203263962e-06,
+      "loss": 2.66,
+      "step": 3784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2508813400526948e-06,
+      "loss": 2.6585,
+      "step": 3785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2484466920593829e-06,
+      "loss": 2.7413,
+      "step": 3786
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2460140776662045e-06,
+      "loss": 2.6704,
+      "step": 3787
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2435834981918054e-06,
+      "loss": 2.6546,
+      "step": 3788
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2411549549537283e-06,
+      "loss": 2.7519,
+      "step": 3789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2387284492684087e-06,
+      "loss": 2.6666,
+      "step": 3790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2363039824511803e-06,
+      "loss": 2.6548,
+      "step": 3791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.233881555816272e-06,
+      "loss": 2.5572,
+      "step": 3792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2314611706768043e-06,
+      "loss": 2.6744,
+      "step": 3793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.229042828344793e-06,
+      "loss": 2.7226,
+      "step": 3794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2266265301311482e-06,
+      "loss": 2.6795,
+      "step": 3795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2242122773456666e-06,
+      "loss": 2.6697,
+      "step": 3796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2218000712970407e-06,
+      "loss": 2.6787,
+      "step": 3797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2193899132928539e-06,
+      "loss": 2.6356,
+      "step": 3798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2169818046395743e-06,
+      "loss": 2.6421,
+      "step": 3799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2145757466425667e-06,
+      "loss": 2.6939,
+      "step": 3800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2121717406060756e-06,
+      "loss": 2.5732,
+      "step": 3801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.209769787833241e-06,
+      "loss": 2.6695,
+      "step": 3802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2073698896260855e-06,
+      "loss": 2.6704,
+      "step": 3803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2049720472855192e-06,
+      "loss": 2.5532,
+      "step": 3804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.202576262111339e-06,
+      "loss": 2.6267,
+      "step": 3805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2001825354022234e-06,
+      "loss": 2.6918,
+      "step": 3806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1977908684557377e-06,
+      "loss": 2.6685,
+      "step": 3807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1954012625683303e-06,
+      "loss": 2.6346,
+      "step": 3808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1930137190353335e-06,
+      "loss": 2.669,
+      "step": 3809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.190628239150957e-06,
+      "loss": 2.6147,
+      "step": 3810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1882448242082971e-06,
+      "loss": 2.6226,
+      "step": 3811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1858634754993303e-06,
+      "loss": 2.6143,
+      "step": 3812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1834841943149078e-06,
+      "loss": 2.7133,
+      "step": 3813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1811069819447658e-06,
+      "loss": 2.6897,
+      "step": 3814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1787318396775188e-06,
+      "loss": 2.6433,
+      "step": 3815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1763587688006535e-06,
+      "loss": 2.6945,
+      "step": 3816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1739877706005404e-06,
+      "loss": 2.6362,
+      "step": 3817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.171618846362424e-06,
+      "loss": 2.7162,
+      "step": 3818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.169251997370422e-06,
+      "loss": 2.6289,
+      "step": 3819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1668872249075325e-06,
+      "loss": 2.7176,
+      "step": 3820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.164524530255622e-06,
+      "loss": 2.6001,
+      "step": 3821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1621639146954344e-06,
+      "loss": 2.7348,
+      "step": 3822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1598053795065884e-06,
+      "loss": 2.6148,
+      "step": 3823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1574489259675685e-06,
+      "loss": 2.634,
+      "step": 3824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.155094555355737e-06,
+      "loss": 2.6439,
+      "step": 3825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1527422689473255e-06,
+      "loss": 2.5375,
+      "step": 3826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1503920680174335e-06,
+      "loss": 2.6446,
+      "step": 3827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1480439538400317e-06,
+      "loss": 2.6557,
+      "step": 3828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1456979276879605e-06,
+      "loss": 2.5843,
+      "step": 3829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1433539908329266e-06,
+      "loss": 2.6532,
+      "step": 3830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.141012144545507e-06,
+      "loss": 2.6637,
+      "step": 3831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.138672390095143e-06,
+      "loss": 2.6222,
+      "step": 3832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1363347287501409e-06,
+      "loss": 2.6527,
+      "step": 3833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1339991617776746e-06,
+      "loss": 2.6444,
+      "step": 3834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1316656904437844e-06,
+      "loss": 2.611,
+      "step": 3835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1293343160133697e-06,
+      "loss": 2.6501,
+      "step": 3836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.127005039750198e-06,
+      "loss": 2.7348,
+      "step": 3837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1246778629168947e-06,
+      "loss": 2.7025,
+      "step": 3838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1223527867749518e-06,
+      "loss": 2.6772,
+      "step": 3839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1200298125847215e-06,
+      "loss": 2.583,
+      "step": 3840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1177089416054132e-06,
+      "loss": 2.6742,
+      "step": 3841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1153901750951002e-06,
+      "loss": 2.6231,
+      "step": 3842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1130735143107142e-06,
+      "loss": 2.6099,
+      "step": 3843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1107589605080432e-06,
+      "loss": 2.5543,
+      "step": 3844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.108446514941735e-06,
+      "loss": 2.6457,
+      "step": 3845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1061361788652963e-06,
+      "loss": 2.6827,
+      "step": 3846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1038279535310852e-06,
+      "loss": 2.6913,
+      "step": 3847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1015218401903205e-06,
+      "loss": 2.6518,
+      "step": 3848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0992178400930753e-06,
+      "loss": 2.636,
+      "step": 3849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.096915954488274e-06,
+      "loss": 2.6391,
+      "step": 3850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0946161846236974e-06,
+      "loss": 2.6298,
+      "step": 3851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0923185317459805e-06,
+      "loss": 2.5437,
+      "step": 3852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.09002299710061e-06,
+      "loss": 2.6309,
+      "step": 3853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0877295819319205e-06,
+      "loss": 2.6835,
+      "step": 3854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0854382874831032e-06,
+      "loss": 2.6966,
+      "step": 3855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0831491149961964e-06,
+      "loss": 2.6256,
+      "step": 3856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0808620657120915e-06,
+      "loss": 2.7065,
+      "step": 3857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.078577140870523e-06,
+      "loss": 2.6472,
+      "step": 3858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0762943417100797e-06,
+      "loss": 2.5941,
+      "step": 3859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0740136694681968e-06,
+      "loss": 2.673,
+      "step": 3860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.071735125381153e-06,
+      "loss": 2.6134,
+      "step": 3861
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.069458710684077e-06,
+      "loss": 2.6838,
+      "step": 3862
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0671844266109443e-06,
+      "loss": 2.639,
+      "step": 3863
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.06491227439457e-06,
+      "loss": 2.6258,
+      "step": 3864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0626422552666194e-06,
+      "loss": 2.6707,
+      "step": 3865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0603743704575992e-06,
+      "loss": 2.6977,
+      "step": 3866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0581086211968571e-06,
+      "loss": 2.6536,
+      "step": 3867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0558450087125865e-06,
+      "loss": 2.6429,
+      "step": 3868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0535835342318223e-06,
+      "loss": 2.6969,
+      "step": 3869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0513241989804374e-06,
+      "loss": 2.7071,
+      "step": 3870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0490670041831474e-06,
+      "loss": 2.5936,
+      "step": 3871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0468119510635088e-06,
+      "loss": 2.6697,
+      "step": 3872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0445590408439133e-06,
+      "loss": 2.6666,
+      "step": 3873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0423082747455943e-06,
+      "loss": 2.611,
+      "step": 3874
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0400596539886233e-06,
+      "loss": 2.6697,
+      "step": 3875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0378131797919056e-06,
+      "loss": 2.7054,
+      "step": 3876
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.035568853373185e-06,
+      "loss": 2.6497,
+      "step": 3877
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0333266759490418e-06,
+      "loss": 2.6383,
+      "step": 3878
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0310866487348903e-06,
+      "loss": 2.6246,
+      "step": 3879
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0288487729449803e-06,
+      "loss": 2.6658,
+      "step": 3880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0266130497923926e-06,
+      "loss": 2.6651,
+      "step": 3881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.024379480489044e-06,
+      "loss": 2.6323,
+      "step": 3882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0221480662456845e-06,
+      "loss": 2.6254,
+      "step": 3883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0199188082718908e-06,
+      "loss": 2.623,
+      "step": 3884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0176917077760763e-06,
+      "loss": 2.7127,
+      "step": 3885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0154667659654832e-06,
+      "loss": 2.7064,
+      "step": 3886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0132439840461817e-06,
+      "loss": 2.6557,
+      "step": 3887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0110233632230726e-06,
+      "loss": 2.6887,
+      "step": 3888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0088049046998876e-06,
+      "loss": 2.6126,
+      "step": 3889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.006588609679181e-06,
+      "loss": 2.689,
+      "step": 3890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0043744793623384e-06,
+      "loss": 2.6119,
+      "step": 3891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0021625149495723e-06,
+      "loss": 2.6638,
+      "step": 3892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.999527176399165e-07,
+      "loss": 2.6088,
+      "step": 3893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.977450886312374e-07,
+      "loss": 2.5789,
+      "step": 3894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.955396291202174e-07,
+      "loss": 2.6903,
+      "step": 3895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.933363403023694e-07,
+      "loss": 2.6705,
+      "step": 3896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.911352233720295e-07,
+      "loss": 2.6853,
+      "step": 3897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.889362795223512e-07,
+      "loss": 2.6169,
+      "step": 3898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.867395099453153e-07,
+      "loss": 2.6731,
+      "step": 3899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.845449158317216e-07,
+      "loss": 2.6599,
+      "step": 3900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.823524983711935e-07,
+      "loss": 2.6456,
+      "step": 3901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.80162258752168e-07,
+      "loss": 2.6223,
+      "step": 3902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.779741981619084e-07,
+      "loss": 2.5874,
+      "step": 3903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.757883177864929e-07,
+      "loss": 2.5619,
+      "step": 3904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.736046188108205e-07,
+      "loss": 2.6778,
+      "step": 3905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.714231024186072e-07,
+      "loss": 2.6788,
+      "step": 3906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.692437697923818e-07,
+      "loss": 2.6739,
