diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,30025 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 5000,
+  "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": 0.2521,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 0.2544,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 0.2512,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 0.254,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 0.2489,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 0.2617,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-07,
+      "loss": 0.2524,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 0.2484,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 0.2576,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 0.2591,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-07,
+      "loss": 0.2631,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 0.2568,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-07,
+      "loss": 0.2407,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8e-07,
+      "loss": 0.2551,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 0.2448,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2e-07,
+      "loss": 0.2651,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 0.2568,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 0.253,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-07,
+      "loss": 0.2515,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.2548,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 0.2551,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-07,
+      "loss": 0.2576,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 0.2545,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 0.2483,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.2252,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2e-07,
+      "loss": 0.2481,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4e-07,
+      "loss": 0.2527,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6e-07,
+      "loss": 0.2255,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.2555,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 0.2378,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 0.2543,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4e-07,
+      "loss": 0.246,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6e-07,
+      "loss": 0.2456,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 0.2514,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 0.252,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-07,
+      "loss": 0.2423,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4e-07,
+      "loss": 0.2361,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6e-07,
+      "loss": 0.2351,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.8e-07,
+      "loss": 0.2444,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.2421,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 0.228,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 0.2197,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.6e-07,
+      "loss": 0.2178,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8e-07,
+      "loss": 0.2254,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 0.1851,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 0.2081,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 0.2081,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 0.1699,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.2084,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.2176,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02e-06,
+      "loss": 0.2019,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.04e-06,
+      "loss": 0.2068,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06e-06,
+      "loss": 0.2071,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.08e-06,
+      "loss": 0.2094,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1e-06,
+      "loss": 0.1984,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.12e-06,
+      "loss": 0.2004,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.14e-06,
+      "loss": 0.1927,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 0.1813,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.1791,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.1687,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 0.1661,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 0.1455,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26e-06,
+      "loss": 0.1405,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.28e-06,
+      "loss": 0.106,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3e-06,
+      "loss": 0.1442,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.32e-06,
+      "loss": 0.1445,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.34e-06,
+      "loss": 0.1375,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 0.1367,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.1507,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 0.1347,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.42e-06,
+      "loss": 0.1338,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-06,
+      "loss": 0.1305,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.46e-06,
+      "loss": 0.1284,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.48e-06,
+      "loss": 0.1208,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 0.1173,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.52e-06,
+      "loss": 0.1207,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.54e-06,
+      "loss": 0.1076,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.56e-06,
+      "loss": 0.1161,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.1132,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.1045,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 0.1007,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 0.0956,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 0.0919,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 0.0939,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 0.0876,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.72e-06,
+      "loss": 0.0925,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.74e-06,
+      "loss": 0.0849,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.76e-06,
+      "loss": 0.0838,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.0818,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 0.0717,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 0.0717,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 0.0787,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 0.0807,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 0.0667,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 0.0676,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 0.067,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.94e-06,
+      "loss": 0.0605,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 0.059,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.98e-06,
+      "loss": 0.0565,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.0554,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.02e-06,
+      "loss": 0.0543,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.04e-06,
+      "loss": 0.0539,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.06e-06,
+      "loss": 0.0577,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.08e-06,
+      "loss": 0.0607,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 0.0517,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.12e-06,
+      "loss": 0.0493,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 0.0532,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.16e-06,
+      "loss": 0.0521,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.0483,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2e-06,
+      "loss": 0.0481,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 0.0492,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.24e-06,
+      "loss": 0.0313,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 0.0389,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.28e-06,
+      "loss": 0.0428,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 0.0453,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 0.0561,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 0.0387,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 0.0406,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.38e-06,
+      "loss": 0.0421,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.0376,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.42e-06,
+      "loss": 0.0353,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 0.0321,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.46e-06,
+      "loss": 0.0358,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 0.0348,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5e-06,
+      "loss": 0.0314,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.52e-06,
+      "loss": 0.0345,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 0.0312,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.56e-06,
+      "loss": 0.029,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.0308,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6e-06,
+      "loss": 0.0286,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 0.034,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.64e-06,
+      "loss": 0.034,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 0.0383,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.68e-06,
+      "loss": 0.0294,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 0.0291,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 0.025,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 0.0281,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 0.0259,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.0568,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.025,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.82e-06,
+      "loss": 0.0274,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.84e-06,
+      "loss": 0.0256,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.86e-06,
+      "loss": 0.0224,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.88e-06,
+      "loss": 0.0271,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9e-06,
+      "loss": 0.0197,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.92e-06,
+      "loss": 0.0229,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 0.0219,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.96e-06,
+      "loss": 0.0204,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.0244,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3e-06,
+      "loss": 0.0212,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 0.0192,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.04e-06,
+      "loss": 0.0196,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 0.0176,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.08e-06,
+      "loss": 0.0175,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 0.0184,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.12e-06,
+      "loss": 0.0213,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 0.0179,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 0.0282,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.0169,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.0171,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 0.0165,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 0.0163,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 0.0157,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 0.0156,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 0.0159,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 0.0151,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 0.016,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 0.0149,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.016,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.0158,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 0.0171,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.44e-06,
+      "loss": 0.0145,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.46e-06,
+      "loss": 0.0149,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.48e-06,
+      "loss": 0.0153,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5e-06,
+      "loss": 0.0151,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.52e-06,
+      "loss": 0.0175,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.54e-06,
+      "loss": 0.0155,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 0.013,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.58e-06,
+      "loss": 0.0135,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.0129,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.62e-06,
+      "loss": 0.0128,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 0.0126,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.66e-06,
+      "loss": 0.0134,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 0.0126,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7e-06,
+      "loss": 0.0116,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 0.0145,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.74e-06,
+      "loss": 0.0126,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 0.0118,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.0107,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.0169,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 0.0178,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 0.0115,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.86e-06,
+      "loss": 0.0125,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.88e-06,
+      "loss": 0.0107,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 0.0117,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 0.0097,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.94e-06,
+      "loss": 0.0192,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.96e-06,
+      "loss": 0.0116,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.0119,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.0088,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 0.0097,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.04e-06,
+      "loss": 0.0194,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 0.0117,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.08e-06,
+      "loss": 0.0095,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1e-06,
+      "loss": 0.0093,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.12e-06,
+      "loss": 0.0113,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.14e-06,
+      "loss": 0.0098,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.16e-06,
+      "loss": 0.0126,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.18e-06,
+      "loss": 0.0125,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.0102,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.22e-06,
+      "loss": 0.0101,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.24e-06,
+      "loss": 0.0106,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.26e-06,
+      "loss": 0.0086,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 0.0088,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3e-06,
+      "loss": 0.0102,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.32e-06,
+      "loss": 0.0118,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.34e-06,
+      "loss": 0.0089,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 0.0084,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.38e-06,
+      "loss": 0.0101,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.4e-06,
+      "loss": 0.0083,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.42e-06,
+      "loss": 0.0079,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 0.0088,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 0.0123,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.48e-06,
+      "loss": 0.0087,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.5e-06,
+      "loss": 0.0116,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 0.0098,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 0.0142,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.56e-06,
+      "loss": 0.0087,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.58e-06,
+      "loss": 0.0079,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.008,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 0.009,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 0.0087,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.66e-06,
+      "loss": 0.0113,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 0.0086,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7e-06,
+      "loss": 0.0104,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 0.0099,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74e-06,
+      "loss": 0.0078,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76e-06,
+      "loss": 0.0094,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.78e-06,
+      "loss": 0.009,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.0151,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 0.0076,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.84e-06,
+      "loss": 0.0072,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.86e-06,
+      "loss": 0.0102,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 0.008,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 0.0063,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92e-06,
+      "loss": 0.0081,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.94e-06,
+      "loss": 0.0107,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 0.0084,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.0114,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5e-06,
+      "loss": 0.0085,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.02e-06,
+      "loss": 0.0067,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.04e-06,
+      "loss": 0.006,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 0.008,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 0.0076,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1e-06,
+      "loss": 0.0098,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.12e-06,
+      "loss": 0.0112,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 0.0091,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 0.0059,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.18e-06,
+      "loss": 0.0079,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2e-06,
+      "loss": 0.01,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 0.0162,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 0.0068,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 0.0087,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.28e-06,
+      "loss": 0.0087,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 0.0107,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 0.0058,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 0.0071,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.36e-06,
+      "loss": 0.0065,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.0069,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 0.0065,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 0.0073,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 0.0072,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 0.0084,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 0.0081,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.0062,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 0.0597,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 0.0063,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 0.0061,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.0068,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.006,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 0.0071,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.64e-06,
+      "loss": 0.0078,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.66e-06,
+      "loss": 0.0081,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.68e-06,
+      "loss": 0.0071,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.7e-06,
+      "loss": 0.0072,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.72e-06,
+      "loss": 0.007,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.74e-06,
+      "loss": 0.0079,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.76e-06,
+      "loss": 0.0075,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.78e-06,
+      "loss": 0.0093,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.8e-06,
+      "loss": 0.0079,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.82e-06,
+      "loss": 0.0059,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.84e-06,
+      "loss": 0.0067,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.86e-06,
+      "loss": 0.0069,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 0.0065,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9e-06,
+      "loss": 0.0123,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.92e-06,
+      "loss": 0.0064,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.94e-06,
+      "loss": 0.0065,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 0.0065,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.98e-06,
+      "loss": 0.0109,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6e-06,
+      "loss": 0.0071,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.02e-06,
+      "loss": 0.0054,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 0.0056,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 0.0086,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.08e-06,
+      "loss": 0.0058,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1e-06,
+      "loss": 0.0073,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 0.0098,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 0.0102,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.16e-06,
+      "loss": 0.0068,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.18e-06,
+      "loss": 0.0061,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 0.0074,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 0.0408,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.24e-06,
+      "loss": 0.0125,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.26e-06,
+      "loss": 0.0063,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 0.0061,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 0.0074,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 0.0065,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.34e-06,
+      "loss": 0.0061,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 0.0088,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.0065,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.0093,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.42e-06,
+      "loss": 0.0069,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 0.0077,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 0.0065,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 0.0089,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.0073,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 0.0082,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 0.0122,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 0.0068,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.0075,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 0.0076,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 0.0068,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 0.007,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 0.0072,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 0.0058,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 0.0084,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 0.0078,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 0.0107,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 0.0105,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.0105,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.0076,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 0.0071,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 0.0064,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 0.0072,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.88e-06,
+      "loss": 0.0062,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.9e-06,
+      "loss": 0.0074,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.92e-06,
+      "loss": 0.0068,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 0.0074,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.96e-06,
+      "loss": 0.0063,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.98e-06,
+      "loss": 0.0076,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7e-06,
+      "loss": 0.0073,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 0.0076,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.04e-06,
+      "loss": 0.0063,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.06e-06,
+      "loss": 0.0097,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.08e-06,
+      "loss": 0.0073,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 0.0068,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 0.0067,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.14e-06,
+      "loss": 0.0052,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.16e-06,
+      "loss": 0.0079,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.0077,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.0071,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.22e-06,
+      "loss": 0.0109,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.24e-06,
+      "loss": 0.0063,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 0.0069,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 0.0076,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.3e-06,
+      "loss": 0.0087,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.32e-06,
+      "loss": 0.0103,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 0.011,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 0.0073,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.0068,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.4e-06,
+      "loss": 0.0063,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 0.0056,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 0.0065,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 0.0067,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.48e-06,
+      "loss": 0.0078,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.0085,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 0.0065,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 0.0085,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 0.0187,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.58e-06,
+      "loss": 0.0064,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.0089,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 0.0075,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 0.0113,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 0.0104,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 0.0069,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.7e-06,
+      "loss": 0.0077,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.72e-06,
+      "loss": 0.0062,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.74e-06,
+      "loss": 0.0081,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.76e-06,
+      "loss": 0.0112,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.78e-06,
+      "loss": 0.0064,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 0.0074,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 0.0068,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 0.0065,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 0.0138,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.88e-06,
+      "loss": 0.0067,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.9e-06,