+      "step": 3907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.670666221134944e-07,
+      "loss": 2.69,
+      "step": 3908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.648916605621106e-07,
+      "loss": 2.6171,
+      "step": 3909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.62718886317206e-07,
+      "loss": 2.6313,
+      "step": 3910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.605483005565752e-07,
+      "loss": 2.6438,
+      "step": 3911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.583799044568264e-07,
+      "loss": 2.5991,
+      "step": 3912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.562136991933773e-07,
+      "loss": 2.6351,
+      "step": 3913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.54049685940463e-07,
+      "loss": 2.6256,
+      "step": 3914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.518878658711244e-07,
+      "loss": 2.6693,
+      "step": 3915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.497282401572189e-07,
+      "loss": 2.7007,
+      "step": 3916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.475708099694125e-07,
+      "loss": 2.5783,
+      "step": 3917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.454155764771794e-07,
+      "loss": 2.6318,
+      "step": 3918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.432625408488049e-07,
+      "loss": 2.6464,
+      "step": 3919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.411117042513845e-07,
+      "loss": 2.6332,
+      "step": 3920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.389630678508166e-07,
+      "loss": 2.676,
+      "step": 3921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.368166328118117e-07,
+      "loss": 2.6563,
+      "step": 3922
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.346724002978863e-07,
+      "loss": 2.6029,
+      "step": 3923
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.325303714713596e-07,
+      "loss": 2.644,
+      "step": 3924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.3039054749336e-07,
+      "loss": 2.5807,
+      "step": 3925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.28252929523819e-07,
+      "loss": 2.7255,
+      "step": 3926
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.261175187214738e-07,
+      "loss": 2.6133,
+      "step": 3927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.239843162438639e-07,
+      "loss": 2.6464,
+      "step": 3928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.218533232473303e-07,
+      "loss": 2.6637,
+      "step": 3929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.197245408870192e-07,
+      "loss": 2.6989,
+      "step": 3930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.175979703168775e-07,
+      "loss": 2.691,
+      "step": 3931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.154736126896513e-07,
+      "loss": 2.6875,
+      "step": 3932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.133514691568895e-07,
+      "loss": 2.6099,
+      "step": 3933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.112315408689415e-07,
+      "loss": 2.7305,
+      "step": 3934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.091138289749512e-07,
+      "loss": 2.6586,
+      "step": 3935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.069983346228662e-07,
+      "loss": 2.6729,
+      "step": 3936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.048850589594311e-07,
+      "loss": 2.6361,
+      "step": 3937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.027740031301835e-07,
+      "loss": 2.6299,
+      "step": 3938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.006651682794637e-07,
+      "loss": 2.658,
+      "step": 3939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.985585555504056e-07,
+      "loss": 2.6757,
+      "step": 3940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.964541660849358e-07,
+      "loss": 2.6755,
+      "step": 3941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.943520010237799e-07,
+      "loss": 2.6271,
+      "step": 3942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.922520615064573e-07,
+      "loss": 2.6621,
+      "step": 3943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.901543486712772e-07,
+      "loss": 2.6979,
+      "step": 3944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.88058863655345e-07,
+      "loss": 2.6683,
+      "step": 3945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.8596560759456e-07,
+      "loss": 2.5795,
+      "step": 3946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.838745816236082e-07,
+      "loss": 2.5769,
+      "step": 3947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.817857868759705e-07,
+      "loss": 2.6097,
+      "step": 3948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.796992244839192e-07,
+      "loss": 2.6493,
+      "step": 3949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.776148955785124e-07,
+      "loss": 2.6431,
+      "step": 3950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.755328012896002e-07,
+      "loss": 2.6604,
+      "step": 3951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.734529427458221e-07,
+      "loss": 2.6675,
+      "step": 3952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.71375321074604e-07,
+      "loss": 2.6792,
+      "step": 3953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.69299937402161e-07,
+      "loss": 2.722,
+      "step": 3954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.672267928534916e-07,
+      "loss": 2.6433,
+      "step": 3955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.651558885523842e-07,
+      "loss": 2.6749,
+      "step": 3956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.630872256214129e-07,
+      "loss": 2.6535,
+      "step": 3957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.610208051819324e-07,
+      "loss": 2.6556,
+      "step": 3958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.589566283540862e-07,
+      "loss": 2.5884,
+      "step": 3959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.56894696256802e-07,
+      "loss": 2.6272,
+      "step": 3960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.548350100077857e-07,
+      "loss": 2.6796,
+      "step": 3961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.527775707235308e-07,
+      "loss": 2.6307,
+      "step": 3962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.50722379519312e-07,
+      "loss": 2.6348,
+      "step": 3963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.48669437509182e-07,
+      "loss": 2.6157,
+      "step": 3964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.466187458059782e-07,
+      "loss": 2.616,
+      "step": 3965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.445703055213172e-07,
+      "loss": 2.6623,
+      "step": 3966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.42524117765593e-07,
+      "loss": 2.656,
+      "step": 3967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.404801836479809e-07,
+      "loss": 2.7,
+      "step": 3968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.384385042764354e-07,
+      "loss": 2.6086,
+      "step": 3969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.363990807576844e-07,
+      "loss": 2.6215,
+      "step": 3970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.343619141972386e-07,
+      "loss": 2.6481,
+      "step": 3971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.323270056993798e-07,
+      "loss": 2.6944,
+      "step": 3972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.302943563671705e-07,
+      "loss": 2.7285,
+      "step": 3973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.282639673024456e-07,
+      "loss": 2.7034,
+      "step": 3974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.262358396058162e-07,
+      "loss": 2.74,
+      "step": 3975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.242099743766685e-07,
+      "loss": 2.6735,
+      "step": 3976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.22186372713158e-07,
+      "loss": 2.6682,
+      "step": 3977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.201650357122171e-07,
+      "loss": 2.636,
+      "step": 3978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.181459644695505e-07,
+      "loss": 2.6288,
+      "step": 3979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.161291600796345e-07,
+      "loss": 2.6554,
+      "step": 3980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.141146236357133e-07,
+      "loss": 2.6336,
+      "step": 3981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.121023562298064e-07,
+      "loss": 2.6785,
+      "step": 3982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.100923589527016e-07,
+      "loss": 2.6443,
+      "step": 3983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.080846328939528e-07,
+      "loss": 2.7175,
+      "step": 3984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.060791791418887e-07,
+      "loss": 2.6491,
+      "step": 3985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.040759987836033e-07,
+      "loss": 2.6641,
+      "step": 3986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.020750929049559e-07,
+      "loss": 2.6238,
+      "step": 3987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.000764625905777e-07,
+      "loss": 2.6425,
+      "step": 3988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.980801089238616e-07,
+      "loss": 2.6454,
+      "step": 3989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.960860329869701e-07,
+      "loss": 2.6912,
+      "step": 3990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.940942358608306e-07,
+      "loss": 2.7095,
+      "step": 3991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.921047186251312e-07,
+      "loss": 2.6252,
+      "step": 3992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.901174823583296e-07,
+      "loss": 2.7093,
+      "step": 3993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.881325281376457e-07,
+      "loss": 2.6397,
+      "step": 3994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.86149857039058e-07,
+      "loss": 2.6502,
+      "step": 3995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.841694701373126e-07,
+      "loss": 2.6803,
+      "step": 3996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.821913685059168e-07,
+      "loss": 2.6242,
+      "step": 3997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.802155532171346e-07,
+      "loss": 2.6621,
+      "step": 3998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.782420253419964e-07,
+      "loss": 2.7179,
+      "step": 3999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.762707859502888e-07,
+      "loss": 2.6486,
+      "step": 4000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.743018361105603e-07,
+      "loss": 2.6026,
+      "step": 4001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.723351768901172e-07,
+      "loss": 2.6391,
+      "step": 4002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.703708093550244e-07,
+      "loss": 2.6563,
+      "step": 4003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.68408734570103e-07,
+      "loss": 2.6075,
+      "step": 4004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.664489535989328e-07,
+      "loss": 2.6847,
+      "step": 4005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.644914675038523e-07,
+      "loss": 2.6339,
+      "step": 4006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.625362773459505e-07,
+      "loss": 2.647,
+      "step": 4007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.605833841850779e-07,
+      "loss": 2.7416,
+      "step": 4008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.58632789079834e-07,
+      "loss": 2.687,
+      "step": 4009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.566844930875778e-07,
+      "loss": 2.5859,
+      "step": 4010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.5473849726442e-07,
+      "loss": 2.6619,
+      "step": 4011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.527948026652226e-07,
+      "loss": 2.6026,
+      "step": 4012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.508534103436027e-07,
+      "loss": 2.6404,
+      "step": 4013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.489143213519301e-07,
+      "loss": 2.5823,
+      "step": 4014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.469775367413218e-07,
+      "loss": 2.6003,
+      "step": 4015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.450430575616496e-07,
+      "loss": 2.6206,
+      "step": 4016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.43110884861536e-07,
+      "loss": 2.6221,
+      "step": 4017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.41181019688349e-07,
+      "loss": 2.6313,
+      "step": 4018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.392534630882092e-07,
+      "loss": 2.6311,
+      "step": 4019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.373282161059864e-07,
+      "loss": 2.5594,
+      "step": 4020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.354052797852945e-07,
+      "loss": 2.6598,
+      "step": 4021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.334846551684993e-07,
+      "loss": 2.6294,
+      "step": 4022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.315663432967101e-07,
+      "loss": 2.6625,
+      "step": 4023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.296503452097864e-07,
+      "loss": 2.6748,
+      "step": 4024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.277366619463283e-07,
+      "loss": 2.707,
+      "step": 4025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.25825294543685e-07,
+      "loss": 2.6289,
+      "step": 4026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.239162440379493e-07,
+      "loss": 2.6572,
+      "step": 4027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.220095114639591e-07,
+      "loss": 2.6906,
+      "step": 4028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.20105097855292e-07,
+      "loss": 2.6241,
+      "step": 4029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.182030042442728e-07,
+      "loss": 2.6064,
+      "step": 4030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.16303231661969e-07,
+      "loss": 2.5989,
+      "step": 4031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.144057811381844e-07,
+      "loss": 2.6247,
+      "step": 4032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.125106537014698e-07,
+      "loss": 2.6302,
+      "step": 4033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.106178503791156e-07,
+      "loss": 2.6503,
+      "step": 4034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.0872737219715e-07,
+      "loss": 2.5604,
+      "step": 4035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.06839220180342e-07,
+      "loss": 2.6437,
+      "step": 4036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.049533953522025e-07,
+      "loss": 2.629,
+      "step": 4037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.030698987349749e-07,
+      "loss": 2.6423,
+      "step": 4038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.011887313496463e-07,
+      "loss": 2.7014,
+      "step": 4039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.993098942159393e-07,
+      "loss": 2.6211,
+      "step": 4040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.974333883523104e-07,
+      "loss": 2.576,
+      "step": 4041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.955592147759572e-07,
+      "loss": 2.6754,
+      "step": 4042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.936873745028111e-07,
+      "loss": 2.5755,
+      "step": 4043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.918178685475363e-07,
+      "loss": 2.6758,
+      "step": 4044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.899506979235354e-07,
+      "loss": 2.674,
+      "step": 4045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.880858636429427e-07,
+      "loss": 2.7234,
+      "step": 4046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.862233667166262e-07,
+      "loss": 2.6427,
+      "step": 4047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.843632081541885e-07,
+      "loss": 2.6312,
+      "step": 4048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.825053889639632e-07,
+      "loss": 2.609,
+      "step": 4049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.806499101530162e-07,
+      "loss": 2.6988,
+      "step": 4050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.787967727271455e-07,
+      "loss": 2.5861,
+      "step": 4051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.769459776908777e-07,
+      "loss": 2.6306,
+      "step": 4052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.750975260474718e-07,
+      "loss": 2.6571,
+      "step": 4053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.732514187989164e-07,
+      "loss": 2.649,
+      "step": 4054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.714076569459277e-07,
+      "loss": 2.7251,
+      "step": 4055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.695662414879517e-07,
+      "loss": 2.7049,
+      "step": 4056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.677271734231639e-07,
+      "loss": 2.626,
+      "step": 4057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.658904537484633e-07,
+      "loss": 2.6958,
+      "step": 4058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.640560834594794e-07,
+      "loss": 2.6282,
+      "step": 4059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.622240635505689e-07,
+      "loss": 2.6488,
+      "step": 4060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.603943950148101e-07,
+      "loss": 2.6288,
+      "step": 4061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.585670788440101e-07,
+      "loss": 2.6053,
+      "step": 4062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.567421160287013e-07,
+      "loss": 2.6505,
+      "step": 4063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.549195075581371e-07,
+      "loss": 2.6694,
+      "step": 4064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.530992544202991e-07,
+      "loss": 2.5794,
+      "step": 4065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.512813576018878e-07,
+      "loss": 2.5954,
+      "step": 4066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.494658180883285e-07,
+      "loss": 2.6373,
+      "step": 4067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.476526368637709e-07,
+      "loss": 2.6003,
+      "step": 4068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.458418149110807e-07,
+      "loss": 2.6188,
+      "step": 4069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.440333532118503e-07,
+      "loss": 2.6746,
+      "step": 4070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.422272527463891e-07,
+      "loss": 2.5899,
+      "step": 4071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.404235144937304e-07,
+      "loss": 2.5459,
+      "step": 4072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.386221394316217e-07,
+      "loss": 2.6941,
+      "step": 4073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.368231285365328e-07,
+      "loss": 2.6908,
+      "step": 4074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.350264827836522e-07,
+      "loss": 2.6595,
+      "step": 4075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.332322031468852e-07,
+      "loss": 2.6987,
+      "step": 4076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.314402905988565e-07,
+      "loss": 2.5768,
+      "step": 4077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.29650746110903e-07,
+      "loss": 2.6629,
+      "step": 4078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.278635706530828e-07,
+      "loss": 2.6595,
+      "step": 4079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.260787651941685e-07,
+      "loss": 2.5659,
+      "step": 4080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.242963307016459e-07,
+      "loss": 2.5718,
+      "step": 4081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.225162681417191e-07,
+      "loss": 2.6182,
+      "step": 4082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.207385784793029e-07,
+      "loss": 2.6381,
+      "step": 4083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.189632626780273e-07,
+      "loss": 2.5968,
+      "step": 4084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.171903217002384e-07,
+      "loss": 2.6849,
+      "step": 4085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.154197565069886e-07,
+      "loss": 2.626,
+      "step": 4086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.136515680580479e-07,
+      "loss": 2.6288,
+      "step": 4087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.118857573118969e-07,
+      "loss": 2.6335,
+      "step": 4088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.101223252257255e-07,
+      "loss": 2.6516,
+      "step": 4089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.08361272755435e-07,
+      "loss": 2.6521,
+      "step": 4090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.066026008556386e-07,
+      "loss": 2.6726,
+      "step": 4091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.048463104796554e-07,
+      "loss": 2.7078,
+      "step": 4092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.03092402579517e-07,
+      "loss": 2.6995,
+      "step": 4093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.013408781059627e-07,
+      "loss": 2.6461,
+      "step": 4094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.995917380084382e-07,
+      "loss": 2.6265,
+      "step": 4095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.978449832350974e-07,
+      "loss": 2.6878,
+      "step": 4096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.96100614732803e-07,
+      "loss": 2.6171,
+      "step": 4097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.943586334471224e-07,
+      "loss": 2.6046,
+      "step": 4098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.926190403223303e-07,
+      "loss": 2.6642,
+      "step": 4099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.908818363014041e-07,
+      "loss": 2.6556,
+      "step": 4100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.891470223260282e-07,
+      "loss": 2.6024,
+      "step": 4101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.874145993365932e-07,
+      "loss": 2.7004,
+      "step": 4102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.85684568272189e-07,
+      "loss": 2.6206,
+      "step": 4103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.839569300706127e-07,
+      "loss": 2.7275,
+      "step": 4104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.822316856683641e-07,
+      "loss": 2.6739,
+      "step": 4105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.805088360006422e-07,
+      "loss": 2.6747,
+      "step": 4106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.787883820013518e-07,
+      "loss": 2.6075,
+      "step": 4107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.770703246030979e-07,
+      "loss": 2.6406,
+      "step": 4108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.753546647371838e-07,
+      "loss": 2.7207,
+      "step": 4109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.736414033336163e-07,
+      "loss": 2.651,
+      "step": 4110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.719305413211018e-07,
+      "loss": 2.5959,
+      "step": 4111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.702220796270441e-07,
+      "loss": 2.6231,
+      "step": 4112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.685160191775474e-07,
+      "loss": 2.7453,
+      "step": 4113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.668123608974152e-07,
+      "loss": 2.6291,
+      "step": 4114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.65111105710145e-07,
+      "loss": 2.6014,
+      "step": 4115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.634122545379355e-07,
+      "loss": 2.6237,
+      "step": 4116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.617158083016821e-07,
+      "loss": 2.5745,
+      "step": 4117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.600217679209741e-07,
+      "loss": 2.6115,
+      "step": 4118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.583301343140973e-07,
+      "loss": 2.6349,
+      "step": 4119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.566409083980362e-07,
+      "loss": 2.728,
+      "step": 4120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.549540910884649e-07,
+      "loss": 2.6561,
+      "step": 4121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.532696832997553e-07,
+      "loss": 2.6235,
+      "step": 4122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.515876859449726e-07,
+      "loss": 2.5549,
+      "step": 4123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.499080999358752e-07,
+      "loss": 2.6034,
+      "step": 4124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.482309261829149e-07,
+      "loss": 2.6306,
+      "step": 4125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.46556165595234e-07,
+      "loss": 2.642,
+      "step": 4126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.448838190806683e-07,
+      "loss": 2.6352,
+      "step": 4127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.432138875457465e-07,
+      "loss": 2.7222,
+      "step": 4128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.415463718956838e-07,
+      "loss": 2.642,
+      "step": 4129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3988127303439e-07,
+      "loss": 2.6562,
+      "step": 4130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.382185918644639e-07,
+      "loss": 2.6709,
+      "step": 4131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.365583292871912e-07,
+      "loss": 2.7409,
+      "step": 4132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.349004862025497e-07,
+      "loss": 2.6233,
+      "step": 4133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.332450635092051e-07,
+      "loss": 2.6872,
+      "step": 4134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.31592062104509e-07,
+      "loss": 2.6657,
+      "step": 4135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.299414828845023e-07,
+      "loss": 2.6715,
+      "step": 4136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.282933267439144e-07,
+      "loss": 2.6631,
+      "step": 4137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.266475945761562e-07,
+      "loss": 2.706,
+      "step": 4138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.250042872733313e-07,
+      "loss": 2.6069,
+      "step": 4139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.233634057262221e-07,
+      "loss": 2.6797,
+      "step": 4140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.217249508243011e-07,
+      "loss": 2.6343,
+      "step": 4141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.200889234557238e-07,
+      "loss": 2.7133,
+      "step": 4142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.184553245073287e-07,
+      "loss": 2.6183,
+      "step": 4143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.168241548646391e-07,
+      "loss": 2.6072,
+      "step": 4144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.151954154118621e-07,
+      "loss": 2.5556,
+      "step": 4145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.135691070318855e-07,
+      "loss": 2.7163,
+      "step": 4146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.119452306062822e-07,
+      "loss": 2.6819,
+      "step": 4147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.103237870153027e-07,
+      "loss": 2.6788,
+      "step": 4148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.087047771378817e-07,
+      "loss": 2.6141,
+      "step": 4149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.070882018516348e-07,
+      "loss": 2.6637,
+      "step": 4150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.054740620328575e-07,
+      "loss": 2.694,
+      "step": 4151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.038623585565228e-07,
+      "loss": 2.6348,
+      "step": 4152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.022530922962859e-07,
+      "loss": 2.647,
+      "step": 4153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.006462641244808e-07,
+      "loss": 2.6979,