+      "loss": 0.0075,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.92e-06,
+      "loss": 0.0067,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.94e-06,
+      "loss": 0.0065,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 0.0066,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.0062,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.009,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 0.0063,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 0.0086,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.06e-06,
+      "loss": 0.0065,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.08e-06,
+      "loss": 0.0059,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.1e-06,
+      "loss": 0.0076,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 0.0067,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.14e-06,
+      "loss": 0.007,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.16e-06,
+      "loss": 0.0065,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.18e-06,
+      "loss": 0.0078,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.2e-06,
+      "loss": 0.011,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 0.0064,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.24e-06,
+      "loss": 0.007,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.26e-06,
+      "loss": 0.009,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.28e-06,
+      "loss": 0.0063,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.3e-06,
+      "loss": 0.0071,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.32e-06,
+      "loss": 0.0149,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.34e-06,
+      "loss": 0.0076,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.36e-06,
+      "loss": 0.0115,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.0067,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.0065,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.42e-06,
+      "loss": 0.0069,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.44e-06,
+      "loss": 0.0075,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.46e-06,
+      "loss": 0.0065,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.48e-06,
+      "loss": 0.0074,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.5e-06,
+      "loss": 0.008,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.52e-06,
+      "loss": 0.006,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 0.0089,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 0.0083,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.0107,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.6e-06,
+      "loss": 0.007,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.62e-06,
+      "loss": 0.0152,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.64e-06,
+      "loss": 0.0078,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.66e-06,
+      "loss": 0.0074,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.68e-06,
+      "loss": 0.0076,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 0.0071,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 0.0079,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 0.0081,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.76e-06,
+      "loss": 0.0064,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.78e-06,
+      "loss": 0.0104,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.8e-06,
+      "loss": 0.0125,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.82e-06,
+      "loss": 0.0114,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.84e-06,
+      "loss": 0.0074,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 0.0101,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 0.0095,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 0.01,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 0.0085,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.94e-06,
+      "loss": 0.0073,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.96e-06,
+      "loss": 0.0107,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.98e-06,
+      "loss": 0.0066,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9e-06,
+      "loss": 0.0062,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 0.0085,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 0.006,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 0.0087,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 0.045,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 0.0101,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.12e-06,
+      "loss": 0.0067,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.14e-06,
+      "loss": 0.0084,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.16e-06,
+      "loss": 0.008,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.0081,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.0131,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 0.0121,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 0.0152,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 0.013,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 0.0089,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.3e-06,
+      "loss": 0.0077,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.32e-06,
+      "loss": 0.0076,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 0.0082,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 0.0151,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.38e-06,
+      "loss": 0.0081,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.4e-06,
+      "loss": 0.0098,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.42e-06,
+      "loss": 0.0073,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 0.0105,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 0.01,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.48e-06,
+      "loss": 0.0098,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.5e-06,
+      "loss": 0.0157,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.52e-06,
+      "loss": 0.0089,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.54e-06,
+      "loss": 0.0104,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.56e-06,
+      "loss": 0.0099,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.58e-06,
+      "loss": 0.0087,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.0088,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 0.0106,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 0.0113,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.66e-06,
+      "loss": 0.0114,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.68e-06,
+      "loss": 0.0102,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7e-06,
+      "loss": 0.0088,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.72e-06,
+      "loss": 0.0097,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.74e-06,
+      "loss": 0.01,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 0.0128,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.0244,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.0392,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 0.0103,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.84e-06,
+      "loss": 0.0118,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86e-06,
+      "loss": 0.0108,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88e-06,
+      "loss": 0.0108,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9e-06,
+      "loss": 0.016,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 0.0104,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 0.01,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 0.0092,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.0077,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 0.0123,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999998781530372e-06,
+      "loss": 0.0106,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999995126122076e-06,
+      "loss": 0.013,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999989033776898e-06,
+      "loss": 0.012,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999980504497803e-06,
+      "loss": 0.0106,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999969538288953e-06,
+      "loss": 0.0164,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999956135155688e-06,
+      "loss": 0.0099,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999940295104546e-06,
+      "loss": 0.0122,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999922018143242e-06,
+      "loss": 0.0155,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999901304280686e-06,
+      "loss": 0.0136,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999878153526974e-06,
+      "loss": 0.0159,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.99985256589339e-06,
+      "loss": 0.0128,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999824541392404e-06,
+      "loss": 0.0111,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999794080037675e-06,
+      "loss": 0.0089,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.99976118184405e-06,
+      "loss": 0.0129,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999725846827562e-06,
+      "loss": 0.0109,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999688075005434e-06,
+      "loss": 0.0113,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999647866396073e-06,
+      "loss": 0.0136,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999605221019082e-06,
+      "loss": 0.0102,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999560138895238e-06,
+      "loss": 0.0176,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999512620046523e-06,
+      "loss": 0.0368,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999462664496088e-06,
+      "loss": 0.0195,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999410272268285e-06,
+      "loss": 0.0168,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999355443388649e-06,
+      "loss": 0.0099,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999298177883902e-06,
+      "loss": 0.0202,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.999238475781957e-06,
+      "loss": 0.0099,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999176337111908e-06,
+      "loss": 0.0221,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999111761904046e-06,
+      "loss": 0.0089,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99904475018984e-06,
+      "loss": 0.0087,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99897530200195e-06,
+      "loss": 0.0101,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998903417374228e-06,
+      "loss": 0.0092,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998829096341706e-06,
+      "loss": 0.0132,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998752338940612e-06,
+      "loss": 0.0099,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998673145208351e-06,
+      "loss": 0.0096,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998591515183524e-06,
+      "loss": 0.0114,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998507448905917e-06,
+      "loss": 0.0516,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.9984209464165e-06,
+      "loss": 0.013,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998332007757436e-06,
+      "loss": 0.0117,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998240632972073e-06,
+      "loss": 0.0117,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998146822104943e-06,
+      "loss": 0.0119,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998050575201772e-06,
+      "loss": 0.0095,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997951892309468e-06,
+      "loss": 0.0103,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997850773476126e-06,
+      "loss": 0.0092,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997747218751032e-06,
+      "loss": 0.0096,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997641228184656e-06,
+      "loss": 0.0076,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997532801828659e-06,
+      "loss": 0.0094,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997421939735885e-06,
+      "loss": 0.0153,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997308641960365e-06,
+      "loss": 0.0187,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997192908557322e-06,
+      "loss": 0.0094,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997074739583162e-06,
+      "loss": 0.0139,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99695413509548e-06,
+      "loss": 0.0142,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.996831095153054e-06,
+      "loss": 0.0212,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.996705619815857e-06,
+      "loss": 0.0116,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99657770914504e-06,
+      "loss": 0.0101,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.996447363202947e-06,
+      "loss": 0.0106,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.996314582053106e-06,
+      "loss": 0.0135,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.996179365760235e-06,
+      "loss": 0.0139,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.996041714390235e-06,
+      "loss": 0.0117,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.995901628010196e-06,
+      "loss": 0.009,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.995759106688394e-06,
+      "loss": 0.0208,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.995614150494293e-06,
+      "loss": 0.0146,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.995466759498543e-06,
+      "loss": 0.0102,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.995316933772978e-06,
+      "loss": 0.011,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.995164673390624e-06,
+      "loss": 0.0083,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.995009978425692e-06,
+      "loss": 0.01,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.994852848953574e-06,
+      "loss": 0.0109,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.994693285050858e-06,
+      "loss": 0.0107,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.994531286795309e-06,
+      "loss": 0.0115,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.994366854265886e-06,
+      "loss": 0.0117,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99419998754273e-06,
+      "loss": 0.037,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.994030686707171e-06,
+      "loss": 0.0107,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.993858951841724e-06,
+      "loss": 0.0171,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99368478303009e-06,
+      "loss": 0.01,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.993508180357154e-06,
+      "loss": 0.0155,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.993329143908994e-06,
+      "loss": 0.02,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993147673772869e-06,
+      "loss": 0.0135,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992963770037227e-06,
+      "loss": 0.0151,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992777432791697e-06,
+      "loss": 0.0089,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.9925886621271e-06,
+      "loss": 0.0155,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992397458135438e-06,
+      "loss": 0.0092,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992203820909906e-06,
+      "loss": 0.0154,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992007750544876e-06,
+      "loss": 0.0092,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991809247135912e-06,
+      "loss": 0.0109,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991608310779762e-06,
+      "loss": 0.0121,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99140494157436e-06,
+      "loss": 0.0113,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991199139618828e-06,
+      "loss": 0.0094,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990990905013466e-06,
+      "loss": 0.0127,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99078023785977e-06,
+      "loss": 0.0122,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990567138260414e-06,
+      "loss": 0.0098,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990351606319261e-06,
+      "loss": 0.0109,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990133642141359e-06,
+      "loss": 0.0094,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98991324583294e-06,
+      "loss": 0.012,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989690417501423e-06,
+      "loss": 0.0138,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989465157255413e-06,
+      "loss": 0.017,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989237465204698e-06,
+      "loss": 0.0204,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989007341460251e-06,
+      "loss": 0.0107,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988774786134235e-06,
+      "loss": 0.0109,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988539799339989e-06,
+      "loss": 0.008,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98830238119205e-06,
+      "loss": 0.0143,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988062531806127e-06,
+      "loss": 0.0087,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.987820251299121e-06,
+      "loss": 0.0149,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.987575539789119e-06,
+      "loss": 0.0122,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.987328397395389e-06,
+      "loss": 0.0111,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.987078824238384e-06,
+      "loss": 0.0127,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.986826820439743e-06,
+      "loss": 0.011,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98657238612229e-06,
+      "loss": 0.0086,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.986315521410035e-06,
+      "loss": 0.0151,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98605622642817e-06,
+      "loss": 0.0135,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98579450130307e-06,
+      "loss": 0.0117,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.9855303461623e-06,
+      "loss": 0.0102,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.985263761134602e-06,
+      "loss": 0.0118,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98499474634991e-06,
+      "loss": 0.0088,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.984723301939337e-06,
+      "loss": 0.0134,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98444942803518e-06,
+      "loss": 0.0085,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.984173124770924e-06,
+      "loss": 0.0107,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.983894392281237e-06,
+      "loss": 0.0113,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.983613230701967e-06,
+      "loss": 0.0119,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98332964017015e-06,
+      "loss": 0.0124,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.983043620824005e-06,
+      "loss": 0.0121,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.982755172802933e-06,
+      "loss": 0.0123,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.982464296247523e-06,
+      "loss": 0.0133,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98217099129954e-06,
+      "loss": 0.0102,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.981875258101944e-06,
+      "loss": 0.0107,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.981577096798864e-06,
+      "loss": 0.0122,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.981276507535625e-06,
+      "loss": 0.0105,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.980973490458728e-06,
+      "loss": 0.011,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.980668045715864e-06,
+      "loss": 0.0245,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.980360173455899e-06,
+      "loss": 0.0108,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.980049873828887e-06,
+      "loss": 0.0126,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.979737146986064e-06,
+      "loss": 0.0098,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.979421993079853e-06,
+      "loss": 0.0106,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.979104412263851e-06,
+      "loss": 0.0101,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.978784404692847e-06,
+      "loss": 0.0093,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.978461970522807e-06,
+      "loss": 0.0114,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97813710991088e-06,
+      "loss": 0.0127,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.9778098230154e-06,
+      "loss": 0.0151,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.977480109995886e-06,
+      "loss": 0.0103,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.977147971013033e-06,
+      "loss": 0.014,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97681340622872e-06,
+      "loss": 0.0182,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.976476415806013e-06,
+      "loss": 0.0141,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.976136999909156e-06,
+      "loss": 0.0102,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.975795158703576e-06,
+      "loss": 0.0154,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.975450892355882e-06,
+      "loss": 0.0097,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.975104201033868e-06,
+      "loss": 0.0105,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974755084906503e-06,
+      "loss": 0.0284,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974403544143942e-06,
+      "loss": 0.0111,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974049578917524e-06,
+      "loss": 0.0167,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973693189399767e-06,
+      "loss": 0.0316,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973334375764372e-06,
+      "loss": 0.0132,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.972973138186217e-06,
+      "loss": 0.0096,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.972609476841368e-06,
+      "loss": 0.0179,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.972243391907068e-06,
+      "loss": 0.0119,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97187488356174e-06,
+      "loss": 0.0138,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.971503951984996e-06,
+      "loss": 0.013,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.971130597357618e-06,
+      "loss": 0.0129,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.970754819861577e-06,
+      "loss": 0.0186,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.970376619680024e-06,
+      "loss": 0.0116,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.969995996997285e-06,
+      "loss": 0.0193,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.969612951998874e-06,
+      "loss": 0.0122,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.969227484871485e-06,
+      "loss": 0.0127,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.968839595802982e-06,
+      "loss": 0.0137,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.968449284982424e-06,
+      "loss": 0.1863,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.968056552600043e-06,
+      "loss": 0.0201,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.96766139884725e-06,
+      "loss": 0.0186,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.967263823916638e-06,
+      "loss": 0.0151,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.966863828001982e-06,
+      "loss": 0.0152,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.966461411298235e-06,
+      "loss": 0.0216,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.966056574001528e-06,
+      "loss": 0.0152,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.965649316309178e-06,
+      "loss": 0.0111,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.965239638419673e-06,
+      "loss": 0.0186,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.964827540532685e-06,
+      "loss": 0.013,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.964413022849069e-06,
+      "loss": 0.0131,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.963996085570854e-06,
+      "loss": 0.0127,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.96357672890125e-06,
+      "loss": 0.0181,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.963154953044646e-06,
+      "loss": 0.0123,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.962730758206612e-06,
+      "loss": 0.0102,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.962304144593893e-06,
+      "loss": 0.0166,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.961875112414417e-06,
+      "loss": 0.0209,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96144366187729e-06,
+      "loss": 0.0165,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.961009793192793e-06,
+      "loss": 0.0131,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.960573506572391e-06,
+      "loss": 0.0423,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.960134802228722e-06,
+      "loss": 0.0192,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.959693680375608e-06,
+      "loss": 0.0131,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.959250141228046e-06,
+      "loss": 0.011,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.958804185002209e-06,
+      "loss": 0.0096,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.958355811915452e-06,
+      "loss": 0.0128,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.957905022186309e-06,
+      "loss": 0.0118,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.957451816034487e-06,
+      "loss": 0.0101,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.956996193680874e-06,
+      "loss": 0.0128,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.956538155347534e-06,
+      "loss": 0.012,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.95607770125771e-06,
+      "loss": 0.0127,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.95561483163582e-06,
+      "loss": 0.0138,
+      "step": 691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.955149546707465e-06,
+      "loss": 0.0343,
+      "step": 692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.954681846699414e-06,
+      "loss": 0.0106,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.954211731839623e-06,
+      "loss": 0.0095,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.953739202357219e-06,
+      "loss": 0.0092,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.953264258482505e-06,
+      "loss": 0.0143,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.952786900446964e-06,
+      "loss": 0.0131,
+      "step": 697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.952307128483257e-06,
+      "loss": 0.0165,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.951824942825215e-06,
+      "loss": 0.0114,
+      "step": 699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.951340343707852e-06,
+      "loss": 0.0101,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.950853331367356e-06,
+      "loss": 0.0133,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.950363906041089e-06,
+      "loss": 0.0135,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.94987206796759e-06,
+      "loss": 0.018,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.94937781738658e-06,
+      "loss": 0.0115,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.948881154538946e-06,
+      "loss": 0.063,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.948382079666756e-06,
+      "loss": 0.0177,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.947880593013256e-06,
+      "loss": 0.0162,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.947376694822861e-06,
+      "loss": 0.0102,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.946870385341167e-06,
+      "loss": 0.0125,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.946361664814942e-06,
+      "loss": 0.0112,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.945850533492132e-06,
+      "loss": 0.0107,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.945336991621854e-06,
+      "loss": 0.0111,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.944821039454403e-06,
+      "loss": 0.0671,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.944302677241247e-06,
+      "loss": 0.0195,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.94378190523503e-06,
+      "loss": 0.014,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.94325872368957e-06,
+      "loss": 0.0139,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.942733132859861e-06,
+      "loss": 0.0174,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.942205133002067e-06,
+      "loss": 0.0128,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.94167472437353e-06,
+      "loss": 0.0162,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.941141907232766e-06,
+      "loss": 0.0106,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.94060668183946e-06,
+      "loss": 0.0115,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.940069048454478e-06,
+      "loss": 0.0235,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.939529007339852e-06,
+      "loss": 0.0402,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.938986558758795e-06,
+      "loss": 0.0119,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.938441702975689e-06,
+      "loss": 0.0118,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.937894440256091e-06,
+      "loss": 0.0135,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.937344770866727e-06,
+      "loss": 0.0143,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.936792695075502e-06,
+      "loss": 0.016,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.936238213151491e-06,
+      "loss": 0.0386,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.93568132536494e-06,
+      "loss": 0.0125,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.93512203198727e-06,
+      "loss": 0.0103,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.934560333291077e-06,
+      "loss": 0.0119,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.93399622955012e-06,
+      "loss": 0.0123,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.93342972103934e-06,
+      "loss": 0.0395,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.932860808034847e-06,
+      "loss": 0.0137,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.932289490813922e-06,
+      "loss": 0.0112,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.931715769655017e-06,
+      "loss": 0.0108,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.931139644837755e-06,
+      "loss": 0.0174,
+      "step": 738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.930561116642936e-06,
+      "loss": 0.0346,
+      "step": 739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.929980185352525e-06,
+      "loss": 0.0096,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.929396851249661e-06,
+      "loss": 0.0151,
+      "step": 741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.928811114618658e-06,
+      "loss": 0.0114,
+      "step": 742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.928222975744992e-06,
+      "loss": 0.0149,
+      "step": 743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.927632434915315e-06,
+      "loss": 0.0124,
+      "step": 744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.927039492417452e-06,
+      "loss": 0.0275,
+      "step": 745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.926444148540394e-06,
+      "loss": 0.0146,
+      "step": 746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.925846403574306e-06,
+      "loss": 0.031,
+      "step": 747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.925246257810519e-06,
+      "loss": 0.0106,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.92464371154154e-06,
+      "loss": 0.0104,
+      "step": 749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.924038765061042e-06,
+      "loss": 0.0257,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.923431418663866e-06,
+      "loss": 0.0138,
+      "step": 751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.922821672646028e-06,
+      "loss": 0.0123,
+      "step": 752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.922209527304709e-06,
+      "loss": 0.0143,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.921594982938262e-06,
+      "loss": 0.0342,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.92097803984621e-06,
+      "loss": 0.0143,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.920358698329242e-06,
+      "loss": 0.0156,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.919736958689216e-06,
+      "loss": 0.0098,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.919112821229165e-06,
+      "loss": 0.0152,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.918486286253279e-06,
+      "loss": 0.01,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.91785735406693e-06,
+      "loss": 0.0147,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.91722602497665e-06,
+      "loss": 0.0115,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.91659229929014e-06,
+      "loss": 0.0177,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.915956177316269e-06,
+      "loss": 0.0115,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.915317659365078e-06,
+      "loss": 0.0353,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.914676745747772e-06,
+      "loss": 0.0111,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.914033436776724e-06,
+      "loss": 0.0106,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.913387732765475e-06,
+      "loss": 0.01,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.912739634028734e-06,
+      "loss": 0.0088,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.912089140882377e-06,
+      "loss": 0.0197,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.911436253643445e-06,
+      "loss": 0.0123,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.910780972630146e-06,
+      "loss": 0.0135,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.91012329816186e-06,
+      "loss": 0.0175,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.909463230559127e-06,
+      "loss": 0.0382,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.908800770143654e-06,
+      "loss": 0.0141,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.908135917238321e-06,
+      "loss": 0.0162,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.907468672167165e-06,
+      "loss": 0.0133,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.906799035255395e-06,
+      "loss": 0.0179,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.906127006829385e-06,
+      "loss": 0.0108,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.90545258721667e-06,
+      "loss": 0.0138,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.904775776745959e-06,
+      "loss": 0.0115,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.904096575747117e-06,
+      "loss": 0.015,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.903414984551178e-06,
+      "loss": 0.0129,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.902731003490344e-06,
+      "loss": 0.0189,
+      "step": 783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.90204463289798e-06,
+      "loss": 0.0145,
+      "step": 784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.901355873108611e-06,
+      "loss": 0.0152,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.900664724457932e-06,
+      "loss": 0.1867,
+      "step": 786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.899971187282799e-06,
+      "loss": 0.01,
+      "step": 787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.899275261921236e-06,
+      "loss": 0.0152,
+      "step": 788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.898576948712427e-06,
+      "loss": 0.0275,
+      "step": 789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.89787624799672e-06,
+      "loss": 0.0133,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.897173160115633e-06,
+      "loss": 0.0169,
+      "step": 791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.896467685411838e-06,
+      "loss": 0.0121,
+      "step": 792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.895759824229176e-06,
+      "loss": 0.0141,
+      "step": 793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.89504957691265e-06,
+      "loss": 0.0172,
+      "step": 794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.894336943808426e-06,
+      "loss": 0.0133,
+      "step": 795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.893621925263832e-06,
+      "loss": 0.0105,
+      "step": 796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.89290452162736e-06,
+      "loss": 0.0143,
+      "step": 797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.892184733248666e-06,
+      "loss": 0.0178,
+      "step": 798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.891462560478562e-06,
+      "loss": 0.0442,
+      "step": 799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.890738003669029e-06,
+      "loss": 0.0146,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.890011063173207e-06,
+      "loss": 0.0183,
+      "step": 801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.889281739345395e-06,
+      "loss": 0.0158,
+      "step": 802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.88855003254106e-06,
+      "loss": 0.0128,
+      "step": 803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.887815943116827e-06,
+      "loss": 0.0229,
+      "step": 804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.887079471430481e-06,
+      "loss": 0.0183,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.886340617840968e-06,
+      "loss": 0.0542,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.8855993827084e-06,
+      "loss": 0.0151,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.884855766394041e-06,
+      "loss": 0.0172,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.884109769260326e-06,
+      "loss": 0.0168,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.883361391670841e-06,
+      "loss": 0.0661,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.882610633990337e-06,
+      "loss": 0.0106,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.881857496584726e-06,
+      "loss": 0.0218,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.881101979821075e-06,
+      "loss": 0.0148,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.880344084067616e-06,
+      "loss": 0.019,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.879583809693737e-06,
+      "loss": 0.0158,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.878821157069988e-06,
+      "loss": 0.0148,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.878056126568077e-06,
+      "loss": 0.0149,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.877288718560866e-06,
+      "loss": 0.0166,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.876518933422385e-06,
+      "loss": 0.0373,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.875746771527817e-06,
+      "loss": 0.0174,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.874972233253503e-06,
+      "loss": 0.0152,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.874195318976945e-06,
+      "loss": 0.0119,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.873416029076801e-06,
+      "loss": 0.0136,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.872634363932887e-06,
+      "loss": 0.0137,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.871850323926178e-06,
+      "loss": 0.0187,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.871063909438803e-06,
+      "loss": 0.0134,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.870275120854055e-06,
+      "loss": 0.0142,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.869483958556376e-06,
+      "loss": 0.0198,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.868690422931372e-06,
+      "loss": 0.0143,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.867894514365802e-06,
+      "loss": 0.0145,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.867096233247581e-06,
+      "loss": 0.0149,
+      "step": 831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.866295579965782e-06,
+      "loss": 0.0131,
+      "step": 832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.865492554910634e-06,
+      "loss": 0.0124,
+      "step": 833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.86468715847352e-06,
+      "loss": 0.0247,
+      "step": 834
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.863879391046985e-06,
+      "loss": 0.0127,
+      "step": 835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.863069253024719e-06,
+      "loss": 0.0157,
+      "step": 836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.862256744801576e-06,
+      "loss": 0.0192,
+      "step": 837
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.861441866773564e-06,
+      "loss": 0.0128,
+      "step": 838
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.860624619337844e-06,
+      "loss": 0.0112,
+      "step": 839
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.859805002892733e-06,
+      "loss": 0.0114,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.8589830178377e-06,
+      "loss": 0.0122,
+      "step": 841
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.85815866457337e-06,
+      "loss": 0.0132,
+      "step": 842
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.857331943501527e-06,
+      "loss": 0.0193,
+      "step": 843
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.856502855025101e-06,
+      "loss": 0.027,
+      "step": 844
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.85567139954818e-06,
+      "loss": 0.0159,
+      "step": 845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.854837577476008e-06,
+      "loss": 0.0199,
+      "step": 846
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.854001389214979e-06,
+      "loss": 0.0121,
+      "step": 847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.853162835172638e-06,
+      "loss": 0.0133,
+      "step": 848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.852321915757688e-06,
+      "loss": 0.0167,
+      "step": 849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.851478631379982e-06,
+      "loss": 0.0137,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.85063298245053e-06,
+      "loss": 0.0141,
+      "step": 851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.849784969381488e-06,
+      "loss": 0.0116,
+      "step": 852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.848934592586165e-06,
+      "loss": 0.0116,
+      "step": 853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.84808185247903e-06,
+      "loss": 0.0163,
+      "step": 854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.847226749475696e-06,
+      "loss": 0.0131,
+      "step": 855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.846369283992927e-06,
+      "loss": 0.0131,
+      "step": 856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.845509456448642e-06,
+      "loss": 0.0151,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.844647267261915e-06,
+      "loss": 0.0609,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.843782716852963e-06,
+      "loss": 0.0142,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.842915805643156e-06,
+      "loss": 0.0124,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.84204653405502e-06,
+      "loss": 0.0136,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.841174902512223e-06,
+      "loss": 0.0201,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.84030091143959e-06,
+      "loss": 0.017,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.839424561263094e-06,
+      "loss": 0.0162,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.838545852409857e-06,
+      "loss": 0.0127,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.83766478530815e-06,
+      "loss": 0.0102,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.836781360387396e-06,
+      "loss": 0.0133,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.835895578078165e-06,
+      "loss": 0.0108,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.835007438812177e-06,
+      "loss": 0.0179,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.834116943022299e-06,
+      "loss": 0.0119,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.833224091142548e-06,
+      "loss": 0.0143,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.832328883608088e-06,
+      "loss": 0.0165,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.831431320855235e-06,
+      "loss": 0.0268,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.830531403321451e-06,
+      "loss": 0.0135,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.829629131445342e-06,
+      "loss": 0.009,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.828724505666664e-06,
+      "loss": 0.0428,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.827817526426324e-06,
+      "loss": 0.0281,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.82690819416637e-06,
+      "loss": 0.0117,
+      "step": 878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.825996509330001e-06,
+      "loss": 0.0135,
+      "step": 879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.825082472361558e-06,
+      "loss": 0.0112,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.824166083706534e-06,
+      "loss": 0.0111,
+      "step": 881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.823247343811567e-06,
+      "loss": 0.0872,
+      "step": 882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.822326253124436e-06,
+      "loss": 0.015,
+      "step": 883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.821402812094074e-06,
+      "loss": 0.0221,
+      "step": 884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.82047702117055e-06,
+      "loss": 0.0125,
+      "step": 885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.819548880805087e-06,
+      "loss": 0.0241,
+      "step": 886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.81861839145005e-06,
+      "loss": 0.0138,
+      "step": 887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.817685553558945e-06,
+      "loss": 0.018,
+      "step": 888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.816750367586424e-06,
+      "loss": 0.0117,
+      "step": 889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.815812833988292e-06,
+      "loss": 0.0112,
+      "step": 890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.814872953221487e-06,
+      "loss": 0.011,
+      "step": 891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.813930725744095e-06,
+      "loss": 0.0113,
+      "step": 892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.812986152015349e-06,
+      "loss": 0.0188,
+      "step": 893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.81203923249562e-06,
+      "loss": 0.0177,
+      "step": 894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.811089967646427e-06,
+      "loss": 0.0366,
+      "step": 895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.81013835793043e-06,
+      "loss": 0.0137,
+      "step": 896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.809184403811432e-06,
+      "loss": 0.0138,
+      "step": 897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.808228105754378e-06,
+      "loss": 0.0133,
+      "step": 898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.807269464225355e-06,
+      "loss": 0.0274,
+      "step": 899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.806308479691595e-06,
+      "loss": 0.0112,
+      "step": 900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.80534515262147e-06,
+      "loss": 0.0181,
+      "step": 901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.804379483484493e-06,
+      "loss": 0.0213,
+      "step": 902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.803411472751321e-06,
+      "loss": 0.013,
+      "step": 903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.80244112089375e-06,
+      "loss": 0.0126,
+      "step": 904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.801468428384716e-06,
+      "loss": 0.0132,
+      "step": 905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.8004933956983e-06,
+      "loss": 0.0189,
+      "step": 906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.799516023309719e-06,
+      "loss": 0.016,
+      "step": 907
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.798536311695334e-06,
+      "loss": 0.0131,
+      "step": 908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.797554261332644e-06,
+      "loss": 0.0113,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.796569872700287e-06,
+      "loss": 0.0101,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.795583146278047e-06,
+      "loss": 0.0146,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.794594082546835e-06,
+      "loss": 0.0887,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.793602681988714e-06,
+      "loss": 0.0128,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.79260894508688e-06,
+      "loss": 0.0129,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.791612872325667e-06,
+      "loss": 0.014,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.79061446419055e-06,
+      "loss": 0.0125,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.789613721168138e-06,
+      "loss": 0.0128,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.788610643746184e-06,
+      "loss": 0.0269,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.787605232413575e-06,
+      "loss": 0.0876,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.786597487660336e-06,
+      "loss": 0.0128,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.785587409977632e-06,
+      "loss": 0.0146,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.784574999857757e-06,
+      "loss": 0.0195,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.783560257794153e-06,
+      "loss": 0.0137,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.78254318428139e-06,
+      "loss": 0.014,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.781523779815178e-06,
+      "loss": 0.012,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.780502044892363e-06,
+      "loss": 0.0141,
+      "step": 926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.779477980010924e-06,
+      "loss": 0.0138,
+      "step": 927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.778451585669982e-06,
+      "loss": 0.0137,
+      "step": 928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.777422862369782e-06,
+      "loss": 0.0132,
+      "step": 929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.776391810611719e-06,
+      "loss": 0.0158,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.775358430898311e-06,
+      "loss": 0.0118,
+      "step": 931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.774322723733216e-06,
+      "loss": 0.015,
+      "step": 932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.773284689621223e-06,
+      "loss": 0.2284,
+      "step": 933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.772244329068261e-06,
+      "loss": 0.0142,
+      "step": 934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.771201642581384e-06,
+      "loss": 0.0125,
+      "step": 935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.77015663066879e-06,
+      "loss": 0.0168,
+      "step": 936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.769109293839803e-06,
+      "loss": 0.0185,
+      "step": 937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.768059632604881e-06,
+      "loss": 0.0128,
+      "step": 938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.767007647475618e-06,
+      "loss": 0.0134,
+      "step": 939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.765953338964736e-06,
+      "loss": 0.0138,
+      "step": 940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.764896707586095e-06,
+      "loss": 0.0124,
+      "step": 941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.763837753854684e-06,
+      "loss": 0.0222,
+      "step": 942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.762776478286622e-06,
+      "loss": 0.0093,
+      "step": 943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.761712881399164e-06,
+      "loss": 0.0119,
+      "step": 944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.760646963710694e-06,
+      "loss": 0.0148,
+      "step": 945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.759578725740726e-06,
+      "loss": 0.0327,
+      "step": 946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.758508168009908e-06,
+      "loss": 0.0131,
+      "step": 947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.757435291040016e-06,
+      "loss": 0.0094,
+      "step": 948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.756360095353957e-06,
+      "loss": 0.0119,
+      "step": 949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.755282581475769e-06,
+      "loss": 0.0159,
+      "step": 950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.754202749930618e-06,
+      "loss": 0.0109,
+      "step": 951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.7531206012448e-06,
+      "loss": 0.0175,
+      "step": 952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.752036135945743e-06,
+      "loss": 0.0116,
+      "step": 953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.750949354562006e-06,
+      "loss": 0.0141,
+      "step": 954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.749860257623262e-06,
+      "loss": 0.0146,
+      "step": 955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.748768845660335e-06,
+      "loss": 0.0151,
+      "step": 956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.74767511920516e-06,
+      "loss": 0.039,
+      "step": 957
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.746579078790808e-06,
+      "loss": 0.018,
+      "step": 958
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.745480724951473e-06,
+      "loss": 0.0157,
+      "step": 959
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.744380058222483e-06,
+      "loss": 0.013,
+      "step": 960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.743277079140288e-06,
+      "loss": 0.0138,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.742171788242468e-06,
+      "loss": 0.0122,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.741064186067723e-06,
+      "loss": 0.0335,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.739954273155892e-06,
+      "loss": 0.0162,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.73884205004793e-06,
+      "loss": 0.0115,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.73772751728592e-06,
+      "loss": 0.0128,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.736610675413073e-06,
+      "loss": 0.0119,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.735491524973723e-06,
+      "loss": 0.0139,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.73437006651333e-06,
+      "loss": 0.0197,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.733246300578482e-06,
+      "loss": 0.0151,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.732120227716887e-06,
+      "loss": 0.0171,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.73099184847738e-06,
+      "loss": 0.0112,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.72986116340992e-06,
+      "loss": 0.0119,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.728728173065584e-06,
+      "loss": 0.0152,
+      "step": 974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.727592877996585e-06,
+      "loss": 0.0118,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.726455278756249e-06,
+      "loss": 0.0117,