+      "step": 4154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.990418749121179e-07,
+      "loss": 2.6408,
+      "step": 4155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.974399255288859e-07,
+      "loss": 2.5438,
+      "step": 4156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.958404168431541e-07,
+      "loss": 2.6647,
+      "step": 4157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.94243349721964e-07,
+      "loss": 2.6854,
+      "step": 4158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.926487250310369e-07,
+      "loss": 2.5904,
+      "step": 4159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.910565436347687e-07,
+      "loss": 2.7029,
+      "step": 4160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.89466806396231e-07,
+      "loss": 2.7123,
+      "step": 4161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.878795141771725e-07,
+      "loss": 2.6605,
+      "step": 4162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.862946678380131e-07,
+      "loss": 2.7012,
+      "step": 4163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.847122682378491e-07,
+      "loss": 2.7078,
+      "step": 4164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.831323162344509e-07,
+      "loss": 2.6615,
+      "step": 4165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.815548126842595e-07,
+      "loss": 2.6579,
+      "step": 4166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.799797584423916e-07,
+      "loss": 2.7061,
+      "step": 4167
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.784071543626356e-07,
+      "loss": 2.6099,
+      "step": 4168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.768370012974494e-07,
+      "loss": 2.682,
+      "step": 4169
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7526930009796523e-07,
+      "loss": 2.596,
+      "step": 4170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.737040516139846e-07,
+      "loss": 2.5864,
+      "step": 4171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.721412566939804e-07,
+      "loss": 2.6411,
+      "step": 4172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.705809161850949e-07,
+      "loss": 2.6552,
+      "step": 4173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.690230309331406e-07,
+      "loss": 2.6642,
+      "step": 4174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.674676017825974e-07,
+      "loss": 2.6236,
+      "step": 4175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.659146295766159e-07,
+      "loss": 2.6542,
+      "step": 4176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.643641151570133e-07,
+      "loss": 2.6023,
+      "step": 4177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.62816059364275e-07,
+      "loss": 2.6338,
+      "step": 4178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.612704630375553e-07,
+      "loss": 2.6307,
+      "step": 4179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5972732701467124e-07,
+      "loss": 2.5851,
+      "step": 4180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.58186652132111e-07,
+      "loss": 2.6266,
+      "step": 4181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.566484392250259e-07,
+      "loss": 2.6136,
+      "step": 4182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5511268912723225e-07,
+      "loss": 2.6516,
+      "step": 4183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.535794026712131e-07,
+      "loss": 2.5788,
+      "step": 4184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.520485806881164e-07,
+      "loss": 2.6663,
+      "step": 4185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.505202240077505e-07,
+      "loss": 2.6667,
+      "step": 4186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.489943334585917e-07,
+      "loss": 2.6469,
+      "step": 4187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.474709098677776e-07,
+      "loss": 2.6479,
+      "step": 4188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4594995406110785e-07,
+      "loss": 2.6029,
+      "step": 4189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.444314668630456e-07,
+      "loss": 2.6566,
+      "step": 4190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4291544909671635e-07,
+      "loss": 2.6771,
+      "step": 4191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4140190158390384e-07,
+      "loss": 2.5748,
+      "step": 4192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3989082514505734e-07,
+      "loss": 2.6658,
+      "step": 4193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.383822205992827e-07,
+      "loss": 2.5816,
+      "step": 4194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3687608876434805e-07,
+      "loss": 2.65,
+      "step": 4195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.35372430456682e-07,
+      "loss": 2.5803,
+      "step": 4196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.33871246491368e-07,
+      "loss": 2.6309,
+      "step": 4197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3237253768215305e-07,
+      "loss": 2.6481,
+      "step": 4198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.308763048414405e-07,
+      "loss": 2.6187,
+      "step": 4199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2938254878028996e-07,
+      "loss": 2.6347,
+      "step": 4200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.278912703084215e-07,
+      "loss": 2.6244,
+      "step": 4201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2640247023421066e-07,
+      "loss": 2.5901,
+      "step": 4202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2491614936468864e-07,
+      "loss": 2.6698,
+      "step": 4203
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.234323085055442e-07,
+      "loss": 2.6288,
+      "step": 4204
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.219509484611223e-07,
+      "loss": 2.6222,
+      "step": 4205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2047207003442003e-07,
+      "loss": 2.6254,
+      "step": 4206
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.189956740270917e-07,
+      "loss": 2.6958,
+      "step": 4207
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.175217612394478e-07,
+      "loss": 2.6374,
+      "step": 4208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1605033247044735e-07,
+      "loss": 2.6769,
+      "step": 4209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.14581388517708e-07,
+      "loss": 2.639,
+      "step": 4210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1311493017749894e-07,
+      "loss": 2.6535,
+      "step": 4211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.116509582447392e-07,
+      "loss": 2.6416,
+      "step": 4212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.101894735130052e-07,
+      "loss": 2.6811,
+      "step": 4213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0873047677451985e-07,
+      "loss": 2.6904,
+      "step": 4214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0727396882016047e-07,
+      "loss": 2.7131,
+      "step": 4215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.058199504394561e-07,
+      "loss": 2.7162,
+      "step": 4216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0436842242058236e-07,
+      "loss": 2.6095,
+      "step": 4217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0291938555036935e-07,
+      "loss": 2.5522,
+      "step": 4218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.014728406142937e-07,
+      "loss": 2.7107,
+      "step": 4219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.0002878839648264e-07,
+      "loss": 2.6354,
+      "step": 4220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.985872296797122e-07,
+      "loss": 2.6029,
+      "step": 4221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9714816524540677e-07,
+      "loss": 2.6472,
+      "step": 4222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.957115958736374e-07,
+      "loss": 2.6143,
+      "step": 4223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.942775223431233e-07,
+      "loss": 2.5929,
+      "step": 4224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9284594543123335e-07,
+      "loss": 2.6331,
+      "step": 4225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.914168659139778e-07,
+      "loss": 2.71,
+      "step": 4226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8999028456601694e-07,
+      "loss": 2.6871,
+      "step": 4227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.885662021606579e-07,
+      "loss": 2.7003,
+      "step": 4228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8714461946984916e-07,
+      "loss": 2.6506,
+      "step": 4229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8572553726418715e-07,
+      "loss": 2.6583,
+      "step": 4230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.843089563129132e-07,
+      "loss": 2.6391,
+      "step": 4231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.828948773839103e-07,
+      "loss": 2.5746,
+      "step": 4232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.814833012437069e-07,
+      "loss": 2.6991,
+      "step": 4233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8007422865747436e-07,
+      "loss": 2.5906,
+      "step": 4234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.786676603890271e-07,
+      "loss": 2.6523,
+      "step": 4235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.772635972008226e-07,
+      "loss": 2.6654,
+      "step": 4236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.758620398539586e-07,
+      "loss": 2.6121,
+      "step": 4237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.744629891081758e-07,
+      "loss": 2.6096,
+      "step": 4238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.73066445721858e-07,
+      "loss": 2.6386,
+      "step": 4239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.7167241045202474e-07,
+      "loss": 2.6836,
+      "step": 4240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.702808840543409e-07,
+      "loss": 2.6644,
+      "step": 4241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6889186728310934e-07,
+      "loss": 2.6275,
+      "step": 4242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.675053608912732e-07,
+      "loss": 2.6311,
+      "step": 4243
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.661213656304141e-07,
+      "loss": 2.6489,
+      "step": 4244
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6473988225075195e-07,
+      "loss": 2.6295,
+      "step": 4245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.633609115011466e-07,
+      "loss": 2.6659,
+      "step": 4246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.619844541290951e-07,
+      "loss": 2.59,
+      "step": 4247
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.606105108807323e-07,
+      "loss": 2.6024,
+      "step": 4248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5923908250082907e-07,
+      "loss": 2.5795,
+      "step": 4249
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.578701697327952e-07,
+      "loss": 2.6602,
+      "step": 4250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5650377331867494e-07,
+      "loss": 2.6508,
+      "step": 4251
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.551398939991491e-07,
+      "loss": 2.5966,
+      "step": 4252
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.537785325135351e-07,
+      "loss": 2.6318,
+      "step": 4253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5241968959978335e-07,
+      "loss": 2.6094,
+      "step": 4254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.510633659944801e-07,
+      "loss": 2.609,
+      "step": 4255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4970956243284784e-07,
+      "loss": 2.6149,
+      "step": 4256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.483582796487395e-07,
+      "loss": 2.579,
+      "step": 4257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4700951837464414e-07,
+      "loss": 2.6345,
+      "step": 4258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4566327934168364e-07,
+      "loss": 2.7233,
+      "step": 4259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4431956327961146e-07,
+      "loss": 2.7268,
+      "step": 4260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4297837091681383e-07,
+      "loss": 2.637,
+      "step": 4261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.416397029803115e-07,
+      "loss": 2.6729,
+      "step": 4262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4030356019575164e-07,
+      "loss": 2.6956,
+      "step": 4263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3896994328741774e-07,
+      "loss": 2.6837,
+      "step": 4264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3763885297822153e-07,
+      "loss": 2.6677,
+      "step": 4265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.363102899897053e-07,
+      "loss": 2.6283,
+      "step": 4266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.3498425504204124e-07,
+      "loss": 2.639,
+      "step": 4267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.3366074885403166e-07,
+      "loss": 2.5631,
+      "step": 4268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.3233977214310875e-07,
+      "loss": 2.6708,
+      "step": 4269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.310213256253336e-07,
+      "loss": 2.5692,
+      "step": 4270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.297054100153929e-07,
+      "loss": 2.6286,
+      "step": 4271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2839202602660433e-07,
+      "loss": 2.6371,
+      "step": 4272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2708117437091346e-07,
+      "loss": 2.6667,
+      "step": 4273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.257728557588902e-07,
+      "loss": 2.5971,
+      "step": 4274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.244670708997344e-07,
+      "loss": 2.7,
+      "step": 4275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2316382050127217e-07,
+      "loss": 2.6517,
+      "step": 4276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.218631052699528e-07,
+      "loss": 2.6369,