+      "step": 976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.725315375899025e-06,
+      "loss": 0.012,
+      "step": 977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.724173169980492e-06,
+      "loss": 0.0261,
+      "step": 978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.723028661557345e-06,
+      "loss": 0.0135,
+      "step": 979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.721881851187406e-06,
+      "loss": 0.011,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.720732739429614e-06,
+      "loss": 0.013,
+      "step": 981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.719581326844033e-06,
+      "loss": 0.012,
+      "step": 982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.718427613991848e-06,
+      "loss": 0.0128,
+      "step": 983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.717271601435363e-06,
+      "loss": 0.0109,
+      "step": 984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.716113289738005e-06,
+      "loss": 0.0138,
+      "step": 985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.714952679464324e-06,
+      "loss": 0.01,
+      "step": 986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.713789771179983e-06,
+      "loss": 0.0224,
+      "step": 987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.712624565451772e-06,
+      "loss": 0.0332,
+      "step": 988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.711457062847596e-06,
+      "loss": 0.0148,
+      "step": 989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.710287263936485e-06,
+      "loss": 0.0171,
+      "step": 990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.709115169288582e-06,
+      "loss": 0.0239,
+      "step": 991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.707940779475151e-06,
+      "loss": 0.0185,
+      "step": 992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.706764095068579e-06,
+      "loss": 0.0115,
+      "step": 993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.705585116642364e-06,
+      "loss": 0.0113,
+      "step": 994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.704403844771128e-06,
+      "loss": 0.0124,
+      "step": 995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.703220280030607e-06,
+      "loss": 0.0125,
+      "step": 996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.702034422997658e-06,
+      "loss": 0.0126,
+      "step": 997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.700846274250252e-06,
+      "loss": 0.0119,
+      "step": 998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.699655834367479e-06,
+      "loss": 0.0135,
+      "step": 999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.698463103929542e-06,
+      "loss": 0.0095,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.697268083517767e-06,
+      "loss": 0.0129,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.696070773714592e-06,
+      "loss": 0.0101,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.69487117510357e-06,
+      "loss": 0.019,
+      "step": 1003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.693669288269371e-06,
+      "loss": 0.014,
+      "step": 1004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.69246511379778e-06,
+      "loss": 0.014,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.691258652275698e-06,
+      "loss": 0.0099,
+      "step": 1006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.690049904291139e-06,
+      "loss": 0.0133,
+      "step": 1007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.68883887043323e-06,
+      "loss": 0.0107,
+      "step": 1008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.687625551292219e-06,
+      "loss": 0.0258,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.68640994745946e-06,
+      "loss": 0.0257,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.68519205952742e-06,
+      "loss": 0.0148,
+      "step": 1011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.68397188808969e-06,
+      "loss": 0.1039,
+      "step": 1012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.682749433740963e-06,
+      "loss": 0.0122,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.681524697077047e-06,
+      "loss": 0.0147,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.680297678694867e-06,
+      "loss": 0.0231,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.679068379192455e-06,
+      "loss": 0.0101,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.677836799168958e-06,
+      "loss": 0.0115,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.67660293922463e-06,
+      "loss": 0.0151,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.675366799960842e-06,
+      "loss": 0.0105,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.674128381980073e-06,
+      "loss": 0.0105,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.672887685885913e-06,
+      "loss": 0.0097,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.671644712283061e-06,
+      "loss": 0.0218,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.670399461777328e-06,
+      "loss": 0.0135,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.669151934975635e-06,
+      "loss": 0.0136,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.667902132486009e-06,
+      "loss": 0.0169,
+      "step": 1025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.666650054917591e-06,
+      "loss": 0.0169,
+      "step": 1026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.665395702880627e-06,
+      "loss": 0.0137,
+      "step": 1027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.664139076986473e-06,
+      "loss": 0.0751,
+      "step": 1028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.662880177847595e-06,
+      "loss": 0.013,
+      "step": 1029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.661619006077562e-06,
+      "loss": 0.0157,
+      "step": 1030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.660355562291055e-06,
+      "loss": 0.0136,
+      "step": 1031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.659089847103863e-06,
+      "loss": 0.0249,
+      "step": 1032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.65782186113288e-06,
+      "loss": 0.0157,
+      "step": 1033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.656551604996102e-06,
+      "loss": 0.0126,
+      "step": 1034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.655279079312643e-06,
+      "loss": 0.0115,
+      "step": 1035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.654004284702712e-06,
+      "loss": 0.0134,
+      "step": 1036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.65272722178763e-06,
+      "loss": 0.0134,
+      "step": 1037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.651447891189824e-06,
+      "loss": 0.0107,
+      "step": 1038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.650166293532822e-06,
+      "loss": 0.0163,
+      "step": 1039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.648882429441258e-06,
+      "loss": 0.0126,
+      "step": 1040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.647596299540874e-06,
+      "loss": 0.0162,
+      "step": 1041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.646307904458513e-06,
+      "loss": 0.0154,
+      "step": 1042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.645017244822124e-06,
+      "loss": 0.0159,
+      "step": 1043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.643724321260757e-06,
+      "loss": 0.0161,
+      "step": 1044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.642429134404568e-06,
+      "loss": 0.0135,
+      "step": 1045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.641131684884817e-06,
+      "loss": 0.016,
+      "step": 1046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.639831973333864e-06,
+      "loss": 0.0157,
+      "step": 1047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.638530000385171e-06,
+      "loss": 0.0163,
+      "step": 1048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.637225766673309e-06,
+      "loss": 0.0192,
+      "step": 1049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.635919272833938e-06,
+      "loss": 0.0162,
+      "step": 1050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.634610519503833e-06,
+      "loss": 0.016,
+      "step": 1051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.633299507320862e-06,
+      "loss": 0.0171,
+      "step": 1052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.631986236923998e-06,
+      "loss": 0.0152,
+      "step": 1053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.630670708953311e-06,
+      "loss": 0.0227,
+      "step": 1054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.629352924049975e-06,
+      "loss": 0.0125,
+      "step": 1055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.628032882856262e-06,
+      "loss": 0.0115,
+      "step": 1056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.626710586015543e-06,
+      "loss": 0.0144,
+      "step": 1057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.62538603417229e-06,
+      "loss": 0.0174,
+      "step": 1058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.624059227972077e-06,
+      "loss": 0.0141,
+      "step": 1059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.622730168061568e-06,
+      "loss": 0.0158,
+      "step": 1060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.62139885508853e-06,
+      "loss": 0.0158,
+      "step": 1061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.620065289701835e-06,
+      "loss": 0.017,
+      "step": 1062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.61872947255144e-06,
+      "loss": 0.0109,
+      "step": 1063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.617391404288412e-06,
+      "loss": 0.0159,
+      "step": 1064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.616051085564905e-06,
+      "loss": 0.0145,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.614708517034176e-06,
+      "loss": 0.0131,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.613363699350575e-06,
+      "loss": 0.0135,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.612016633169552e-06,
+      "loss": 0.0254,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.610667319147648e-06,
+      "loss": 0.0141,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.609315757942504e-06,
+      "loss": 0.0128,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.607961950212855e-06,
+      "loss": 0.0166,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.606605896618528e-06,
+      "loss": 0.0107,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.605247597820448e-06,
+      "loss": 0.0129,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.603887054480636e-06,
+      "loss": 0.0149,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.602524267262202e-06,
+      "loss": 0.0139,
+      "step": 1075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.601159236829353e-06,
+      "loss": 0.0117,
+      "step": 1076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.599791963847388e-06,
+      "loss": 0.0144,
+      "step": 1077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.598422448982697e-06,
+      "loss": 0.0102,
+      "step": 1078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.597050692902765e-06,
+      "loss": 0.0263,
+      "step": 1079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.595676696276173e-06,
+      "loss": 0.0123,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.594300459772588e-06,
+      "loss": 0.0125,
+      "step": 1081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.592921984062771e-06,
+      "loss": 0.0118,
+      "step": 1082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.591541269818574e-06,
+      "loss": 0.0189,
+      "step": 1083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.590158317712941e-06,
+      "loss": 0.0213,
+      "step": 1084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.588773128419907e-06,
+      "loss": 0.0107,
+      "step": 1085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.587385702614593e-06,
+      "loss": 0.015,
+      "step": 1086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.585996040973218e-06,
+      "loss": 0.0149,
+      "step": 1087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.584604144173084e-06,
+      "loss": 0.0105,
+      "step": 1088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.583210012892582e-06,
+      "loss": 0.0121,
+      "step": 1089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.581813647811199e-06,
+      "loss": 0.0141,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.580415049609503e-06,
+      "loss": 0.0171,
+      "step": 1091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.579014218969158e-06,
+      "loss": 0.0124,
+      "step": 1092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.577611156572908e-06,
+      "loss": 0.0145,
+      "step": 1093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.576205863104588e-06,
+      "loss": 0.013,
+      "step": 1094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.574798339249124e-06,
+      "loss": 0.0162,
+      "step": 1095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.573388585692525e-06,
+      "loss": 0.0099,
+      "step": 1096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.571976603121889e-06,
+      "loss": 0.0151,
+      "step": 1097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.570562392225395e-06,
+      "loss": 0.0175,
+      "step": 1098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.569145953692316e-06,
+      "loss": 0.0148,
+      "step": 1099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.567727288213005e-06,
+      "loss": 0.0126,
+      "step": 1100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.566306396478904e-06,
+      "loss": 0.0393,
+      "step": 1101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.564883279182538e-06,
+      "loss": 0.0128,
+      "step": 1102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.563457937017514e-06,
+      "loss": 0.0144,
+      "step": 1103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.562030370678533e-06,
+      "loss": 0.0114,
+      "step": 1104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.560600580861366e-06,
+      "loss": 0.0176,
+      "step": 1105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.55916856826288e-06,
+      "loss": 0.0113,
+      "step": 1106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.557734333581019e-06,
+      "loss": 0.0149,
+      "step": 1107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.556297877514812e-06,
+      "loss": 0.0221,
+      "step": 1108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.554859200764371e-06,
+      "loss": 0.0114,
+      "step": 1109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.553418304030886e-06,
+      "loss": 0.0144,
+      "step": 1110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.551975188016638e-06,
+      "loss": 0.0107,
+      "step": 1111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.550529853424979e-06,
+      "loss": 0.0143,
+      "step": 1112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.549082300960351e-06,
+      "loss": 0.0119,
+      "step": 1113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.547632531328273e-06,
+      "loss": 0.0132,
+      "step": 1114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.546180545235344e-06,
+      "loss": 0.0144,
+      "step": 1115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.544726343389245e-06,
+      "loss": 0.0169,
+      "step": 1116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.543269926498735e-06,
+      "loss": 0.0746,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.541811295273657e-06,
+      "loss": 0.0839,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.540350450424927e-06,
+      "loss": 0.0159,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.538887392664544e-06,
+      "loss": 0.02,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.537422122705585e-06,
+      "loss": 0.0143,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.535954641262206e-06,
+      "loss": 0.0175,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.534484949049636e-06,
+      "loss": 0.0196,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.53301304678419e-06,
+      "loss": 0.0185,
+      "step": 1124
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.531538935183252e-06,
+      "loss": 0.0235,
+      "step": 1125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.530062614965286e-06,
+      "loss": 0.0139,
+      "step": 1126
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.528584086849832e-06,
+      "loss": 0.0189,
+      "step": 1127
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.52710335155751e-06,
+      "loss": 0.0138,
+      "step": 1128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.525620409810009e-06,
+      "loss": 0.0211,
+      "step": 1129
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.524135262330098e-06,
+      "loss": 0.225,
+      "step": 1130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.52264790984162e-06,
+      "loss": 0.0308,
+      "step": 1131
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.521158353069494e-06,
+      "loss": 0.0628,
+      "step": 1132
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.51966659273971e-06,
+      "loss": 0.0177,
+      "step": 1133
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.518172629579334e-06,
+      "loss": 0.0138,
+      "step": 1134
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.516676464316505e-06,
+      "loss": 0.0172,
+      "step": 1135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.515178097680437e-06,
+      "loss": 0.0135,
+      "step": 1136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.513677530401415e-06,
+      "loss": 0.0147,
+      "step": 1137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.512174763210798e-06,
+      "loss": 0.0271,
+      "step": 1138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.510669796841014e-06,
+      "loss": 0.0123,
+      "step": 1139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.50916263202557e-06,
+      "loss": 0.0163,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.507653269499035e-06,
+      "loss": 0.0244,
+      "step": 1141
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.506141709997058e-06,
+      "loss": 0.0141,
+      "step": 1142
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.504627954256352e-06,
+      "loss": 0.0133,
+      "step": 1143
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.503112003014702e-06,
+      "loss": 0.0145,
+      "step": 1144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.501593857010968e-06,
+      "loss": 0.0144,
+      "step": 1145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.500073516985074e-06,
+      "loss": 0.0143,
+      "step": 1146
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.498550983678016e-06,
+      "loss": 0.0133,
+      "step": 1147
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.497026257831856e-06,
+      "loss": 0.0157,
+      "step": 1148
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.495499340189729e-06,
+      "loss": 0.0138,
+      "step": 1149
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.493970231495836e-06,
+      "loss": 0.0125,
+      "step": 1150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.492438932495444e-06,
+      "loss": 0.0125,
+      "step": 1151
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.490905443934892e-06,
+      "loss": 0.0114,
+      "step": 1152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.489369766561584e-06,
+      "loss": 0.0125,
+      "step": 1153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.487831901123989e-06,
+      "loss": 0.0128,
+      "step": 1154
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.486291848371642e-06,
+      "loss": 0.2382,
+      "step": 1155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.484749609055151e-06,
+      "loss": 0.0125,
+      "step": 1156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.48320518392618e-06,
+      "loss": 0.0231,
+      "step": 1157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.481658573737465e-06,
+      "loss": 0.0154,
+      "step": 1158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.480109779242805e-06,
+      "loss": 0.0147,
+      "step": 1159
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.478558801197065e-06,
+      "loss": 0.0131,
+      "step": 1160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.47700564035617e-06,
+      "loss": 0.0152,
+      "step": 1161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.475450297477113e-06,
+      "loss": 0.0168,
+      "step": 1162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.473892773317952e-06,
+      "loss": 0.013,
+      "step": 1163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.4723330686378e-06,
+      "loss": 0.0106,
+      "step": 1164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.470771184196842e-06,
+      "loss": 0.0116,
+      "step": 1165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.46920712075632e-06,
+      "loss": 0.0123,
+      "step": 1166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.46764087907854e-06,
+      "loss": 0.0148,
+      "step": 1167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.46607245992687e-06,
+      "loss": 0.0139,
+      "step": 1168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.464501864065735e-06,
+      "loss": 0.0175,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.46292909226063e-06,
+      "loss": 0.0143,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.461354145278098e-06,
+      "loss": 0.012,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.459777023885754e-06,
+      "loss": 0.0134,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.458197728852268e-06,
+      "loss": 0.0139,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.456616260947367e-06,
+      "loss": 0.0161,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.45503262094184e-06,
+      "loss": 0.0153,
+      "step": 1175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.453446809607534e-06,
+      "loss": 0.0138,
+      "step": 1176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.451858827717354e-06,
+      "loss": 0.0164,
+      "step": 1177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.450268676045261e-06,
+      "loss": 0.0287,
+      "step": 1178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.448676355366282e-06,
+      "loss": 0.0174,
+      "step": 1179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.44708186645649e-06,
+      "loss": 0.0109,
+      "step": 1180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.445485210093018e-06,
+      "loss": 0.0123,
+      "step": 1181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.443886387054058e-06,
+      "loss": 0.015,
+      "step": 1182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.44228539811886e-06,
+      "loss": 0.0146,
+      "step": 1183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.440682244067724e-06,
+      "loss": 0.0149,
+      "step": 1184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.439076925682006e-06,
+      "loss": 0.0146,
+      "step": 1185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.437469443744124e-06,
+      "loss": 0.0125,
+      "step": 1186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.435859799037541e-06,
+      "loss": 0.018,
+      "step": 1187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.43424799234678e-06,
+      "loss": 0.0144,
+      "step": 1188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.432634024457414e-06,
+      "loss": 0.0135,
+      "step": 1189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.431017896156074e-06,
+      "loss": 0.0112,
+      "step": 1190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.429399608230441e-06,
+      "loss": 0.0307,
+      "step": 1191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.427779161469246e-06,
+      "loss": 0.0115,
+      "step": 1192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.426156556662276e-06,
+      "loss": 0.0179,
+      "step": 1193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.424531794600372e-06,
+      "loss": 0.0146,
+      "step": 1194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.42290487607542e-06,
+      "loss": 0.0166,
+      "step": 1195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.421275801880363e-06,
+      "loss": 0.0143,
+      "step": 1196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.419644572809189e-06,
+      "loss": 0.0133,
+      "step": 1197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.418011189656942e-06,
+      "loss": 0.0108,
+      "step": 1198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.41637565321971e-06,
+      "loss": 0.0122,
+      "step": 1199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.414737964294636e-06,
+      "loss": 0.0151,
+      "step": 1200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.41309812367991e-06,
+      "loss": 0.0149,
+      "step": 1201
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.411456132174768e-06,
+      "loss": 0.0116,
+      "step": 1202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.409811990579498e-06,
+      "loss": 0.0114,
+      "step": 1203
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.408165699695435e-06,
+      "loss": 0.0173,
+      "step": 1204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.406517260324962e-06,
+      "loss": 0.0146,
+      "step": 1205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.404866673271506e-06,
+      "loss": 0.0134,
+      "step": 1206
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.403213939339546e-06,
+      "loss": 0.0131,
+      "step": 1207
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.401559059334601e-06,
+      "loss": 0.0119,
+      "step": 1208
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.399902034063244e-06,
+      "loss": 0.017,
+      "step": 1209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.398242864333084e-06,
+      "loss": 0.0185,
+      "step": 1210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.396581550952781e-06,
+      "loss": 0.0124,
+      "step": 1211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.394918094732044e-06,
+      "loss": 0.0116,
+      "step": 1212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.393252496481615e-06,
+      "loss": 0.0291,
+      "step": 1213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.39158475701329e-06,
+      "loss": 0.0189,
+      "step": 1214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.389914877139903e-06,
+      "loss": 0.011,
+      "step": 1215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.388242857675336e-06,
+      "loss": 0.0122,
+      "step": 1216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.386568699434509e-06,
+      "loss": 0.0127,
+      "step": 1217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.384892403233384e-06,
+      "loss": 0.0096,
+      "step": 1218
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.383213969888972e-06,
+      "loss": 0.0171,
+      "step": 1219
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.381533400219319e-06,
+      "loss": 0.0122,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.379850695043513e-06,
+      "loss": 0.0116,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.378165855181687e-06,
+      "loss": 0.0177,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.376478881455008e-06,
+      "loss": 0.016,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.37478977468569e-06,
+      "loss": 0.0156,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.37309853569698e-06,