+      "step": 4277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.205649259108551e-07,
+      "loss": 2.6521,
+      "step": 4278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1926928312768166e-07,
+      "loss": 2.6958,
+      "step": 4279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.179761776227591e-07,
+      "loss": 2.6608,
+      "step": 4280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1668561009704003e-07,
+      "loss": 2.6723,
+      "step": 4281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.15397581250102e-07,
+      "loss": 2.5899,
+      "step": 4282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1411209178014324e-07,
+      "loss": 2.6905,
+      "step": 4283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.128291423839891e-07,
+      "loss": 2.5977,
+      "step": 4284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1154873375708606e-07,
+      "loss": 2.641,
+      "step": 4285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1027086659350334e-07,
+      "loss": 2.6483,
+      "step": 4286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0899554158593303e-07,
+      "loss": 2.6286,
+      "step": 4287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0772275942568986e-07,
+      "loss": 2.6465,
+      "step": 4288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0645252080270806e-07,
+      "loss": 2.6136,
+      "step": 4289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0518482640554515e-07,
+      "loss": 2.6265,
+      "step": 4290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.039196769213787e-07,
+      "loss": 2.6827,
+      "step": 4291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.026570730360079e-07,
+      "loss": 2.5709,
+      "step": 4292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.013970154338486e-07,
+      "loss": 2.6567,
+      "step": 4293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.001395047979405e-07,
+      "loss": 2.6182,
+      "step": 4294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.988845418099401e-07,
+      "loss": 2.6476,
+      "step": 4295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9763212715012535e-07,
+      "loss": 2.6589,
+      "step": 4296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.963822614973888e-07,
+      "loss": 2.7302,
+      "step": 4297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.951349455292446e-07,
+      "loss": 2.702,
+      "step": 4298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.938901799218252e-07,
+      "loss": 2.7167,
+      "step": 4299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.926479653498776e-07,
+      "loss": 2.7049,
+      "step": 4300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.914083024867681e-07,
+      "loss": 2.6505,
+      "step": 4301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9017119200447974e-07,
+      "loss": 2.7217,
+      "step": 4302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8893663457361056e-07,
+      "loss": 2.6952,
+      "step": 4303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.877046308633763e-07,
+      "loss": 2.6816,
+      "step": 4304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.864751815416084e-07,
+      "loss": 2.6828,
+      "step": 4305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8524828727475194e-07,
+      "loss": 2.6701,
+      "step": 4306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8402394872786896e-07,
+      "loss": 2.6651,
+      "step": 4307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.828021665646341e-07,
+      "loss": 2.6446,
+      "step": 4308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.815829414473381e-07,
+      "loss": 2.6866,
+      "step": 4309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8036627403688564e-07,
+      "loss": 2.6452,
+      "step": 4310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.791521649927925e-07,
+      "loss": 2.6822,
+      "step": 4311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.779406149731906e-07,
+      "loss": 2.6008,
+      "step": 4312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.767316246348234e-07,
+      "loss": 2.647,
+      "step": 4313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.755251946330467e-07,
+      "loss": 2.671,
+      "step": 4314
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.74321325621828e-07,
+      "loss": 2.6641,
+      "step": 4315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7312001825374747e-07,
+      "loss": 2.6177,
+      "step": 4316
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7192127317999696e-07,
+      "loss": 2.6428,
+      "step": 4317
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7072509105037837e-07,
+      "loss": 2.6535,
+      "step": 4318
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6953147251330583e-07,
+      "loss": 2.6949,
+      "step": 4319
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6834041821580114e-07,
+      "loss": 2.6725,
+      "step": 4320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6715192880349795e-07,
+      "loss": 2.6113,
+      "step": 4321
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6596600492064085e-07,
+      "loss": 2.6449,
+      "step": 4322
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.647826472100806e-07,
+      "loss": 2.6857,
+      "step": 4323
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.636018563132786e-07,
+      "loss": 2.6263,
+      "step": 4324
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6242363287030617e-07,
+      "loss": 2.6337,
+      "step": 4325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6124797751983953e-07,
+      "loss": 2.6961,
+      "step": 4326
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.600748908991674e-07,
+      "loss": 2.6535,
+      "step": 4327
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.589043736441804e-07,
+      "loss": 2.65,
+      "step": 4328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5773642638938166e-07,
+      "loss": 2.6328,
+      "step": 4329
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.565710497678786e-07,
+      "loss": 2.6753,
+      "step": 4330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.554082444113848e-07,
+      "loss": 2.6551,
+      "step": 4331
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5424801095022154e-07,
+      "loss": 2.6504,
+      "step": 4332
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.530903500133153e-07,
+      "loss": 2.5852,
+      "step": 4333
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5193526222819733e-07,
+      "loss": 2.6636,
+      "step": 4334
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.507827482210051e-07,
+      "loss": 2.6715,
+      "step": 4335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.496328086164812e-07,
+      "loss": 2.6208,
+      "step": 4336
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.484854440379708e-07,
+      "loss": 2.6579,
+      "step": 4337
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.473406551074248e-07,
+      "loss": 2.6694,
+      "step": 4338
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.461984424453978e-07,
+      "loss": 2.5841,
+      "step": 4339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.450588066710485e-07,
+      "loss": 2.7186,
+      "step": 4340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.439217484021367e-07,
+      "loss": 2.6409,
+      "step": 4341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4278726825502696e-07,
+      "loss": 2.6504,
+      "step": 4342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.416553668446847e-07,
+      "loss": 2.5554,
+      "step": 4343
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4052604478467967e-07,
+      "loss": 2.6588,
+      "step": 4344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3939930268718216e-07,
+      "loss": 2.6393,
+      "step": 4345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3827514116296246e-07,
+      "loss": 2.6986,
+      "step": 4346
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.371535608213954e-07,
+      "loss": 2.6887,
+      "step": 4347
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3603456227045297e-07,
+      "loss": 2.7137,
+      "step": 4348
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3491814611671049e-07,
+      "loss": 2.717,
+      "step": 4349
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3380431296534224e-07,
+      "loss": 2.6969,
+      "step": 4350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3269306342012188e-07,
+      "loss": 2.673,
+      "step": 4351
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.315843980834237e-07,
+      "loss": 2.6752,
+      "step": 4352
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3047831755622085e-07,
+      "loss": 2.6429,
+      "step": 4353
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.293748224380843e-07,
+      "loss": 2.6342,
+      "step": 4354
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2827391332718495e-07,
+      "loss": 2.6966,
+      "step": 4355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.271755908202916e-07,
+      "loss": 2.5888,
+      "step": 4356
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2607985551276969e-07,
+      "loss": 2.5794,
+      "step": 4357
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2498670799858347e-07,
+      "loss": 2.6297,
+      "step": 4358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2389614887029564e-07,
+      "loss": 2.5724,
+      "step": 4359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2280817871906168e-07,
+      "loss": 2.6199,
+      "step": 4360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2172279813463815e-07,
+      "loss": 2.7195,
+      "step": 4361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2064000770537663e-07,
+      "loss": 2.6343,
+      "step": 4362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1955980801822207e-07,
+      "loss": 2.648,
+      "step": 4363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1848219965871774e-07,
+      "loss": 2.6706,
+      "step": 4364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1740718321100195e-07,
+      "loss": 2.662,
+      "step": 4365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1633475925780746e-07,
+      "loss": 2.6595,
+      "step": 4366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1526492838046308e-07,
+      "loss": 2.6577,
+      "step": 4367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1419769115888833e-07,
+      "loss": 2.5576,
+      "step": 4368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1313304817160096e-07,
+      "loss": 2.5673,
+      "step": 4369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1207099999571102e-07,
+      "loss": 2.6389,
+      "step": 4370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1101154720692018e-07,
+      "loss": 2.644,
+      "step": 4371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.099546903795252e-07,
+      "loss": 2.6271,
+      "step": 4372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0890043008641726e-07,
+      "loss": 2.6223,
+      "step": 4373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0784876689907528e-07,
+      "loss": 2.6358,
+      "step": 4374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0679970138757387e-07,
+      "loss": 2.6096,
+      "step": 4375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0575323412058036e-07,
+      "loss": 2.5712,
+      "step": 4376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0470936566534982e-07,
+      "loss": 2.6439,
+      "step": 4377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.036680965877319e-07,
+      "loss": 2.7107,
+      "step": 4378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0262942745216675e-07,
+      "loss": 2.6368,
+      "step": 4379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.015933588216834e-07,
+      "loss": 2.6632,
+      "step": 4380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0055989125790265e-07,
+      "loss": 2.6747,
+      "step": 4381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9952902532103525e-07,
+      "loss": 2.6731,
+      "step": 4382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.985007615698814e-07,
+      "loss": 2.6254,
+      "step": 4383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.974751005618314e-07,
+      "loss": 2.5954,
+      "step": 4384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.964520428528627e-07,
+      "loss": 2.6102,
+      "step": 4385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9543158899754389e-07,
+      "loss": 2.6206,
+      "step": 4386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9441373954903142e-07,
+      "loss": 2.6167,
+      "step": 4387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9339849505907006e-07,
+      "loss": 2.62,
+      "step": 4388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9238585607799121e-07,
+      "loss": 2.6131,
+      "step": 4389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9137582315471525e-07,
+      "loss": 2.6473,
+      "step": 4390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.903683968367498e-07,
+      "loss": 2.642,
+      "step": 4391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.893635776701891e-07,
+      "loss": 2.7174,
+      "step": 4392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8836136619971468e-07,
+      "loss": 2.6709,
+      "step": 4393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8736176296859365e-07,
+      "loss": 2.6626,
+      "step": 4394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8636476851867924e-07,
+      "loss": 2.7001,
+      "step": 4395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8537038339041303e-07,
+      "loss": 2.6842,
+      "step": 4396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8437860812281826e-07,
+      "loss": 2.5388,
+      "step": 4397
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8338944325350605e-07,
+      "loss": 2.682,
+      "step": 4398
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8240288931867245e-07,
+      "loss": 2.6417,
+      "step": 4399
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8141894685309636e-07,
+      "loss": 2.6708,