+      "loss": 0.0144,
+      "step": 1225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.371405165313169e-06,
+      "loss": 0.0193,
+      "step": 1226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.369709664359585e-06,
+      "loss": 0.0112,
+      "step": 1227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.368012033662594e-06,
+      "loss": 0.0119,
+      "step": 1228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.366312274049602e-06,
+      "loss": 0.0095,
+      "step": 1229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.364610386349048e-06,
+      "loss": 0.0168,
+      "step": 1230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.362906371390416e-06,
+      "loss": 0.0175,
+      "step": 1231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.361200230004219e-06,
+      "loss": 0.0217,
+      "step": 1232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.35949196302201e-06,
+      "loss": 0.0166,
+      "step": 1233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.357781571276379e-06,
+      "loss": 0.0107,
+      "step": 1234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.356069055600949e-06,
+      "loss": 0.0159,
+      "step": 1235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.354354416830377e-06,
+      "loss": 0.0112,
+      "step": 1236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.352637655800362e-06,
+      "loss": 0.0127,
+      "step": 1237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.35091877334763e-06,
+      "loss": 0.0107,
+      "step": 1238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.349197770309942e-06,
+      "loss": 0.0154,
+      "step": 1239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.347474647526095e-06,
+      "loss": 0.0101,
+      "step": 1240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.34574940583592e-06,
+      "loss": 0.0149,
+      "step": 1241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.344022046080277e-06,
+      "loss": 0.0151,
+      "step": 1242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.342292569101061e-06,
+      "loss": 0.0136,
+      "step": 1243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.340560975741198e-06,
+      "loss": 0.022,
+      "step": 1244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.338827266844643e-06,
+      "loss": 0.0121,
+      "step": 1245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.337091443256388e-06,
+      "loss": 0.0141,
+      "step": 1246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.33535350582245e-06,
+      "loss": 0.0191,
+      "step": 1247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.333613455389883e-06,
+      "loss": 0.0112,
+      "step": 1248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.33187129280676e-06,
+      "loss": 0.0148,
+      "step": 1249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.330127018922195e-06,
+      "loss": 0.0151,
+      "step": 1250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.328380634586322e-06,
+      "loss": 0.0122,
+      "step": 1251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.326632140650311e-06,
+      "loss": 0.0146,
+      "step": 1252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.324881537966355e-06,
+      "loss": 0.0138,
+      "step": 1253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.323128827387675e-06,
+      "loss": 0.0118,
+      "step": 1254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.321374009768525e-06,
+      "loss": 0.015,
+      "step": 1255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.319617085964177e-06,
+      "loss": 0.0219,
+      "step": 1256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.317858056830938e-06,
+      "loss": 0.0199,
+      "step": 1257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.316096923226135e-06,
+      "loss": 0.0133,
+      "step": 1258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.314333686008125e-06,
+      "loss": 0.0128,
+      "step": 1259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.312568346036288e-06,
+      "loss": 0.0134,
+      "step": 1260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.31080090417103e-06,
+      "loss": 0.0169,
+      "step": 1261
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.309031361273775e-06,
+      "loss": 0.0117,
+      "step": 1262
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.307259718206984e-06,
+      "loss": 0.0094,
+      "step": 1263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.305485975834132e-06,
+      "loss": 0.01,
+      "step": 1264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.30371013501972e-06,
+      "loss": 0.0095,
+      "step": 1265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.301932196629267e-06,
+      "loss": 0.0132,
+      "step": 1266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.300152161529325e-06,
+      "loss": 0.0137,
+      "step": 1267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.298370030587456e-06,
+      "loss": 0.0199,
+      "step": 1268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.296585804672253e-06,
+      "loss": 0.01,
+      "step": 1269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.294799484653323e-06,
+      "loss": 0.0182,
+      "step": 1270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.293011071401299e-06,
+      "loss": 0.0132,
+      "step": 1271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.291220565787829e-06,
+      "loss": 0.0121,
+      "step": 1272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.289427968685588e-06,
+      "loss": 0.0095,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.287633280968263e-06,
+      "loss": 0.0109,
+      "step": 1274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.285836503510562e-06,
+      "loss": 0.0213,
+      "step": 1275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.284037637188215e-06,
+      "loss": 0.0224,
+      "step": 1276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.282236682877968e-06,
+      "loss": 0.0124,
+      "step": 1277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.280433641457582e-06,
+      "loss": 0.0107,
+      "step": 1278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.278628513805838e-06,
+      "loss": 0.0154,
+      "step": 1279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.276821300802535e-06,
+      "loss": 0.0123,
+      "step": 1280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.275012003328483e-06,
+      "loss": 0.0087,
+      "step": 1281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.273200622265516e-06,
+      "loss": 0.0128,
+      "step": 1282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.271387158496477e-06,
+      "loss": 0.0101,
+      "step": 1283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.269571612905227e-06,
+      "loss": 0.0137,
+      "step": 1284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.267753986376638e-06,
+      "loss": 0.0108,
+      "step": 1285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.265934279796602e-06,
+      "loss": 0.0107,
+      "step": 1286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.264112494052022e-06,
+      "loss": 0.009,
+      "step": 1287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.262288630030814e-06,
+      "loss": 0.0176,
+      "step": 1288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.260462688621906e-06,
+      "loss": 0.0093,
+      "step": 1289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.25863467071524e-06,
+      "loss": 0.0156,
+      "step": 1290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.256804577201768e-06,
+      "loss": 0.0133,
+      "step": 1291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.25497240897346e-06,
+      "loss": 0.0139,
+      "step": 1292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.25313816692329e-06,
+      "loss": 0.0119,
+      "step": 1293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.251301851945244e-06,
+      "loss": 0.0195,
+      "step": 1294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.24946346493432e-06,
+      "loss": 0.0149,
+      "step": 1295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.247623006786529e-06,
+      "loss": 0.0102,
+      "step": 1296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.245780478398883e-06,
+      "loss": 0.0145,
+      "step": 1297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.24393588066941e-06,
+      "loss": 0.0106,
+      "step": 1298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.242089214497146e-06,
+      "loss": 0.0104,
+      "step": 1299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.24024048078213e-06,
+      "loss": 0.0128,
+      "step": 1300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.238389680425417e-06,
+      "loss": 0.0153,
+      "step": 1301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.236536814329062e-06,
+      "loss": 0.0143,
+      "step": 1302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.234681883396129e-06,
+      "loss": 0.0212,
+      "step": 1303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.232824888530689e-06,
+      "loss": 0.0205,
+      "step": 1304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.230965830637821e-06,
+      "loss": 0.0108,
+      "step": 1305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.229104710623604e-06,
+      "loss": 0.0103,
+      "step": 1306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.227241529395127e-06,
+      "loss": 0.0151,
+      "step": 1307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.225376287860484e-06,
+      "loss": 0.0121,
+      "step": 1308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.223508986928766e-06,
+      "loss": 0.0132,
+      "step": 1309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.221639627510076e-06,
+      "loss": 0.0096,
+      "step": 1310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.219768210515518e-06,
+      "loss": 0.0319,
+      "step": 1311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.217894736857195e-06,
+      "loss": 0.0154,
+      "step": 1312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.216019207448216e-06,
+      "loss": 0.0139,
+      "step": 1313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.214141623202694e-06,
+      "loss": 0.0229,
+      "step": 1314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.21226198503574e-06,
+      "loss": 0.0136,
+      "step": 1315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.210380293863462e-06,
+      "loss": 0.0141,
+      "step": 1316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.208496550602979e-06,
+      "loss": 0.0127,
+      "step": 1317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.206610756172402e-06,
+      "loss": 0.0117,
+      "step": 1318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.204722911490847e-06,
+      "loss": 0.0137,
+      "step": 1319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.202833017478421e-06,
+      "loss": 0.0193,
+      "step": 1320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.200941075056242e-06,
+      "loss": 0.0109,
+      "step": 1321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.199047085146415e-06,
+      "loss": 0.0131,
+      "step": 1322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.197151048672051e-06,
+      "loss": 0.0354,
+      "step": 1323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.195252966557252e-06,
+      "loss": 0.0121,
+      "step": 1324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.193352839727122e-06,
+      "loss": 0.0114,
+      "step": 1325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.191450669107758e-06,
+      "loss": 0.0116,
+      "step": 1326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.189546455626258e-06,
+      "loss": 0.0116,
+      "step": 1327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.18764020021071e-06,
+      "loss": 0.0139,
+      "step": 1328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.1857319037902e-06,
+      "loss": 0.0156,
+      "step": 1329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.18382156729481e-06,
+      "loss": 0.0101,
+      "step": 1330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.181909191655613e-06,
+      "loss": 0.0142,
+      "step": 1331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.179994777804677e-06,
+      "loss": 0.012,
+      "step": 1332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.178078326675069e-06,
+      "loss": 0.0114,
+      "step": 1333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.176159839200838e-06,
+      "loss": 0.0098,
+      "step": 1334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.174239316317034e-06,
+      "loss": 0.0138,
+      "step": 1335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.172316758959695e-06,
+      "loss": 0.014,
+      "step": 1336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.170392168065858e-06,
+      "loss": 0.0354,
+      "step": 1337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.168465544573538e-06,
+      "loss": 0.0184,
+      "step": 1338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.16653688942175e-06,
+      "loss": 0.0127,
+      "step": 1339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.164606203550498e-06,
+      "loss": 0.0285,
+      "step": 1340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.162673487900775e-06,
+      "loss": 0.0125,
+      "step": 1341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.160738743414564e-06,
+      "loss": 0.0199,
+      "step": 1342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.158801971034832e-06,
+      "loss": 0.0166,
+      "step": 1343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.156863171705543e-06,
+      "loss": 0.0137,
+      "step": 1344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.154922346371641e-06,
+      "loss": 0.0762,
+      "step": 1345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.152979495979064e-06,
+      "loss": 0.0163,
+      "step": 1346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.15103462147473e-06,
+      "loss": 0.036,
+      "step": 1347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.14908772380655e-06,
+      "loss": 0.0302,
+      "step": 1348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.147138803923417e-06,
+      "loss": 0.0172,
+      "step": 1349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.145187862775208e-06,
+      "loss": 0.0251,
+      "step": 1350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.143234901312794e-06,
+      "loss": 0.0111,
+      "step": 1351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.141279920488021e-06,
+      "loss": 0.0141,
+      "step": 1352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.139322921253724e-06,
+      "loss": 0.0122,
+      "step": 1353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.13736390456372e-06,
+      "loss": 0.0193,
+      "step": 1354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.13540287137281e-06,
+      "loss": 0.0132,
+      "step": 1355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.133439822636779e-06,
+      "loss": 0.0146,
+      "step": 1356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.13147475931239e-06,
+      "loss": 0.0162,
+      "step": 1357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.129507682357393e-06,
+      "loss": 0.0107,
+      "step": 1358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.12753859273052e-06,
+      "loss": 0.0137,
+      "step": 1359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.125567491391476e-06,
+      "loss": 0.0145,
+      "step": 1360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.123594379300956e-06,
+      "loss": 0.0131,
+      "step": 1361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.12161925742063e-06,
+      "loss": 0.0109,
+      "step": 1362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.119642126713147e-06,
+      "loss": 0.026,
+      "step": 1363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.117662988142138e-06,
+      "loss": 0.0151,
+      "step": 1364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.115681842672211e-06,
+      "loss": 0.0178,
+      "step": 1365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.11369869126895e-06,
+      "loss": 0.0151,
+      "step": 1366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.111713534898923e-06,
+      "loss": 0.0139,
+      "step": 1367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.109726374529666e-06,
+      "loss": 0.0109,
+      "step": 1368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.107737211129702e-06,
+      "loss": 0.0127,
+      "step": 1369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.10574604566852e-06,
+      "loss": 0.0142,
+      "step": 1370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.103752879116595e-06,
+      "loss": 0.0104,
+      "step": 1371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.101757712445369e-06,
+      "loss": 0.013,
+      "step": 1372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.099760546627262e-06,
+      "loss": 0.0108,
+      "step": 1373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.09776138263567e-06,
+      "loss": 0.018,
+      "step": 1374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.09576022144496e-06,
+      "loss": 0.0183,
+      "step": 1375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.093757064030473e-06,
+      "loss": 0.0207,
+      "step": 1376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.091751911368524e-06,
+      "loss": 0.0205,
+      "step": 1377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.089744764436404e-06,
+      "loss": 0.0155,
+      "step": 1378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.087735624212365e-06,
+      "loss": 0.0132,
+      "step": 1379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.085724491675642e-06,
+      "loss": 0.0152,
+      "step": 1380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.083711367806438e-06,
+      "loss": 0.0549,
+      "step": 1381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.08169625358592e-06,
+      "loss": 0.0133,
+      "step": 1382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.079679149996235e-06,
+      "loss": 0.0129,
+      "step": 1383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.077660058020492e-06,
+      "loss": 0.0112,
+      "step": 1384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.07563897864277e-06,
+      "loss": 0.013,
+      "step": 1385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.073615912848126e-06,
+      "loss": 0.0179,
+      "step": 1386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.07159086162257e-06,
+      "loss": 0.0163,
+      "step": 1387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.069563825953092e-06,
+      "loss": 0.0176,
+      "step": 1388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.06753480682764e-06,
+      "loss": 0.0165,
+      "step": 1389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.065503805235139e-06,
+      "loss": 0.0181,
+      "step": 1390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.06347082216547e-06,
+      "loss": 0.0144,
+      "step": 1391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.061435858609486e-06,
+      "loss": 0.0137,
+      "step": 1392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.059398915559005e-06,
+      "loss": 0.0105,
+      "step": 1393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.057359994006806e-06,
+      "loss": 0.0276,
+      "step": 1394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.055319094946633e-06,
+      "loss": 0.0212,
+      "step": 1395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.0532762193732e-06,
+      "loss": 0.0121,
+      "step": 1396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.051231368282177e-06,
+      "loss": 0.0132,
+      "step": 1397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.0491845426702e-06,
+      "loss": 0.0172,
+      "step": 1398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.047135743534866e-06,
+      "loss": 0.0188,
+      "step": 1399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 0.0186,
+      "step": 1400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.043032228689333e-06,
+      "loss": 0.015,
+      "step": 1401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.040977514979136e-06,
+      "loss": 0.0171,
+      "step": 1402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.038920831745587e-06,
+      "loss": 0.0372,
+      "step": 1403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.036862179991092e-06,
+      "loss": 0.0214,
+      "step": 1404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.03480156071901e-06,
+      "loss": 0.0111,
+      "step": 1405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.032738974933663e-06,
+      "loss": 0.0116,
+      "step": 1406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.03067442364033e-06,
+      "loss": 0.0122,
+      "step": 1407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.028607907845247e-06,
+      "loss": 0.0145,
+      "step": 1408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.026539428555609e-06,
+      "loss": 0.0182,
+      "step": 1409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.02446898677957e-06,
+      "loss": 0.0096,
+      "step": 1410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.022396583526238e-06,
+      "loss": 0.0882,
+      "step": 1411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.020322219805674e-06,
+      "loss": 0.0155,
+      "step": 1412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.0182458966289e-06,
+      "loss": 0.0204,
+      "step": 1413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.01616761500789e-06,
+      "loss": 0.0274,
+      "step": 1414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.014087375955574e-06,
+      "loss": 0.0123,
+      "step": 1415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.012005180485834e-06,
+      "loss": 0.0263,
+      "step": 1416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.009921029613506e-06,
+      "loss": 0.0259,
+      "step": 1417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.007834924354384e-06,
+      "loss": 0.0127,
+      "step": 1418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.005746865725206e-06,
+      "loss": 0.0156,
+      "step": 1419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.003656854743667e-06,
+      "loss": 0.0124,
+      "step": 1420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.001564892428416e-06,
+      "loss": 0.0135,
+      "step": 1421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.999470979799048e-06,
+      "loss": 0.0288,
+      "step": 1422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.99737511787611e-06,
+      "loss": 0.0145,
+      "step": 1423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.9952773076811e-06,
+      "loss": 0.0275,
+      "step": 1424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.993177550236464e-06,
+      "loss": 0.0114,
+      "step": 1425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.991075846565603e-06,
+      "loss": 0.0306,
+      "step": 1426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.988972197692857e-06,
+      "loss": 0.0129,
+      "step": 1427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.986866604643518e-06,
+      "loss": 0.0141,
+      "step": 1428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.984759068443832e-06,
+      "loss": 0.0141,
+      "step": 1429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.982649590120982e-06,
+      "loss": 0.0229,
+      "step": 1430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.980538170703104e-06,
+      "loss": 0.0131,
+      "step": 1431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.978424811219277e-06,
+      "loss": 0.0333,
+      "step": 1432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.97630951269953e-06,
+      "loss": 0.017,
+      "step": 1433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.97419227617483e-06,
+      "loss": 0.0137,
+      "step": 1434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.972073102677091e-06,
+      "loss": 0.0105,
+      "step": 1435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.969951993239177e-06,
+      "loss": 0.0159,
+      "step": 1436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.96782894889489e-06,
+      "loss": 0.0218,
+      "step": 1437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.965703970678974e-06,
+      "loss": 0.0189,
+      "step": 1438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.963577059627117e-06,
+      "loss": 0.0137,
+      "step": 1439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.961448216775955e-06,
+      "loss": 0.017,
+      "step": 1440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.959317443163054e-06,
+      "loss": 0.0138,
+      "step": 1441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.957184739826929e-06,
+      "loss": 0.0148,
+      "step": 1442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.955050107807035e-06,
+      "loss": 0.0145,
+      "step": 1443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.952913548143766e-06,
+      "loss": 0.0163,
+      "step": 1444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.950775061878453e-06,
+      "loss": 0.0138,
+      "step": 1445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.94863465005337e-06,
+      "loss": 0.0185,
+      "step": 1446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.946492313711725e-06,
+      "loss": 0.0118,
+      "step": 1447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.944348053897672e-06,
+      "loss": 0.0155,
+      "step": 1448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.942201871656292e-06,
+      "loss": 0.0115,
+      "step": 1449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.94005376803361e-06,
+      "loss": 0.0309,
+      "step": 1450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.937903744076587e-06,
+      "loss": 0.0136,
+      "step": 1451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.935751800833117e-06,
+      "loss": 0.0186,
+      "step": 1452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.933597939352031e-06,
+      "loss": 0.0152,
+      "step": 1453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.931442160683094e-06,
+      "loss": 0.0099,
+      "step": 1454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.92928446587701e-06,
+      "loss": 0.0138,
+      "step": 1455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.92712485598541e-06,
+      "loss": 0.0153,
+      "step": 1456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.924963332060863e-06,
+      "loss": 0.0217,
+      "step": 1457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.922799895156868e-06,
+      "loss": 0.0151,
+      "step": 1458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.920634546327857e-06,
+      "loss": 0.0499,
+      "step": 1459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.9184672866292e-06,
+      "loss": 0.0239,
+      "step": 1460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.916298117117188e-06,
+      "loss": 0.0187,
+      "step": 1461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.91412703884905e-06,
+      "loss": 0.0414,
+      "step": 1462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.911954052882941e-06,
+      "loss": 0.0121,
+      "step": 1463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.909779160277951e-06,
+      "loss": 0.0208,
+      "step": 1464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.907602362094094e-06,
+      "loss": 0.0132,
+      "step": 1465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.905423659392316e-06,