+      "step": 4400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8043761639014335e-07,
+      "loss": 2.6089,
+      "step": 4401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7945889846176178e-07,
+      "loss": 2.6047,
+      "step": 4402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7848279359848343e-07,
+      "loss": 2.6527,
+      "step": 4403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7750930232942608e-07,
+      "loss": 2.5918,
+      "step": 4404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7653842518228715e-07,
+      "loss": 2.6136,
+      "step": 4405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7557016268334948e-07,
+      "loss": 2.6171,
+      "step": 4406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.746045153574788e-07,
+      "loss": 2.7161,
+      "step": 4407
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.736414837281214e-07,
+      "loss": 2.6197,
+      "step": 4408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.726810683173069e-07,
+      "loss": 2.6638,
+      "step": 4409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7172326964564777e-07,
+      "loss": 2.6972,
+      "step": 4410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7076808823233482e-07,
+      "loss": 2.6404,
+      "step": 4411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6981552459514328e-07,
+      "loss": 2.5847,
+      "step": 4412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6886557925042845e-07,
+      "loss": 2.5849,
+      "step": 4413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.679182527131251e-07,
+      "loss": 2.6095,
+      "step": 4414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6697354549675082e-07,
+      "loss": 2.6298,
+      "step": 4415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6603145811340094e-07,
+      "loss": 2.6742,
+      "step": 4416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.650919910737514e-07,
+      "loss": 2.6703,
+      "step": 4417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6415514488705874e-07,
+      "loss": 2.6966,
+      "step": 4418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6322092006115843e-07,
+      "loss": 2.6628,
+      "step": 4419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6228931710246366e-07,
+      "loss": 2.6681,
+      "step": 4420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6136033651596828e-07,
+      "loss": 2.6694,
+      "step": 4421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6043397880524224e-07,
+      "loss": 2.6259,
+      "step": 4422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5951024447243614e-07,
+      "loss": 2.593,
+      "step": 4423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5858913401827881e-07,
+      "loss": 2.6884,
+      "step": 4424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.576706479420731e-07,
+      "loss": 2.6749,
+      "step": 4425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5675478674170297e-07,
+      "loss": 2.6519,
+      "step": 4426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5584155091362907e-07,
+      "loss": 2.6452,
+      "step": 4427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5493094095288596e-07,
+      "loss": 2.6719,
+      "step": 4428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.540229573530888e-07,
+      "loss": 2.6502,
+      "step": 4429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5311760060642666e-07,
+      "loss": 2.5696,
+      "step": 4430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5221487120366417e-07,
+      "loss": 2.7013,
+      "step": 4431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5131476963414438e-07,
+      "loss": 2.5874,
+      "step": 4432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5041729638578362e-07,
+      "loss": 2.7409,
+      "step": 4433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4952245194507386e-07,
+      "loss": 2.6452,
+      "step": 4434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.486302367970832e-07,
+      "loss": 2.6097,
+      "step": 4435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4774065142545314e-07,
+      "loss": 2.6065,
+      "step": 4436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4685369631239955e-07,
+      "loss": 2.7001,
+      "step": 4437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4596937193871452e-07,
+      "loss": 2.6408,
+      "step": 4438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4508767878376129e-07,
+      "loss": 2.6882,
+      "step": 4439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.442086173254792e-07,
+      "loss": 2.6474,
+      "step": 4440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.433321880403804e-07,
+      "loss": 2.633,
+      "step": 4441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4245839140354935e-07,
+      "loss": 2.6962,
+      "step": 4442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4158722788864321e-07,
+      "loss": 2.6498,
+      "step": 4443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4071869796789427e-07,
+      "loss": 2.6638,
+      "step": 4444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3985280211210372e-07,
+      "loss": 2.6981,
+      "step": 4445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.389895407906483e-07,
+      "loss": 2.6447,
+      "step": 4446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3812891447147425e-07,
+      "loss": 2.683,
+      "step": 4447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3727092362110006e-07,
+      "loss": 2.6165,
+      "step": 4448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3641556870461591e-07,
+      "loss": 2.6831,
+      "step": 4449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3556285018568428e-07,
+      "loss": 2.6065,
+      "step": 4450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3471276852653593e-07,
+      "loss": 2.6637,
+      "step": 4451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3386532418797337e-07,
+      "loss": 2.6309,
+      "step": 4452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.330205176293714e-07,
+      "loss": 2.5735,
+      "step": 4453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3217834930867202e-07,
+      "loss": 2.6687,
+      "step": 4454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3133881968238836e-07,
+      "loss": 2.6913,
+      "step": 4455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3050192920560422e-07,
+      "loss": 2.6235,
+      "step": 4456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2966767833197113e-07,
+      "loss": 2.6435,
+      "step": 4457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.288360675137107e-07,
+      "loss": 2.5979,
+      "step": 4458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2800709720161287e-07,
+      "loss": 2.6257,
+      "step": 4459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2718076784503708e-07,
+      "loss": 2.6292,
+      "step": 4460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.263570798919106e-07,
+      "loss": 2.685,
+      "step": 4461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2553603378872903e-07,
+      "loss": 2.63,
+      "step": 4462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.247176299805558e-07,
+      "loss": 2.6498,
+      "step": 4463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2390186891102217e-07,
+      "loss": 2.6396,
+      "step": 4464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2308875102232664e-07,
+      "loss": 2.631,
+      "step": 4465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2227827675523551e-07,
+      "loss": 2.6362,
+      "step": 4466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2147044654908124e-07,
+      "loss": 2.6315,
+      "step": 4467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2066526084176355e-07,
+      "loss": 2.6724,
+      "step": 4468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1986272006974775e-07,
+      "loss": 2.6413,
+      "step": 4469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1906282466806806e-07,
+      "loss": 2.6864,
+      "step": 4470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.18265575070321e-07,
+      "loss": 2.6317,
+      "step": 4471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1747097170867084e-07,
+      "loss": 2.6293,
+      "step": 4472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1667901501384914e-07,
+      "loss": 2.6882,
+      "step": 4473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1588970541514865e-07,
+      "loss": 2.5115,
+      "step": 4474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1510304334043099e-07,
+      "loss": 2.6582,
+      "step": 4475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.143190292161206e-07,
+      "loss": 2.6678,
+      "step": 4476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1353766346720696e-07,
+      "loss": 2.6424,
+      "step": 4477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1275894651724517e-07,
+      "loss": 2.6674,
+      "step": 4478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1198287878835201e-07,
+      "loss": 2.7087,
+      "step": 4479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1120946070120986e-07,
+      "loss": 2.7515,
+      "step": 4480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1043869267506557e-07,
+      "loss": 2.5834,
+      "step": 4481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0967057512772717e-07,
+      "loss": 2.6214,
+      "step": 4482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0890510847556768e-07,
+      "loss": 2.689,
+      "step": 4483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0814229313352354e-07,
+      "loss": 2.6123,
+      "step": 4484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0738212951509231e-07,
+      "loss": 2.6912,
+      "step": 4485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0662461803233492e-07,
+      "loss": 2.6271,
+      "step": 4486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0586975909587516e-07,
+      "loss": 2.6512,
+      "step": 4487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0511755311489791e-07,
+      "loss": 2.6724,
+      "step": 4488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0436800049715146e-07,
+      "loss": 2.586,
+      "step": 4489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0362110164894523e-07,
+      "loss": 2.663,
+      "step": 4490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0287685697514871e-07,
+      "loss": 2.6653,
+      "step": 4491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0213526687919362e-07,
+      "loss": 2.6738,
+      "step": 4492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0139633176307396e-07,
+      "loss": 2.7399,
+      "step": 4493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0066005202734264e-07,
+      "loss": 2.6713,
+      "step": 4494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.992642807111486e-08,
+      "loss": 2.7039,
+      "step": 4495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.919546029206361e-08,
+      "loss": 2.6227,
+      "step": 4496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.84671490864253e-08,
+      "loss": 2.6547,
+      "step": 4497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.774149484899409e-08,
+      "loss": 2.6644,
+      "step": 4498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.701849797312479e-08,
+      "loss": 2.6163,
+      "step": 4499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.629815885073057e-08,
+      "loss": 2.6959,
+      "step": 4500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.55804778722863e-08,
+      "loss": 2.711,
+      "step": 4501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.486545542682357e-08,
+      "loss": 2.6638,
+      "step": 4502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.415309190193345e-08,
+      "loss": 2.6323,
+      "step": 4503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.34433876837676e-08,
+      "loss": 2.6097,
+      "step": 4504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.273634315703217e-08,
+      "loss": 2.7195,
+      "step": 4505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.20319587049956e-08,
+      "loss": 2.6595,
+      "step": 4506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.133023470948299e-08,
+      "loss": 2.6554,
+      "step": 4507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.06311715508762e-08,
+      "loss": 2.6497,
+      "step": 4508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.993476960811542e-08,
+      "loss": 2.6231,
+      "step": 4509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.924102925869871e-08,
+      "loss": 2.6198,
+      "step": 4510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.85499508786819e-08,
+      "loss": 2.5899,
+      "step": 4511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.78615348426759e-08,
+      "loss": 2.6252,
+      "step": 4512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.717578152384997e-08,
+      "loss": 2.6838,
+      "step": 4513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.649269129393012e-08,
+      "loss": 2.6379,
+      "step": 4514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.581226452319791e-08,
+      "loss": 2.681,
+      "step": 4515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.513450158049109e-08,
+      "loss": 2.5795,
+      "step": 4516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.445940283320408e-08,
+      "loss": 2.692,
+      "step": 4517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.378696864728753e-08,
+      "loss": 2.5987,
+      "step": 4518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.311719938724594e-08,
+      "loss": 2.6731,
+      "step": 4519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.245009541614113e-08,
+      "loss": 2.7198,
+      "step": 4520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.178565709558939e-08,
+      "loss": 2.6179,
+      "step": 4521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.11238847857615e-08,
+      "loss": 2.6558,
+      "step": 4522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.046477884538384e-08,
+      "loss": 2.5687,
+      "step": 4523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.980833963173784e-08,
+      "loss": 2.5371,
+      "step": 4524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.915456750065775e-08,
+      "loss": 2.6657,
+      "step": 4525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.850346280653287e-08,
+      "loss": 2.6487,
+      "step": 4526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.785502590230864e-08,
+      "loss": 2.6551,
+      "step": 4527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.720925713948002e-08,
+      "loss": 2.6047,
+      "step": 4528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.656615686809976e-08,
+      "loss": 2.6425,
+      "step": 4529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.592572543677235e-08,
+      "loss": 2.724,
+      "step": 4530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.528796319265508e-08,
+      "loss": 2.5822,
+      "step": 4531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.46528704814592e-08,
+      "loss": 2.6238,
+      "step": 4532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.402044764744932e-08,
+      "loss": 2.6647,
+      "step": 4533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.339069503344121e-08,
+      "loss": 2.6948,
+      "step": 4534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.27636129808046e-08,
+      "loss": 2.6299,
+      "step": 4535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.213920182946087e-08,
+      "loss": 2.6571,
+      "step": 4536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.151746191788434e-08,
+      "loss": 2.6234,
+      "step": 4537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.089839358310091e-08,
+      "loss": 2.576,
+      "step": 4538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.028199716068773e-08,
+      "loss": 2.6302,
+      "step": 4539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.966827298477363e-08,
+      "loss": 2.6418,
+      "step": 4540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.905722138804083e-08,
+      "loss": 2.671,
+      "step": 4541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.844884270171992e-08,
+      "loss": 2.5719,
+      "step": 4542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.784313725559432e-08,
+      "loss": 2.6014,
+      "step": 4543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.724010537799918e-08,
+      "loss": 2.7001,
+      "step": 4544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.663974739581802e-08,
+      "loss": 2.6068,
+      "step": 4545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.604206363448662e-08,
+      "loss": 2.5849,
+      "step": 4546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.544705441799082e-08,
+      "loss": 2.5743,
+      "step": 4547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.48547200688665e-08,
+      "loss": 2.6251,
+      "step": 4548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.426506090819961e-08,
+      "loss": 2.6442,
+      "step": 4549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.367807725562614e-08,
+      "loss": 2.7467,
+      "step": 4550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.309376942933154e-08,
+      "loss": 2.5715,
+      "step": 4551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.251213774605081e-08,
+      "loss": 2.6477,
+      "step": 4552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.193318252106839e-08,
+      "loss": 2.6005,
+      "step": 4553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.13569040682177e-08,
+      "loss": 2.6719,
+      "step": 4554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.07833026998822e-08,
+      "loss": 2.6082,
+      "step": 4555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.021237872699204e-08,
+      "loss": 2.6723,
+      "step": 4556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.9644132459028e-08,
+      "loss": 2.6054,
+      "step": 4557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.9078564204018694e-08,
+      "loss": 2.695,
+      "step": 4558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8515674268541103e-08,
+      "loss": 2.6717,
+      "step": 4559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.7955462957719474e-08,
+      "loss": 2.6451,
+      "step": 4560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.7397930575228134e-08,
+      "loss": 2.6532,
+      "step": 4561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.6843077423286984e-08,
+      "loss": 2.6957,
+      "step": 4562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.6290903802665444e-08,
+      "loss": 2.5876,
+      "step": 4563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.5741410012679634e-08,
+      "loss": 2.6477,
+      "step": 4564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.5194596351192396e-08,
+      "loss": 2.676,
+      "step": 4565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.465046311461442e-08,
+      "loss": 2.6429,
+      "step": 4566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.410901059790419e-08,
+      "loss": 2.5823,
+      "step": 4567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.357023909456527e-08,
+      "loss": 2.5998,
+      "step": 4568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.303414889664904e-08,
+      "loss": 2.6805,
+      "step": 4569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.250074029475416e-08,
+      "loss": 2.6551,
+      "step": 4570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.197001357802378e-08,
+      "loss": 2.6398,
+      "step": 4571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1441969034148886e-08,
+      "loss": 2.7272,
+      "step": 4572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.091660694936551e-08,
+      "loss": 2.5933,
+      "step": 4573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0393927608456404e-08,
+      "loss": 2.664,
+      "step": 4574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.987393129474993e-08,
+      "loss": 2.6358,
+      "step": 4575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9356618290119505e-08,
+      "loss": 2.6105,
+      "step": 4576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8841988874984145e-08,
+      "loss": 2.5598,
+      "step": 4577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8330043328309594e-08,
+      "loss": 2.5905,
+      "step": 4578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7820781927604423e-08,
+      "loss": 2.7143,
+      "step": 4579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7314204948923356e-08,
+      "loss": 2.6632,
+      "step": 4580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6810312666867306e-08,
+      "loss": 2.6938,
+      "step": 4581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.630910535457889e-08,
+      "loss": 2.6133,
+      "step": 4582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.581058328374799e-08,
+      "loss": 2.6675,
+      "step": 4583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.531474672460734e-08,
+      "loss": 2.6001,
+      "step": 4584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4821595945934714e-08,
+      "loss": 2.632,
+      "step": 4585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4331131215052395e-08,
+      "loss": 2.5718,
+      "step": 4586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.384335279782492e-08,
+      "loss": 2.6566,
+      "step": 4587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.335826095866247e-08,
+      "loss": 2.681,
+      "step": 4588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.287585596051858e-08,
+      "loss": 2.6247,
+      "step": 4589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.239613806488907e-08,
+      "loss": 2.7065,
+      "step": 4590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.191910753181427e-08,
+      "loss": 2.6542,
+      "step": 4591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.14447646198779e-08,
+      "loss": 2.6481,
+      "step": 4592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.097310958620593e-08,
+      "loss": 2.6543,
+      "step": 4593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.050414268646829e-08,
+      "loss": 2.6098,
+      "step": 4594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.003786417487776e-08,
+      "loss": 2.5925,
+      "step": 4595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.957427430418881e-08,
+      "loss": 2.6041,
+      "step": 4596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.911337332569876e-08,
+      "loss": 2.6421,
+      "step": 4597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.865516148924831e-08,
+      "loss": 2.6121,
+      "step": 4598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8199639043219325e-08,
+      "loss": 2.6115,
+      "step": 4599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7746806234536526e-08,
+      "loss": 2.6607,
+      "step": 4600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.729666330866688e-08,
+      "loss": 2.634,
+      "step": 4601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6849210509617984e-08,
+      "loss": 2.6465,
+      "step": 4602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6404448079940816e-08,
+      "loss": 2.6517,
+      "step": 4603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.596237626072696e-08,
+      "loss": 2.6572,
+      "step": 4604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.552299529160974e-08,
+      "loss": 2.676,
+      "step": 4605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5086305410763634e-08,
+      "loss": 2.5617,
+      "step": 4606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4652306854905394e-08,
+      "loss": 2.5868,
+      "step": 4607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.422099985929128e-08,
+      "loss": 2.6857,
+      "step": 4608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3792384657719276e-08,
+      "loss": 2.6612,
+      "step": 4609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.336646148252909e-08,
+      "loss": 2.6397,
+      "step": 4610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.294323056460047e-08,
+      "loss": 2.6382,
+      "step": 4611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2522692133352686e-08,
+      "loss": 2.639,
+      "step": 4612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.210484641674727e-08,
+      "loss": 2.6532,
+      "step": 4613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.168969364128527e-08,
+      "loss": 2.5288,
+      "step": 4614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.127723403200833e-08,
+      "loss": 2.5864,
+      "step": 4615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.086746781249705e-08,
+      "loss": 2.7139,
+      "step": 4616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.046039520487376e-08,
+      "loss": 2.6474,
+      "step": 4617
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.005601642979972e-08,
+      "loss": 2.6023,
+      "step": 4618
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9654331706476245e-08,
+      "loss": 2.5719,
+      "step": 4619
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9255341252643067e-08,
+      "loss": 2.6286,
+      "step": 4620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.885904528458161e-08,
+      "loss": 2.6384,
+      "step": 4621
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8465444017111154e-08,
+      "loss": 2.6559,
+      "step": 4622
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8074537663590472e-08,
+      "loss": 2.5737,
+      "step": 4623
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7686326435917842e-08,
+      "loss": 2.6972,
+      "step": 4624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.730081054453104e-08,
+      "loss": 2.7099,
+      "step": 4625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.691799019840513e-08,
+      "loss": 2.5969,
+      "step": 4626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6537865605056333e-08,
+      "loss": 2.6444,
+      "step": 4627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6160436970537052e-08,
+      "loss": 2.6721,
+      "step": 4628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5785704499440846e-08,
+      "loss": 2.6503,
+      "step": 4629
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.541366839489745e-08,
+      "loss": 2.639,
+      "step": 4630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5044328858576105e-08,
+      "loss": 2.6037,
+      "step": 4631
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4677686090685548e-08,
+      "loss": 2.639,
+      "step": 4632
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4313740289970134e-08,
+      "loss": 2.6389,
+      "step": 4633
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3952491653713717e-08,
+      "loss": 2.6399,
+      "step": 4634
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3593940377738545e-08,
+      "loss": 2.6594,
+      "step": 4635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3238086656403034e-08,
+      "loss": 2.7032,
+      "step": 4636
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2884930682605112e-08,
+      "loss": 2.6627,
+      "step": 4637
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.253447264777997e-08,
+      "loss": 2.5814,
+      "step": 4638
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2186712741899542e-08,
+      "loss": 2.6604,
+      "step": 4639
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1841651153473588e-08,
+      "loss": 2.7336,
+      "step": 4640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1499288069549153e-08,
+      "loss": 2.6265,
+      "step": 4641
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1159623675711115e-08,
+      "loss": 2.5564,
+      "step": 4642
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.082265815607998e-08,
+      "loss": 2.6012,
+      "step": 4643
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0488391693315178e-08,
+      "loss": 2.7168,
+      "step": 4644
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.015682446861178e-08,
+      "loss": 2.7039,
+      "step": 4645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.982795666170212e-08,
+      "loss": 2.597,
+      "step": 4646
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9501788450855263e-08,
+      "loss": 2.6553,