+      "loss": 0.0269,
+      "step": 1466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.903243053234492e-06,
+      "loss": 0.0136,
+      "step": 1467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.90106054468342e-06,
+      "loss": 0.0131,
+      "step": 1468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.898876134802827e-06,
+      "loss": 0.0134,
+      "step": 1469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.896689824657371e-06,
+      "loss": 0.0357,
+      "step": 1470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.894501615312633e-06,
+      "loss": 0.0154,
+      "step": 1471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.892311507835118e-06,
+      "loss": 0.0466,
+      "step": 1472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.890119503292258e-06,
+      "loss": 0.0184,
+      "step": 1473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.887925602752411e-06,
+      "loss": 0.0176,
+      "step": 1474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.885729807284855e-06,
+      "loss": 0.017,
+      "step": 1475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.883532117959797e-06,
+      "loss": 0.0113,
+      "step": 1476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.88133253584836e-06,
+      "loss": 0.0134,
+      "step": 1477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.879131062022598e-06,
+      "loss": 0.0111,
+      "step": 1478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.87692769755548e-06,
+      "loss": 0.0136,
+      "step": 1479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.874722443520898e-06,
+      "loss": 0.0183,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.872515300993669e-06,
+      "loss": 0.0122,
+      "step": 1481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.870306271049527e-06,
+      "loss": 0.0152,
+      "step": 1482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.868095354765125e-06,
+      "loss": 0.0147,
+      "step": 1483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.865882553218036e-06,
+      "loss": 0.0152,
+      "step": 1484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.863667867486756e-06,
+      "loss": 0.0139,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.861451298650692e-06,
+      "loss": 0.0135,
+      "step": 1486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.859232847790175e-06,
+      "loss": 0.0142,
+      "step": 1487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.857012515986452e-06,
+      "loss": 0.0129,
+      "step": 1488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.854790304321682e-06,
+      "loss": 0.0135,
+      "step": 1489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.852566213878947e-06,
+      "loss": 0.0104,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.85034024574224e-06,
+      "loss": 0.0151,
+      "step": 1491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.848112400996473e-06,
+      "loss": 0.0176,
+      "step": 1492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.84588268072747e-06,
+      "loss": 0.0118,
+      "step": 1493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.843651086021966e-06,
+      "loss": 0.0159,
+      "step": 1494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.841417617967618e-06,
+      "loss": 0.0981,
+      "step": 1495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.83918227765299e-06,
+      "loss": 0.0497,
+      "step": 1496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.836945066167556e-06,
+      "loss": 0.0352,
+      "step": 1497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.834705984601708e-06,
+      "loss": 0.0153,
+      "step": 1498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.83246503404675e-06,
+      "loss": 0.0122,
+      "step": 1499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.83022221559489e-06,
+      "loss": 0.0124,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.827977530339254e-06,
+      "loss": 0.0164,
+      "step": 1501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.825730979373873e-06,
+      "loss": 0.0216,
+      "step": 1502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.823482563793687e-06,
+      "loss": 0.0154,
+      "step": 1503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.821232284694545e-06,
+      "loss": 0.0149,
+      "step": 1504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.818980143173212e-06,
+      "loss": 0.0194,
+      "step": 1505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.81672614032735e-06,
+      "loss": 0.0163,
+      "step": 1506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.814470277255532e-06,
+      "loss": 0.0268,
+      "step": 1507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.81221255505724e-06,
+      "loss": 0.018,
+      "step": 1508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.80995297483286e-06,
+      "loss": 0.0184,
+      "step": 1509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.807691537683685e-06,
+      "loss": 0.0289,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.80542824471191e-06,
+      "loss": 0.0201,
+      "step": 1511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.803163097020637e-06,
+      "loss": 0.0132,
+      "step": 1512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.80089609571387e-06,
+      "loss": 0.0156,
+      "step": 1513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.798627241896524e-06,
+      "loss": 0.024,
+      "step": 1514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.796356536674404e-06,
+      "loss": 0.0152,
+      "step": 1515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.794083981154229e-06,
+      "loss": 0.0209,
+      "step": 1516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.791809576443611e-06,
+      "loss": 0.0132,
+      "step": 1517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.789533323651067e-06,
+      "loss": 0.0136,
+      "step": 1518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.78725522388602e-06,
+      "loss": 0.0137,
+      "step": 1519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.784975278258783e-06,
+      "loss": 0.0176,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.782693487880575e-06,
+      "loss": 0.0164,
+      "step": 1521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.780409853863517e-06,
+      "loss": 0.018,
+      "step": 1522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.778124377320619e-06,
+      "loss": 0.0275,
+      "step": 1523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.775837059365796e-06,
+      "loss": 0.0177,
+      "step": 1524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.773547901113862e-06,
+      "loss": 0.0215,
+      "step": 1525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.77125690368052e-06,
+      "loss": 0.0141,
+      "step": 1526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.768964068182378e-06,
+      "loss": 0.0338,
+      "step": 1527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.766669395736936e-06,
+      "loss": 0.0139,
+      "step": 1528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.764372887462587e-06,
+      "loss": 0.0178,
+      "step": 1529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.762074544478622e-06,
+      "loss": 0.0155,
+      "step": 1530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.759774367905228e-06,
+      "loss": 0.0125,
+      "step": 1531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.757472358863481e-06,
+      "loss": 0.0138,
+      "step": 1532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.755168518475351e-06,
+      "loss": 0.0209,
+      "step": 1533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.752862847863707e-06,
+      "loss": 0.013,
+      "step": 1534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.750555348152299e-06,
+      "loss": 0.0153,
+      "step": 1535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.748246020465776e-06,
+      "loss": 0.0217,
+      "step": 1536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.745934865929676e-06,
+      "loss": 0.0145,
+      "step": 1537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.743621885670431e-06,
+      "loss": 0.0149,
+      "step": 1538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.741307080815357e-06,
+      "loss": 0.0226,
+      "step": 1539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.73899045249266e-06,
+      "loss": 0.011,
+      "step": 1540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.736672001831438e-06,
+      "loss": 0.0266,
+      "step": 1541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.73435172996168e-06,
+      "loss": 0.0303,
+      "step": 1542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.732029638014249e-06,
+      "loss": 0.0168,
+      "step": 1543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.729705727120911e-06,
+      "loss": 0.0202,
+      "step": 1544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.727379998414311e-06,
+      "loss": 0.0115,
+      "step": 1545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.725052453027982e-06,
+      "loss": 0.0225,
+      "step": 1546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.722723092096337e-06,
+      "loss": 0.0136,
+      "step": 1547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.720391916754683e-06,
+      "loss": 0.0172,
+      "step": 1548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.718058928139205e-06,
+      "loss": 0.0147,
+      "step": 1549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.715724127386971e-06,
+      "loss": 0.0137,
+      "step": 1550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.713387515635938e-06,
+      "loss": 0.0168,
+      "step": 1551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.711049094024942e-06,
+      "loss": 0.021,
+      "step": 1552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.708708863693696e-06,
+      "loss": 0.0114,
+      "step": 1553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.706366825782805e-06,
+      "loss": 0.0136,
+      "step": 1554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.70402298143375e-06,
+      "loss": 0.0178,
+      "step": 1555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.701677331788891e-06,
+      "loss": 0.0184,
+      "step": 1556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.699329877991469e-06,
+      "loss": 0.0128,
+      "step": 1557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.696980621185602e-06,
+      "loss": 0.0093,
+      "step": 1558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.694629562516295e-06,
+      "loss": 0.0205,
+      "step": 1559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.692276703129421e-06,
+      "loss": 0.0151,
+      "step": 1560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.689922044171735e-06,
+      "loss": 0.0252,
+      "step": 1561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.68756558679087e-06,
+      "loss": 0.0281,
+      "step": 1562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.685207332135337e-06,
+      "loss": 0.0113,
+      "step": 1563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.682847281354517e-06,
+      "loss": 0.0113,
+      "step": 1564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.680485435598674e-06,
+      "loss": 0.0231,
+      "step": 1565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.678121796018938e-06,
+      "loss": 0.0175,
+      "step": 1566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.675756363767322e-06,
+      "loss": 0.0137,
+      "step": 1567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.673389139996708e-06,
+      "loss": 0.0968,
+      "step": 1568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.671020125860851e-06,
+      "loss": 0.024,
+      "step": 1569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.668649322514382e-06,
+      "loss": 0.0145,
+      "step": 1570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.666276731112802e-06,
+      "loss": 0.0161,
+      "step": 1571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.66390235281248e-06,
+      "loss": 0.0159,
+      "step": 1572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.66152618877066e-06,
+      "loss": 0.0143,
+      "step": 1573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.659148240145456e-06,
+      "loss": 0.0123,
+      "step": 1574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.656768508095853e-06,
+      "loss": 0.0234,
+      "step": 1575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.654386993781703e-06,
+      "loss": 0.011,
+      "step": 1576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.652003698363724e-06,
+      "loss": 0.0122,
+      "step": 1577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.649618623003509e-06,
+      "loss": 0.0134,
+      "step": 1578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.647231768863513e-06,
+      "loss": 0.0119,
+      "step": 1579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.644843137107058e-06,
+      "loss": 0.0112,
+      "step": 1580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.642452728898339e-06,
+      "loss": 0.0116,
+      "step": 1581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.640060545402407e-06,
+      "loss": 0.0134,
+      "step": 1582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.637666587785185e-06,
+      "loss": 0.015,
+      "step": 1583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.63527085721346e-06,
+      "loss": 0.0182,
+      "step": 1584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.632873354854881e-06,
+      "loss": 0.0145,
+      "step": 1585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.630474081877959e-06,
+      "loss": 0.0106,
+      "step": 1586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.628073039452076e-06,
+      "loss": 0.0358,
+      "step": 1587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.625670228747467e-06,
+      "loss": 0.0637,
+      "step": 1588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.623265650935233e-06,
+      "loss": 0.0159,
+      "step": 1589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.620859307187339e-06,
+      "loss": 0.0137,
+      "step": 1590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.618451198676602e-06,
+      "loss": 0.0126,
+      "step": 1591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.616041326576711e-06,
+      "loss": 0.0202,
+      "step": 1592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.613629692062204e-06,
+      "loss": 0.0287,
+      "step": 1593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.611216296308485e-06,
+      "loss": 0.012,
+      "step": 1594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.608801140491811e-06,
+      "loss": 0.0327,
+      "step": 1595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.606384225789304e-06,
+      "loss": 0.0135,
+      "step": 1596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.603965553378934e-06,
+      "loss": 0.0143,
+      "step": 1597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.601545124439535e-06,
+      "loss": 0.0147,
+      "step": 1598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.599122940150795e-06,
+      "loss": 0.0134,
+      "step": 1599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.596699001693257e-06,
+      "loss": 0.018,
+      "step": 1600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.594273310248317e-06,
+      "loss": 0.0123,
+      "step": 1601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.591845866998231e-06,
+      "loss": 0.0154,
+      "step": 1602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.589416673126104e-06,
+      "loss": 0.0122,
+      "step": 1603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.586985729815895e-06,
+      "loss": 0.0356,
+      "step": 1604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.584553038252415e-06,
+      "loss": 0.0699,
+      "step": 1605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.58211859962133e-06,
+      "loss": 0.0363,
+      "step": 1606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.579682415109156e-06,
+      "loss": 0.0132,
+      "step": 1607
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.57724448590326e-06,
+      "loss": 0.0139,
+      "step": 1608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.574804813191859e-06,
+      "loss": 0.0141,
+      "step": 1609
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.572363398164017e-06,
+      "loss": 0.0148,
+      "step": 1610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.569920242009655e-06,
+      "loss": 0.0285,
+      "step": 1611
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.567475345919532e-06,
+      "loss": 0.0116,
+      "step": 1612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.565028711085266e-06,
+      "loss": 0.012,
+      "step": 1613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.562580338699313e-06,
+      "loss": 0.0196,
+      "step": 1614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.560130229954985e-06,
+      "loss": 0.0107,
+      "step": 1615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.557678386046429e-06,
+      "loss": 0.0163,
+      "step": 1616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.555224808168644e-06,
+      "loss": 0.0221,
+      "step": 1617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.55276949751748e-06,
+      "loss": 0.0286,
+      "step": 1618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.550312455289624e-06,
+      "loss": 0.0163,
+      "step": 1619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.547853682682605e-06,
+      "loss": 0.0136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.545393180894801e-06,
+      "loss": 0.0166,
+      "step": 1621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.542930951125432e-06,
+      "loss": 0.0183,
+      "step": 1622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.540466994574556e-06,
+      "loss": 0.0234,
+      "step": 1623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.538001312443078e-06,
+      "loss": 0.0127,
+      "step": 1624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 0.019,
+      "step": 1625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.533064776246126e-06,
+      "loss": 0.0137,
+      "step": 1626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.530593924586659e-06,
+      "loss": 0.0137,
+      "step": 1627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.528121352158604e-06,
+      "loss": 0.0184,
+      "step": 1628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.525647060167063e-06,
+      "loss": 0.0157,
+      "step": 1629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.523171049817974e-06,
+      "loss": 0.0138,
+      "step": 1630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.520693322318116e-06,
+      "loss": 0.0177,
+      "step": 1631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.518213878875103e-06,
+      "loss": 0.0355,
+      "step": 1632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.515732720697383e-06,
+      "loss": 0.0166,
+      "step": 1633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.513249848994248e-06,
+      "loss": 0.016,
+      "step": 1634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.510765264975813e-06,
+      "loss": 0.0117,
+      "step": 1635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.508278969853037e-06,
+      "loss": 0.0155,
+      "step": 1636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.505790964837712e-06,
+      "loss": 0.0132,
+      "step": 1637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.50330125114246e-06,
+      "loss": 0.0117,
+      "step": 1638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.500809829980734e-06,
+      "loss": 0.0121,
+      "step": 1639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.498316702566828e-06,
+      "loss": 0.0151,
+      "step": 1640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.495821870115857e-06,
+      "loss": 0.0125,
+      "step": 1641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.493325333843776e-06,
+      "loss": 0.0193,
+      "step": 1642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.490827094967364e-06,
+      "loss": 0.0527,
+      "step": 1643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.488327154704232e-06,
+      "loss": 0.0141,
+      "step": 1644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.485825514272824e-06,
+      "loss": 0.0138,
+      "step": 1645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.483322174892404e-06,
+      "loss": 0.0134,
+      "step": 1646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.480817137783073e-06,
+      "loss": 0.0117,
+      "step": 1647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.478310404165756e-06,
+      "loss": 0.0203,
+      "step": 1648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.4758019752622e-06,
+      "loss": 0.0113,
+      "step": 1649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.473291852294986e-06,
+      "loss": 0.0155,
+      "step": 1650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.47078003648752e-06,
+      "loss": 0.0207,
+      "step": 1651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.468266529064025e-06,
+      "loss": 0.0139,
+      "step": 1652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.465751331249558e-06,
+      "loss": 0.0145,
+      "step": 1653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.463234444269994e-06,
+      "loss": 0.0193,
+      "step": 1654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.460715869352035e-06,
+      "loss": 0.0181,
+      "step": 1655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.458195607723201e-06,
+      "loss": 0.0549,
+      "step": 1656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.45567366061184e-06,
+      "loss": 0.0183,
+      "step": 1657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.453150029247115e-06,
+      "loss": 0.0157,
+      "step": 1658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.450624714859016e-06,
+      "loss": 0.0515,
+      "step": 1659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.44809771867835e-06,
+      "loss": 0.0136,
+      "step": 1660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.445569041936743e-06,
+      "loss": 0.0121,
+      "step": 1661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.443038685866643e-06,
+      "loss": 0.0178,
+      "step": 1662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.440506651701315e-06,
+      "loss": 0.0133,
+      "step": 1663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.437972940674838e-06,
+      "loss": 0.0176,
+      "step": 1664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.435437554022116e-06,
+      "loss": 0.0241,
+      "step": 1665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.432900492978864e-06,
+      "loss": 0.0132,
+      "step": 1666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.430361758781616e-06,
+      "loss": 0.0133,
+      "step": 1667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.427821352667719e-06,
+      "loss": 0.0316,
+      "step": 1668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.425279275875336e-06,
+      "loss": 0.0159,
+      "step": 1669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.422735529643445e-06,
+      "loss": 0.0161,
+      "step": 1670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.420190115211835e-06,
+      "loss": 0.0164,
+      "step": 1671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.417643033821114e-06,
+      "loss": 0.0155,
+      "step": 1672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.415094286712694e-06,
+      "loss": 0.0156,
+      "step": 1673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.412543875128809e-06,
+      "loss": 0.0113,
+      "step": 1674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.409991800312493e-06,
+      "loss": 0.0133,
+      "step": 1675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.4074380635076e-06,
+      "loss": 0.0197,
+      "step": 1676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.404882665958788e-06,
+      "loss": 0.0315,
+      "step": 1677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.402325608911527e-06,
+      "loss": 0.0137,
+      "step": 1678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.399766893612096e-06,
+      "loss": 0.0144,
+      "step": 1679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.397206521307584e-06,
+      "loss": 0.0149,
+      "step": 1680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.394644493245882e-06,
+      "loss": 0.013,
+      "step": 1681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.392080810675692e-06,
+      "loss": 0.0126,
+      "step": 1682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.389515474846522e-06,
+      "loss": 0.0142,
+      "step": 1683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.386948487008687e-06,
+      "loss": 0.0176,
+      "step": 1684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.384379848413304e-06,
+      "loss": 0.0525,
+      "step": 1685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.381809560312298e-06,
+      "loss": 0.0188,
+      "step": 1686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.379237623958393e-06,
+      "loss": 0.0467,
+      "step": 1687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.376664040605122e-06,
+      "loss": 0.0321,
+      "step": 1688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.374088811506819e-06,
+      "loss": 0.0208,
+      "step": 1689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.371511937918616e-06,
+      "loss": 0.0136,
+      "step": 1690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.368933421096454e-06,
+      "loss": 0.0161,
+      "step": 1691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.366353262297069e-06,
+      "loss": 0.0165,
+      "step": 1692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.363771462778e-06,
+      "loss": 0.0146,
+      "step": 1693
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.361188023797581e-06,
+      "loss": 0.014,
+      "step": 1694
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.358602946614952e-06,
+      "loss": 0.0139,
+      "step": 1695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.356016232490047e-06,
+      "loss": 0.0151,
+      "step": 1696
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.353427882683601e-06,
+      "loss": 0.0184,
+      "step": 1697
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.350837898457142e-06,
+      "loss": 0.0129,
+      "step": 1698
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.348246281072998e-06,
+      "loss": 0.0129,
+      "step": 1699
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.345653031794292e-06,
+      "loss": 0.0211,
+      "step": 1700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.343058151884942e-06,
+      "loss": 0.0121,
+      "step": 1701
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.34046164260966e-06,
+      "loss": 0.0125,
+      "step": 1702
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.337863505233954e-06,
+      "loss": 0.0142,
+      "step": 1703
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.335263741024123e-06,
+      "loss": 0.0132,
+      "step": 1704
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.332662351247262e-06,
+      "loss": 0.0133,
+      "step": 1705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.33005933717126e-06,
+      "loss": 0.0124,
+      "step": 1706
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.327454700064788e-06,
+      "loss": 0.0116,