+      "step": 4647
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.917832001287645e-08,
+      "loss": 2.6626,
+      "step": 4648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8857551523108753e-08,
+      "loss": 2.712,
+      "step": 4649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8539483155429196e-08,
+      "loss": 2.6722,
+      "step": 4650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8224115082254303e-08,
+      "loss": 2.6844,
+      "step": 4651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.791144747453455e-08,
+      "loss": 2.6472,
+      "step": 4652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.760148050175714e-08,
+      "loss": 2.6198,
+      "step": 4653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7294214331946003e-08,
+      "loss": 2.6177,
+      "step": 4654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.698964913166068e-08,
+      "loss": 2.6275,
+      "step": 4655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6687785065995776e-08,
+      "loss": 2.641,
+      "step": 4656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.638862229858207e-08,
+      "loss": 2.6676,
+      "step": 4657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6092160991587614e-08,
+      "loss": 2.6174,
+      "step": 4658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5798401305714416e-08,
+      "loss": 2.5873,
+      "step": 4659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5507343400200103e-08,
+      "loss": 2.6101,
+      "step": 4660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5218987432818465e-08,
+      "loss": 2.6255,
+      "step": 4661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4933333559877805e-08,
+      "loss": 2.6511,
+      "step": 4662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4650381936222591e-08,
+      "loss": 2.6708,
+      "step": 4663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4370132715232354e-08,
+      "loss": 2.623,
+      "step": 4664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4092586048820578e-08,
+      "loss": 2.6331,
+      "step": 4665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.381774208743747e-08,
+      "loss": 2.6683,
+      "step": 4666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3545600980067187e-08,
+      "loss": 2.6612,
+      "step": 4667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3276162874228394e-08,
+      "loss": 2.6765,
+      "step": 4668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3009427915975925e-08,
+      "loss": 2.6821,
+      "step": 4669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2745396249898013e-08,
+      "loss": 2.6768,
+      "step": 4670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2484068019117945e-08,
+      "loss": 2.6812,
+      "step": 4671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2225443365294077e-08,
+      "loss": 2.5864,
+      "step": 4672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1969522428618152e-08,
+      "loss": 2.6191,
+      "step": 4673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.171630534781698e-08,
+      "loss": 2.5663,
+      "step": 4674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.146579226015243e-08,
+      "loss": 2.73,
+      "step": 4675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1217983301418656e-08,
+      "loss": 2.6547,
+      "step": 4676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0972878605946535e-08,
+      "loss": 2.699,
+      "step": 4677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0730478306598679e-08,
+      "loss": 2.6557,
+      "step": 4678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0490782534772758e-08,
+      "loss": 2.6538,
+      "step": 4679
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0253791420400949e-08,
+      "loss": 2.6448,
+      "step": 4680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0019505091948268e-08,
+      "loss": 2.661,
+      "step": 4681
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.787923676414235e-09,
+      "loss": 2.6163,
+      "step": 4682
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.55904729933177e-09,
+      "loss": 2.6079,
+      "step": 4683
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.332876084768295e-09,
+      "loss": 2.6661,
+      "step": 4684
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.109410155322962e-09,
+      "loss": 2.6174,
+      "step": 4685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.888649632130542e-09,
+      "loss": 2.7084,
+      "step": 4686
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.670594634858642e-09,
+      "loss": 2.5876,
+      "step": 4687
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.455245281707158e-09,
+      "loss": 2.6565,
+      "step": 4688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.242601689411044e-09,
+      "loss": 2.6599,
+      "step": 4689
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.032663973237543e-09,
+      "loss": 2.6538,
+      "step": 4690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.825432246987286e-09,
+      "loss": 2.6311,
+      "step": 4691
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.620906622994307e-09,
+      "loss": 2.5999,
+      "step": 4692
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.41908721212492e-09,
+      "loss": 2.7005,
+      "step": 4693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.219974123779949e-09,
+      "loss": 2.6337,
+      "step": 4694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.0235674658925e-09,
+      "loss": 2.6326,
+      "step": 4695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.82986734492741e-09,
+      "loss": 2.65,
+      "step": 4696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.638873865884576e-09,
+      "loss": 2.636,
+      "step": 4697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.450587132295072e-09,
+      "loss": 2.641,
+      "step": 4698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.265007246223365e-09,
+      "loss": 2.6929,
+      "step": 4699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.082134308266763e-09,
+      "loss": 2.7121,
+      "step": 4700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.90196841755486e-09,
+      "loss": 2.6686,
+      "step": 4701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.724509671749534e-09,
+      "loss": 2.6996,
+      "step": 4702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.54975816704606e-09,
+      "loss": 2.5781,
+      "step": 4703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.377713998171996e-09,
+      "loss": 2.6589,
+      "step": 4704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.208377258386632e-09,
+      "loss": 2.6224,
+      "step": 4705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.041748039482097e-09,
+      "loss": 2.6722,
+      "step": 4706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.877826431783916e-09,
+      "loss": 2.5234,
+      "step": 4707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.716612524147124e-09,
+      "loss": 2.5935,
+      "step": 4708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.55810640396237e-09,
+      "loss": 2.6116,
+      "step": 4709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.40230815715037e-09,
+      "loss": 2.6776,
+      "step": 4710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.249217868164124e-09,
+      "loss": 2.6236,
+      "step": 4711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.098835619989472e-09,
+      "loss": 2.6569,
+      "step": 4712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.951161494143984e-09,
+      "loss": 2.6456,
+      "step": 4713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.806195570676962e-09,
+      "loss": 2.718,
+      "step": 4714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6639379281699917e-09,
+      "loss": 2.6144,
+      "step": 4715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.524388643736387e-09,
+      "loss": 2.6475,
+      "step": 4716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3875477930217506e-09,
+      "loss": 2.6329,
+      "step": 4717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.253415450203412e-09,
+      "loss": 2.637,
+      "step": 4718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.121991687989323e-09,
+      "loss": 2.7031,
+      "step": 4719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9932765776213845e-09,
+      "loss": 2.6006,
+      "step": 4720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8672701888715627e-09,
+      "loss": 2.6368,
+      "step": 4721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7439725900441083e-09,
+      "loss": 2.6753,
+      "step": 4722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6233838479744478e-09,
+      "loss": 2.6827,
+      "step": 4723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5055040280302923e-09,
+      "loss": 2.6399,
+      "step": 4724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.390333194111083e-09,
+      "loss": 2.5776,
+      "step": 4725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.277871408646326e-09,
+      "loss": 2.6615,
+      "step": 4726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1681187325983675e-09,
+      "loss": 2.6463,
+      "step": 4727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.061075225461284e-09,
+      "loss": 2.6536,
+      "step": 4728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.956740945259772e-09,
+      "loss": 2.6454,
+      "step": 4729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8551159485497016e-09,
+      "loss": 2.6543,
+      "step": 4730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7562002904192299e-09,
+      "loss": 2.7654,
+      "step": 4731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6599940244871327e-09,
+      "loss": 2.6706,
+      "step": 4732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566497202904471e-09,
+      "loss": 2.681,
+      "step": 4733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4757098763523715e-09,
+      "loss": 2.7117,
+      "step": 4734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3876320940442445e-09,
+      "loss": 2.6517,
+      "step": 4735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3022639037241214e-09,
+      "loss": 2.6563,
+      "step": 4736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.219605351667208e-09,
+      "loss": 2.6867,
+      "step": 4737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1396564826804401e-09,
+      "loss": 2.6095,
+      "step": 4738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0624173401013738e-09,
+      "loss": 2.7135,
+      "step": 4739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.878879657992946e-10,
+      "loss": 2.6932,
+      "step": 4740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.160684001741082e-10,
+      "loss": 2.5852,
+      "step": 4741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.469586821563403e-10,
+      "loss": 2.6351,
+      "step": 4742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.805588492093564e-10,
+      "loss": 2.5954,
+      "step": 4743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.168689373254767e-10,
+      "loss": 2.6846,
+      "step": 4744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.558889810293067e-10,
+      "loss": 2.6296,
+      "step": 4745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.976190133760718e-10,
+      "loss": 2.6401,
+      "step": 4746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.420590659527269e-10,
+      "loss": 2.603,
+      "step": 4747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.892091688757372e-10,
+      "loss": 2.5919,
+      "step": 4748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3906935079440773e-10,
+      "loss": 2.723,
+      "step": 4749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.916396388869981e-10,
+      "loss": 2.6647,
+      "step": 4750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.469200588640531e-10,
+      "loss": 2.6439,
+      "step": 4751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0491063496618234e-10,
+      "loss": 2.6017,
+      "step": 4752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6561138996628046e-10,
+      "loss": 2.619,
+      "step": 4753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.290223451667517e-10,
+      "loss": 2.6516,
+      "step": 4754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9514352040173045e-10,
+      "loss": 2.6521,
+      "step": 4755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6397493403541576e-10,
+      "loss": 2.7024,
+      "step": 4756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.355166029631816e-10,
+      "loss": 2.6367,
+      "step": 4757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0976854261213199e-10,
+      "loss": 2.6887,
+      "step": 4758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.67307669383255e-11,
+      "loss": 2.5637,
+      "step": 4759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.640328843121602e-11,
+      "loss": 2.6444,
+      "step": 4760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8786118109211967e-11,
+      "loss": 2.7029,
+      "step": 4761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.387926552134157e-11,
+      "loss": 2.6555,
+      "step": 4762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.168273874891824e-11,
+      "loss": 2.6012,
+      "step": 4763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2196544402764965e-11,
+      "loss": 2.7365,
+      "step": 4764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.420687625434795e-12,
+      "loss": 2.6669,
+      "step": 4765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3551720901006094e-12,
+      "loss": 2.5921,
+      "step": 4766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 2.6295,
+      "step": 4767
+    },
+    {
+      "epoch": 1.0,
+      "step": 4767,
+      "total_flos": 1.5788618829728317e+18,
+      "train_loss": 2.666367991955395,
+      "train_runtime": 12556.6581,
+      "train_samples_per_second": 48.594,
+      "train_steps_per_second": 0.38
+    }
+  ],
+  "max_steps": 4767,
+  "num_train_epochs": 1,
+  "total_flos": 1.5788618829728317e+18,
+  "trial_name": null,
+  "trial_params": null
+}