+      "step": 1707
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.324848441197317e-06,
+      "loss": 0.0115,
+      "step": 1708
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.32224056183911e-06,
+      "loss": 0.0168,
+      "step": 1709
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.319631063261209e-06,
+      "loss": 0.0109,
+      "step": 1710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.317019946735456e-06,
+      "loss": 0.0123,
+      "step": 1711
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.314407213534477e-06,
+      "loss": 0.0125,
+      "step": 1712
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.311792864931686e-06,
+      "loss": 0.0154,
+      "step": 1713
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.309176902201283e-06,
+      "loss": 0.0192,
+      "step": 1714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.30655932661826e-06,
+      "loss": 0.0159,
+      "step": 1715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.303940139458389e-06,
+      "loss": 0.0114,
+      "step": 1716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.301319341998231e-06,
+      "loss": 0.0157,
+      "step": 1717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.298696935515132e-06,
+      "loss": 0.0168,
+      "step": 1718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.296072921287217e-06,
+      "loss": 0.011,
+      "step": 1719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.293447300593402e-06,
+      "loss": 0.0158,
+      "step": 1720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.290820074713383e-06,
+      "loss": 0.0133,
+      "step": 1721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.288191244927637e-06,
+      "loss": 0.0207,
+      "step": 1722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.285560812517423e-06,
+      "loss": 0.0153,
+      "step": 1723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.282928778764783e-06,
+      "loss": 0.013,
+      "step": 1724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.280295144952537e-06,
+      "loss": 0.0294,
+      "step": 1725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.277659912364288e-06,
+      "loss": 0.01,
+      "step": 1726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.275023082284413e-06,
+      "loss": 0.0134,
+      "step": 1727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.272384655998075e-06,
+      "loss": 0.0107,
+      "step": 1728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.269744634791207e-06,
+      "loss": 0.0113,
+      "step": 1729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.267103019950529e-06,
+      "loss": 0.0144,
+      "step": 1730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.264459812763525e-06,
+      "loss": 0.0139,
+      "step": 1731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.261815014518465e-06,
+      "loss": 0.0112,
+      "step": 1732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.259168626504395e-06,
+      "loss": 0.0181,
+      "step": 1733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.256520650011126e-06,
+      "loss": 0.2287,
+      "step": 1734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.253871086329255e-06,
+      "loss": 0.1973,
+      "step": 1735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.251219936750145e-06,
+      "loss": 0.0143,
+      "step": 1736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.248567202565934e-06,
+      "loss": 0.0149,
+      "step": 1737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.24591288506953e-06,
+      "loss": 0.0147,
+      "step": 1738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.243256985554622e-06,
+      "loss": 0.0226,
+      "step": 1739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.240599505315656e-06,
+      "loss": 0.015,
+      "step": 1740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.237940445647858e-06,
+      "loss": 0.0183,
+      "step": 1741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.235279807847223e-06,
+      "loss": 0.0215,
+      "step": 1742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.232617593210512e-06,
+      "loss": 0.0152,
+      "step": 1743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.229953803035256e-06,
+      "loss": 0.0136,
+      "step": 1744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.227288438619754e-06,
+      "loss": 0.0112,
+      "step": 1745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.224621501263073e-06,
+      "loss": 0.0147,
+      "step": 1746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.221952992265046e-06,
+      "loss": 0.0117,
+      "step": 1747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.21928291292627e-06,
+      "loss": 0.0093,
+      "step": 1748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.21661126454811e-06,
+      "loss": 0.0124,
+      "step": 1749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.213938048432697e-06,
+      "loss": 0.0227,
+      "step": 1750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.211263265882923e-06,
+      "loss": 0.0136,
+      "step": 1751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.208586918202444e-06,
+      "loss": 0.0129,
+      "step": 1752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.205909006695679e-06,
+      "loss": 0.0134,
+      "step": 1753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.203229532667808e-06,
+      "loss": 0.0115,
+      "step": 1754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.200548497424779e-06,
+      "loss": 0.0148,
+      "step": 1755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.197865902273291e-06,
+      "loss": 0.0178,
+      "step": 1756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.19518174852081e-06,
+      "loss": 0.0187,
+      "step": 1757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.192496037475562e-06,
+      "loss": 0.0115,
+      "step": 1758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.189808770446528e-06,
+      "loss": 0.0297,
+      "step": 1759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.18711994874345e-06,
+      "loss": 0.0164,
+      "step": 1760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.184429573676825e-06,
+      "loss": 0.0138,
+      "step": 1761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.181737646557912e-06,
+      "loss": 0.016,
+      "step": 1762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.179044168698722e-06,
+      "loss": 0.0186,
+      "step": 1763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.176349141412022e-06,
+      "loss": 0.0122,
+      "step": 1764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.173652566011339e-06,
+      "loss": 0.0155,
+      "step": 1765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.170954443810947e-06,
+      "loss": 0.0118,
+      "step": 1766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.168254776125883e-06,
+      "loss": 0.012,
+      "step": 1767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.165553564271928e-06,
+      "loss": 0.0118,
+      "step": 1768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.162850809565623e-06,
+      "loss": 0.0177,
+      "step": 1769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.160146513324256e-06,
+      "loss": 0.0128,
+      "step": 1770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.157440676865866e-06,
+      "loss": 0.0139,
+      "step": 1771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.154733301509249e-06,
+      "loss": 0.0142,
+      "step": 1772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.152024388573945e-06,
+      "loss": 0.012,
+      "step": 1773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.149313939380244e-06,
+      "loss": 0.0147,
+      "step": 1774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.146601955249187e-06,
+      "loss": 0.0141,
+      "step": 1775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.143888437502565e-06,
+      "loss": 0.0111,
+      "step": 1776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.141173387462908e-06,
+      "loss": 0.0126,
+      "step": 1777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.138456806453503e-06,
+      "loss": 0.0128,
+      "step": 1778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.135738695798377e-06,
+      "loss": 0.0164,
+      "step": 1779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.133019056822303e-06,
+      "loss": 0.017,
+      "step": 1780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.130297890850803e-06,
+      "loss": 0.0117,
+      "step": 1781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.127575199210136e-06,
+      "loss": 0.0116,
+      "step": 1782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.124850983227313e-06,
+      "loss": 0.0143,
+      "step": 1783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.12212524423008e-06,
+      "loss": 0.0131,
+      "step": 1784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.119397983546932e-06,
+      "loss": 0.0138,
+      "step": 1785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.116669202507102e-06,
+      "loss": 0.0133,
+      "step": 1786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.113938902440563e-06,
+      "loss": 0.0155,
+      "step": 1787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.111207084678033e-06,
+      "loss": 0.019,
+      "step": 1788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.108473750550965e-06,
+      "loss": 0.0122,
+      "step": 1789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.105738901391553e-06,
+      "loss": 0.0104,
+      "step": 1790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.103002538532729e-06,
+      "loss": 0.0319,
+      "step": 1791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.100264663308165e-06,
+      "loss": 0.0129,
+      "step": 1792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.097525277052265e-06,
+      "loss": 0.0118,
+      "step": 1793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.094784381100174e-06,
+      "loss": 0.0148,
+      "step": 1794
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.092041976787772e-06,
+      "loss": 0.0125,
+      "step": 1795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.089298065451673e-06,
+      "loss": 0.0157,
+      "step": 1796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.086552648429225e-06,
+      "loss": 0.015,
+      "step": 1797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.083805727058514e-06,
+      "loss": 0.0139,
+      "step": 1798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.081057302678352e-06,
+      "loss": 0.013,
+      "step": 1799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.078307376628292e-06,
+      "loss": 0.0181,
+      "step": 1800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.075555950248613e-06,
+      "loss": 0.0164,
+      "step": 1801
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.072803024880322e-06,
+      "loss": 0.0142,
+      "step": 1802
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.07004860186517e-06,
+      "loss": 0.0168,
+      "step": 1803
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.067292682545622e-06,
+      "loss": 0.0165,
+      "step": 1804
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.064535268264883e-06,
+      "loss": 0.0259,
+      "step": 1805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.061776360366883e-06,
+      "loss": 0.0149,
+      "step": 1806
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.05901596019628e-06,
+      "loss": 0.0169,
+      "step": 1807
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.05625406909846e-06,
+      "loss": 0.0138,
+      "step": 1808
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.053490688419532e-06,
+      "loss": 0.0234,
+      "step": 1809
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.05072581950634e-06,
+      "loss": 0.0156,
+      "step": 1810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.047959463706441e-06,
+      "loss": 0.0179,
+      "step": 1811
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.045191622368128e-06,
+      "loss": 0.0113,
+      "step": 1812
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.04242229684041e-06,
+      "loss": 0.0115,
+      "step": 1813
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.039651488473028e-06,
+      "loss": 0.0126,
+      "step": 1814
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.036879198616434e-06,
+      "loss": 0.0161,
+      "step": 1815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.034105428621812e-06,
+      "loss": 0.0117,
+      "step": 1816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.031330179841062e-06,
+      "loss": 0.0154,
+      "step": 1817
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.028553453626809e-06,
+      "loss": 0.0128,
+      "step": 1818
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.02577525133239e-06,
+      "loss": 0.0149,
+      "step": 1819
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.022995574311876e-06,
+      "loss": 0.018,
+      "step": 1820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.020214423920039e-06,
+      "loss": 0.0162,
+      "step": 1821
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.017431801512384e-06,
+      "loss": 0.0189,
+      "step": 1822
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.014647708445124e-06,
+      "loss": 0.0111,
+      "step": 1823
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.011862146075194e-06,
+      "loss": 0.0111,
+      "step": 1824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.009075115760243e-06,
+      "loss": 0.0481,
+      "step": 1825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.006286618858634e-06,
+      "loss": 0.0143,
+      "step": 1826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.003496656729448e-06,
+      "loss": 0.0149,
+      "step": 1827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.000705230732478e-06,
+      "loss": 0.0115,
+      "step": 1828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.997912342228232e-06,
+      "loss": 0.0093,
+      "step": 1829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.99511799257793e-06,
+      "loss": 0.0188,
+      "step": 1830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.992322183143504e-06,
+      "loss": 0.0096,
+      "step": 1831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.989524915287595e-06,
+      "loss": 0.0078,
+      "step": 1832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.986726190373562e-06,
+      "loss": 0.0109,
+      "step": 1833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.983926009765464e-06,
+      "loss": 0.0124,
+      "step": 1834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.981124374828079e-06,
+      "loss": 0.3054,
+      "step": 1835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.978321286926892e-06,
+      "loss": 0.011,
+      "step": 1836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.975516747428087e-06,
+      "loss": 0.0139,
+      "step": 1837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.972710757698567e-06,
+      "loss": 0.0203,
+      "step": 1838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.969903319105935e-06,
+      "loss": 0.0131,
+      "step": 1839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.967094433018508e-06,
+      "loss": 0.0924,
+      "step": 1840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.964284100805297e-06,
+      "loss": 0.0525,
+      "step": 1841
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.961472323836025e-06,
+      "loss": 0.0223,
+      "step": 1842
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.95865910348112e-06,
+      "loss": 0.0356,
+      "step": 1843
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.95584444111171e-06,
+      "loss": 0.0124,
+      "step": 1844
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.953028338099628e-06,
+      "loss": 0.0169,
+      "step": 1845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.950210795817406e-06,
+      "loss": 0.0088,
+      "step": 1846
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.947391815638284e-06,
+      "loss": 0.0141,
+      "step": 1847
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.944571398936193e-06,
+      "loss": 0.0153,
+      "step": 1848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.941749547085778e-06,
+      "loss": 0.0234,
+      "step": 1849
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.938926261462366e-06,
+      "loss": 0.0178,
+      "step": 1850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.936101543441998e-06,
+      "loss": 0.0135,
+      "step": 1851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.933275394401407e-06,
+      "loss": 0.0142,
+      "step": 1852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.930447815718022e-06,
+      "loss": 0.0138,
+      "step": 1853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.927618808769971e-06,
+      "loss": 0.0111,
+      "step": 1854
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.92478837493608e-06,
+      "loss": 0.0169,
+      "step": 1855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.921956515595861e-06,
+      "loss": 0.013,
+      "step": 1856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.919123232129535e-06,
+      "loss": 0.0444,
+      "step": 1857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.916288525918008e-06,
+      "loss": 0.0453,
+      "step": 1858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.913452398342882e-06,
+      "loss": 0.0146,
+      "step": 1859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.910614850786448e-06,
+      "loss": 0.0125,
+      "step": 1860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.907775884631694e-06,
+      "loss": 0.0181,
+      "step": 1861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.904935501262301e-06,
+      "loss": 0.0135,
+      "step": 1862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.90209370206263e-06,
+      "loss": 0.0188,
+      "step": 1863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.899250488417746e-06,
+      "loss": 0.0161,
+      "step": 1864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.896405861713393e-06,
+      "loss": 0.0129,
+      "step": 1865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.893559823336013e-06,
+      "loss": 0.0113,
+      "step": 1866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.890712374672724e-06,
+      "loss": 0.0171,
+      "step": 1867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.887863517111337e-06,
+      "loss": 0.0129,
+      "step": 1868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.88501325204036e-06,
+      "loss": 0.0162,
+      "step": 1869
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.882161580848966e-06,
+      "loss": 0.0239,
+      "step": 1870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.879308504927034e-06,
+      "loss": 0.0424,
+      "step": 1871
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.876454025665114e-06,
+      "loss": 0.0132,
+      "step": 1872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.873598144454444e-06,
+      "loss": 0.0123,
+      "step": 1873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.87074086268695e-06,
+      "loss": 0.0099,
+      "step": 1874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.86788218175523e-06,
+      "loss": 0.0112,
+      "step": 1875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.865022103052578e-06,
+      "loss": 0.0124,
+      "step": 1876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.862160627972956e-06,
+      "loss": 0.0134,
+      "step": 1877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.859297757911013e-06,
+      "loss": 0.0185,
+      "step": 1878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.856433494262078e-06,
+      "loss": 0.0141,
+      "step": 1879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.85356783842216e-06,
+      "loss": 0.0118,
+      "step": 1880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.850700791787941e-06,
+      "loss": 0.0288,
+      "step": 1881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.847832355756788e-06,
+      "loss": 0.0172,
+      "step": 1882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.844962531726742e-06,
+      "loss": 0.0121,
+      "step": 1883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.842091321096515e-06,
+      "loss": 0.0117,
+      "step": 1884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.839218725265507e-06,
+      "loss": 0.0171,
+      "step": 1885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.836344745633785e-06,
+      "loss": 0.0167,
+      "step": 1886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.833469383602086e-06,
+      "loss": 0.0258,
+      "step": 1887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.830592640571833e-06,
+      "loss": 0.0151,
+      "step": 1888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.827714517945116e-06,
+      "loss": 0.0144,
+      "step": 1889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.82483501712469e-06,
+      "loss": 0.0139,
+      "step": 1890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.821954139513997e-06,
+      "loss": 0.0155,
+      "step": 1891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.819071886517134e-06,
+      "loss": 0.0135,
+      "step": 1892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.816188259538885e-06,
+      "loss": 0.0113,
+      "step": 1893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.813303259984685e-06,
+      "loss": 0.0218,
+      "step": 1894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.810416889260653e-06,
+      "loss": 0.0111,
+      "step": 1895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.807529148773572e-06,
+      "loss": 0.0123,
+      "step": 1896
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.80464003993089e-06,
+      "loss": 0.0282,
+      "step": 1897
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.801749564140724e-06,
+      "loss": 0.0215,
+      "step": 1898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.798857722811857e-06,
+      "loss": 0.0153,
+      "step": 1899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.795964517353734e-06,
+      "loss": 0.0217,
+      "step": 1900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.793069949176474e-06,
+      "loss": 0.0144,
+      "step": 1901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.79017401969085e-06,
+      "loss": 0.0134,
+      "step": 1902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.787276730308304e-06,
+      "loss": 0.0124,
+      "step": 1903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.78437808244094e-06,
+      "loss": 0.0148,
+      "step": 1904
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.781478077501526e-06,
+      "loss": 0.0168,
+      "step": 1905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.778576716903484e-06,
+      "loss": 0.0108,
+      "step": 1906
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.775674002060905e-06,
+      "loss": 0.0217,
+      "step": 1907
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.772769934388537e-06,
+      "loss": 0.0144,
+      "step": 1908
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.769864515301787e-06,
+      "loss": 0.0168,
+      "step": 1909
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.76695774621672e-06,
+      "loss": 0.0158,
+      "step": 1910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.764049628550063e-06,
+      "loss": 0.0119,
+      "step": 1911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.761140163719194e-06,
+      "loss": 0.0188,
+      "step": 1912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.758229353142153e-06,
+      "loss": 0.0144,
+      "step": 1913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.755317198237631e-06,
+      "loss": 0.0116,
+      "step": 1914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.752403700424978e-06,
+      "loss": 0.0129,
+      "step": 1915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.7494888611242e-06,
+      "loss": 0.0151,
+      "step": 1916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.74657268175595e-06,
+      "loss": 0.0233,
+      "step": 1917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.743655163741544e-06,
+      "loss": 0.0231,
+      "step": 1918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.740736308502939e-06,
+      "loss": 0.0141,
+      "step": 1919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.737816117462752e-06,
+      "loss": 0.0168,
+      "step": 1920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.734894592044249e-06,
+      "loss": 0.0179,
+      "step": 1921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.731971733671347e-06,
+      "loss": 0.0185,
+      "step": 1922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.729047543768608e-06,
+      "loss": 0.0126,
+      "step": 1923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.726122023761252e-06,
+      "loss": 0.0189,
+      "step": 1924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.723195175075136e-06,
+      "loss": 0.0216,
+      "step": 1925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.720266999136774e-06,
+      "loss": 0.0165,
+      "step": 1926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.717337497373324e-06,
+      "loss": 0.0139,
+      "step": 1927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.714406671212589e-06,
+      "loss": 0.0186,
+      "step": 1928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.711474522083015e-06,
+      "loss": 0.0193,
+      "step": 1929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.7085410514137e-06,
+      "loss": 0.0163,
+      "step": 1930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.70560626063438e-06,
+      "loss": 0.0101,
+      "step": 1931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.702670151175435e-06,
+      "loss": 0.0112,
+      "step": 1932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.699732724467894e-06,
+      "loss": 0.0194,
+      "step": 1933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.696793981943418e-06,
+      "loss": 0.0137,
+      "step": 1934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.693853925034316e-06,
+      "loss": 0.0189,
+      "step": 1935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.690912555173536e-06,
+      "loss": 0.0153,
+      "step": 1936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.687969873794667e-06,
+      "loss": 0.014,
+      "step": 1937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.685025882331936e-06,
+      "loss": 0.0129,
+      "step": 1938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.682080582220206e-06,
+      "loss": 0.0163,
+      "step": 1939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.679133974894984e-06,
+      "loss": 0.0159,
+      "step": 1940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.676186061792408e-06,
+      "loss": 0.0206,
+      "step": 1941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.673236844349257e-06,
+      "loss": 0.0153,
+      "step": 1942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.670286324002943e-06,
+      "loss": 0.0137,
+      "step": 1943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.667334502191514e-06,
+      "loss": 0.0152,
+      "step": 1944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.66438138035365e-06,
+      "loss": 0.0209,
+      "step": 1945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.66142695992867e-06,
+      "loss": 0.0122,
+      "step": 1946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.658471242356521e-06,
+      "loss": 0.0159,
+      "step": 1947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.655514229077784e-06,
+      "loss": 0.0129,
+      "step": 1948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.652555921533671e-06,
+      "loss": 0.0133,
+      "step": 1949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.649596321166024e-06,
+      "loss": 0.0112,
+      "step": 1950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.646635429417322e-06,
+      "loss": 0.0169,
+      "step": 1951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.64367324773066e-06,
+      "loss": 0.0142,
+      "step": 1952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.640709777549773e-06,
+      "loss": 0.0122,
+      "step": 1953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.637745020319019e-06,
+      "loss": 0.0114,
+      "step": 1954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.634778977483389e-06,
+      "loss": 0.0122,
+      "step": 1955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.63181165048849e-06,
+      "loss": 0.0143,
+      "step": 1956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.628843040780567e-06,
+      "loss": 0.0154,
+      "step": 1957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.6258731498064796e-06,
+      "loss": 0.0145,
+      "step": 1958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.622901979013717e-06,
+      "loss": 0.0125,
+      "step": 1959
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.619929529850397e-06,
+      "loss": 0.0111,
+      "step": 1960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.616955803765249e-06,
+      "loss": 0.0136,
+      "step": 1961
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.613980802207633e-06,
+      "loss": 0.0265,
+      "step": 1962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.6110045266275305e-06,
+      "loss": 0.0147,
+      "step": 1963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.6080269784755405e-06,
+      "loss": 0.01,
+      "step": 1964
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.605048159202884e-06,
+      "loss": 0.0163,
+      "step": 1965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.6020680702613995e-06,
+      "loss": 0.012,
+      "step": 1966
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.5990867131035474e-06,
+      "loss": 0.0157,
+      "step": 1967
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.596104089182408e-06,
+      "loss": 0.0173,
+      "step": 1968
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.5931201999516715e-06,
+      "loss": 0.0158,
+      "step": 1969
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.590135046865652e-06,
+      "loss": 0.0196,
+      "step": 1970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.587148631379276e-06,
+      "loss": 0.0185,
+      "step": 1971
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.5841609549480854e-06,
+      "loss": 0.0153,
+      "step": 1972
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.581172019028238e-06,
+      "loss": 0.0135,
+      "step": 1973
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.578181825076506e-06,
+      "loss": 0.0124,
+      "step": 1974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.575190374550272e-06,
+      "loss": 0.0125,
+      "step": 1975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.572197668907533e-06,
+      "loss": 0.024,
+      "step": 1976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.569203709606898e-06,
+      "loss": 0.0123,
+      "step": 1977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.566208498107586e-06,
+      "loss": 0.0107,
+      "step": 1978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.563212035869426e-06,
+      "loss": 0.0123,
+      "step": 1979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.560214324352858e-06,
+      "loss": 0.0142,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.55721536501893e-06,
+      "loss": 0.0137,
+      "step": 1981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.5542151593293e-06,
+      "loss": 0.0111,
+      "step": 1982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.55121370874623e-06,
+      "loss": 0.0112,
+      "step": 1983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.548211014732589e-06,
+      "loss": 0.014,
+      "step": 1984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.545207078751858e-06,
+      "loss": 0.0132,
+      "step": 1985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.542201902268115e-06,
+      "loss": 0.0193,
+      "step": 1986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.539195486746047e-06,
+      "loss": 0.0138,
+      "step": 1987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.536187833650947e-06,
+      "loss": 0.0097,
+      "step": 1988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.533178944448705e-06,
+      "loss": 0.0117,
+      "step": 1989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.530168820605819e-06,
+      "loss": 0.2457,
+      "step": 1990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.527157463589389e-06,
+      "loss": 0.0131,
+      "step": 1991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.52414487486711e-06,
+      "loss": 0.0153,
+      "step": 1992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.521131055907283e-06,
+      "loss": 0.0176,
+      "step": 1993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.518116008178805e-06,
+      "loss": 0.0112,
+      "step": 1994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.515099733151177e-06,
+      "loss": 0.0195,
+      "step": 1995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.512082232294491e-06,
+      "loss": 0.0141,
+      "step": 1996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.509063507079443e-06,
+      "loss": 0.0127,
+      "step": 1997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.5060435589773215e-06,
+      "loss": 0.0131,
+      "step": 1998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.503022389460014e-06,
+      "loss": 0.0154,
+      "step": 1999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.0156,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.496976392070358e-06,
+      "loss": 0.0192,
+      "step": 2001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.493951567144755e-06,
+      "loss": 0.026,
+      "step": 2002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.490925526697455e-06,
+      "loss": 0.0231,
+      "step": 2003
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.487898272203314e-06,
+      "loss": 0.0119,
+      "step": 2004
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.484869805137778e-06,
+      "loss": 0.0128,
+      "step": 2005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.481840126976885e-06,
+      "loss": 0.0119,
+      "step": 2006
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.478809239197264e-06,
+      "loss": 0.0169,
+      "step": 2007
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.475777143276133e-06,
+      "loss": 0.0207,
+      "step": 2008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.4727438406912986e-06,
+      "loss": 0.0097,
+      "step": 2009
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.469709332921155e-06,
+      "loss": 0.0169,
+      "step": 2010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.4666736214446855e-06,
+      "loss": 0.016,
+      "step": 2011
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.463636707741458e-06,
+      "loss": 0.0173,
+      "step": 2012
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.460598593291628e-06,
+      "loss": 0.0111,
+      "step": 2013
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.4575592795759356e-06,
+      "loss": 0.0284,
+      "step": 2014
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.454518768075705e-06,
+      "loss": 0.0092,
+      "step": 2015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.451477060272844e-06,
+      "loss": 0.0118,
+      "step": 2016
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.448434157649846e-06,
+      "loss": 0.0123,
+      "step": 2017
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.445390061689782e-06,
+      "loss": 0.0124,
+      "step": 2018
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.44234477387631e-06,
+      "loss": 0.0509,
+      "step": 2019
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.4392982956936644e-06,
+      "loss": 0.0141,
+      "step": 2020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.436250628626662e-06,
+      "loss": 0.014,
+      "step": 2021
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.433201774160701e-06,
+      "loss": 0.014,
+      "step": 2022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.430151733781752e-06,
+      "loss": 0.0147,
+      "step": 2023
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.42710050897637e-06,
+      "loss": 0.0105,
+      "step": 2024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.424048101231687e-06,
+      "loss": 0.0181,
+      "step": 2025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.4209945120354045e-06,
+      "loss": 0.0139,
+      "step": 2026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.4179397428758085e-06,
+      "loss": 0.0127,
+      "step": 2027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.414883795241754e-06,
+      "loss": 0.013,
+      "step": 2028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.411826670622676e-06,
+      "loss": 0.0113,
+      "step": 2029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.408768370508577e-06,
+      "loss": 0.0101,
+      "step": 2030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.405708896390037e-06,
+      "loss": 0.0116,
+      "step": 2031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.402648249758204e-06,
+      "loss": 0.0109,
+      "step": 2032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3995864321048036e-06,
+      "loss": 0.0296,
+      "step": 2033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.396523444922126e-06,
+      "loss": 0.0145,
+      "step": 2034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.393459289703035e-06,
+      "loss": 0.0181,
+      "step": 2035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.390393967940962e-06,
+      "loss": 0.0132,
+      "step": 2036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3873274811299065e-06,
+      "loss": 0.0186,
+      "step": 2037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3842598307644396e-06,
+      "loss": 0.0224,
+      "step": 2038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.381191018339697e-06,
+      "loss": 0.0098,
+      "step": 2039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.378121045351378e-06,
+      "loss": 0.016,
+      "step": 2040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.37504991329575e-06,
+      "loss": 0.0148,
+      "step": 2041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.371977623669646e-06,
+      "loss": 0.0143,
+      "step": 2042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.368904177970466e-06,
+      "loss": 0.0106,
+      "step": 2043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.365829577696166e-06,
+      "loss": 0.0107,
+      "step": 2044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.362753824345271e-06,
+      "loss": 0.0119,
+      "step": 2045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3596769194168646e-06,
+      "loss": 0.0176,
+      "step": 2046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3565988644105926e-06,
+      "loss": 0.0139,
+      "step": 2047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.353519660826665e-06,
+      "loss": 0.0182,
+      "step": 2048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.350439310165842e-06,
+      "loss": 0.0161,
+      "step": 2049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.347357813929455e-06,
+      "loss": 0.0103,
+      "step": 2050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.344275173619385e-06,
+      "loss": 0.0148,
+      "step": 2051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.341191390738073e-06,
+      "loss": 0.0115,
+      "step": 2052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.33810646678852e-06,
+      "loss": 0.0101,
+      "step": 2053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.335020403274277e-06,
+      "loss": 0.0116,
+      "step": 2054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3319332016994575e-06,
+      "loss": 0.0211,
+      "step": 2055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3288448635687215e-06,
+      "loss": 0.0097,
+      "step": 2056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.325755390387293e-06,
+      "loss": 0.0088,
+      "step": 2057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.32266478366094e-06,
+      "loss": 0.0161,
+      "step": 2058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.319573044895986e-06,
+      "loss": 0.0143,
+      "step": 2059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.31648017559931e-06,
+      "loss": 0.012,
+      "step": 2060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.313386177278335e-06,
+      "loss": 0.0099,
+      "step": 2061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.310291051441044e-06,
+      "loss": 0.0111,
+      "step": 2062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.307194799595958e-06,
+      "loss": 0.0114,
+      "step": 2063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3040974232521555e-06,
+      "loss": 0.0167,
+      "step": 2064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.300998923919259e-06,
+      "loss": 0.0088,
+      "step": 2065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.297899303107441e-06,
+      "loss": 0.0139,
+      "step": 2066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.294798562327417e-06,
+      "loss": 0.0087,
+      "step": 2067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.291696703090449e-06,
+      "loss": 0.01,
+      "step": 2068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.288593726908351e-06,
+      "loss": 0.0108,
+      "step": 2069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.285489635293472e-06,
+      "loss": 0.013,
+      "step": 2070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.282384429758709e-06,
+      "loss": 0.0131,
+      "step": 2071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.279278111817502e-06,
+      "loss": 0.0155,
+      "step": 2072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.27617068298383e-06,
+      "loss": 0.0082,
+      "step": 2073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.27306214477222e-06,
+      "loss": 0.0182,
+      "step": 2074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.269952498697734e-06,
+      "loss": 0.0095,
+      "step": 2075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.266841746275977e-06,
+      "loss": 0.0158,
+      "step": 2076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.26372988902309e-06,
+      "loss": 0.0125,
+      "step": 2077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.260616928455754e-06,
+      "loss": 0.0148,
+      "step": 2078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.257502866091192e-06,
+      "loss": 0.0401,
+      "step": 2079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.254387703447154e-06,
+      "loss": 0.0117,
+      "step": 2080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.251271442041938e-06,
+      "loss": 0.0157,
+      "step": 2081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.24815408339437e-06,
+      "loss": 0.0149,
+      "step": 2082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.245035629023812e-06,
+      "loss": 0.011,
+      "step": 2083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.241916080450163e-06,
+      "loss": 0.0128,
+      "step": 2084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.238795439193849e-06,
+      "loss": 0.0117,
+      "step": 2085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.235673706775837e-06,
+      "loss": 0.0115,
+      "step": 2086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2325508847176175e-06,
+      "loss": 0.0142,
+      "step": 2087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2294269745412214e-06,
+      "loss": 0.0153,
+      "step": 2088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.226301977769199e-06,
+      "loss": 0.011,
+      "step": 2089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.223175895924638e-06,
+      "loss": 0.0135,
+      "step": 2090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.220048730531154e-06,
+      "loss": 0.0106,
+      "step": 2091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.216920483112886e-06,
+      "loss": 0.0115,
+      "step": 2092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.21379115519451e-06,
+      "loss": 0.0138,
+      "step": 2093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.210660748301214e-06,
+      "loss": 0.0113,
+      "step": 2094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.207529263958727e-06,
+      "loss": 0.0098,
+      "step": 2095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2043967036932935e-06,
+      "loss": 0.0101,
+      "step": 2096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.201263069031686e-06,
+      "loss": 0.0148,
+      "step": 2097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.1981283615012e-06,
+      "loss": 0.0151,
+      "step": 2098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.194992582629654e-06,
+      "loss": 0.0117,
+      "step": 2099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.191855733945388e-06,
+      "loss": 0.0144,
+      "step": 2100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.188717816977264e-06,
+      "loss": 0.0116,
+      "step": 2101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.185578833254665e-06,
+      "loss": 0.0272,
+      "step": 2102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.182438784307495e-06,
+      "loss": 0.0129,
+      "step": 2103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.179297671666171e-06,
+      "loss": 0.0186,
+      "step": 2104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.176155496861639e-06,
+      "loss": 0.0163,
+      "step": 2105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.173012261425352e-06,
+      "loss": 0.0207,
+      "step": 2106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.169867966889288e-06,
+      "loss": 0.0123,
+      "step": 2107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.166722614785937e-06,
+      "loss": 0.0556,
+      "step": 2108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.1635762066483035e-06,
+      "loss": 0.0168,
+      "step": 2109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.160428744009913e-06,
+      "loss": 0.0289,
+      "step": 2110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.157280228404796e-06,
+      "loss": 0.0108,
+      "step": 2111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.154130661367503e-06,
+      "loss": 0.0149,
+      "step": 2112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.150980044433094e-06,
+      "loss": 0.007,
+      "step": 2113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.1478283791371415e-06,
+      "loss": 0.0149,
+      "step": 2114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.1446756670157306e-06,
+      "loss": 0.0105,
+      "step": 2115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.141521909605452e-06,
+      "loss": 0.0108,
+      "step": 2116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.138367108443411e-06,
+      "loss": 0.0105,
+      "step": 2117
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.135211265067217e-06,
+      "loss": 0.0084,
+      "step": 2118
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.1320543810149945e-06,
+      "loss": 0.0131,
+      "step": 2119
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.128896457825364e-06,
+      "loss": 0.0135,
+      "step": 2120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.125737497037464e-06,
+      "loss": 0.0127,
+      "step": 2121
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.12257750019093e-06,
+      "loss": 0.0085,
+      "step": 2122
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.119416468825908e-06,
+      "loss": 0.0134,
+      "step": 2123
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.116254404483049e-06,
+      "loss": 0.0126,
+      "step": 2124
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.113091308703498e-06,
+      "loss": 0.0233,
+      "step": 2125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.1099271830289155e-06,
+      "loss": 0.0152,
+      "step": 2126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.106762029001455e-06,
+      "loss": 0.0249,
+      "step": 2127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.103595848163775e-06,
+      "loss": 0.0101,
+      "step": 2128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.100428642059033e-06,
+      "loss": 0.0153,
+      "step": 2129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0972604122308865e-06,
+      "loss": 0.0106,
+      "step": 2130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.094091160223493e-06,
+      "loss": 0.0148,
+      "step": 2131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.090920887581507e-06,
+      "loss": 0.0132,
+      "step": 2132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.087749595850084e-06,
+      "loss": 0.0188,
+      "step": 2133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0845772865748684e-06,
+      "loss": 0.0121,
+      "step": 2134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.081403961302007e-06,
+      "loss": 0.0114,
+      "step": 2135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.07822962157814e-06,
+      "loss": 0.0096,
+      "step": 2136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.075054268950402e-06,
+      "loss": 0.0162,
+      "step": 2137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.071877904966422e-06,
+      "loss": 0.0121,
+      "step": 2138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0687005311743195e-06,
+      "loss": 0.0121,
+      "step": 2139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.06552214912271e-06,
+      "loss": 0.0105,
+      "step": 2140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0623427603606965e-06,
+      "loss": 0.0105,
+      "step": 2141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.059162366437875e-06,
+      "loss": 0.0107,
+      "step": 2142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0559809689043325e-06,
+      "loss": 0.0122,
+      "step": 2143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.052798569310641e-06,
+      "loss": 0.0123,
+      "step": 2144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.049615169207864e-06,
+      "loss": 0.0123,
+      "step": 2145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0464307701475544e-06,
+      "loss": 0.0109,
+      "step": 2146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.043245373681746e-06,
+      "loss": 0.0101,
+      "step": 2147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0400589813629645e-06,
+      "loss": 0.0087,
+      "step": 2148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.036871594744218e-06,
+      "loss": 0.0103,
+      "step": 2149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.033683215379002e-06,
+      "loss": 0.0129,
+      "step": 2150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.030493844821291e-06,
+      "loss": 0.0108,
+      "step": 2151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.027303484625547e-06,
+      "loss": 0.0118,
+      "step": 2152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.024112136346713e-06,
+      "loss": 0.0112,
+      "step": 2153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0209198015402115e-06,
+      "loss": 0.017,
+      "step": 2154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0177264817619514e-06,
+      "loss": 0.0113,
+      "step": 2155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.014532178568314e-06,
+      "loss": 0.011,
+      "step": 2156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.011336893516167e-06,
+      "loss": 0.0141,
+      "step": 2157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.008140628162851e-06,
+      "loss": 0.024,
+      "step": 2158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0049433840661875e-06,
+      "loss": 0.016,
+      "step": 2159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0017451627844765e-06,
+      "loss": 0.0121,
+      "step": 2160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.998545965876489e-06,
+      "loss": 0.0133,
+      "step": 2161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.995345794901477e-06,
+      "loss": 0.0098,
+      "step": 2162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.992144651419163e-06,
+      "loss": 0.0152,
+      "step": 2163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.98894253698975e-06,
+      "loss": 0.0359,
+      "step": 2164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.985739453173903e-06,
+      "loss": 0.0527,
+      "step": 2165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.9825354015327715e-06,
+      "loss": 0.0123,
+      "step": 2166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.979330383627969e-06,
+      "loss": 0.0103,
+      "step": 2167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.976124401021583e-06,
+      "loss": 0.0199,
+      "step": 2168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.97291745527617e-06,
+      "loss": 0.0113,
+      "step": 2169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.9697095479547564e-06,
+      "loss": 0.0171,
+      "step": 2170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.966500680620837e-06,
+      "loss": 0.0106,
+      "step": 2171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.963290854838376e-06,
+      "loss": 0.0084,
+      "step": 2172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.960080072171802e-06,
+      "loss": 0.0109,
+      "step": 2173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.9568683341860135e-06,
+      "loss": 0.0095,
+      "step": 2174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.953655642446368e-06,
+      "loss": 0.0108,
+      "step": 2175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.950441998518699e-06,
+      "loss": 0.0133,
+      "step": 2176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.947227403969293e-06,
+      "loss": 0.0108,
+      "step": 2177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.944011860364905e-06,
+      "loss": 0.0132,
+      "step": 2178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.940795369272754e-06,
+      "loss": 0.0114,
+      "step": 2179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9375779322605154e-06,
+      "loss": 0.0221,
+      "step": 2180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.934359550896332e-06,
+      "loss": 0.0116,
+      "step": 2181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9311402267488004e-06,
+      "loss": 0.0112,
+      "step": 2182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.927919961386984e-06,
+      "loss": 0.0121,
+      "step": 2183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.924698756380398e-06,
+      "loss": 0.0145,
+      "step": 2184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.921476613299018e-06,
+      "loss": 0.0126,
+      "step": 2185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9182535337132824e-06,
+      "loss": 0.01,
+      "step": 2186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.915029519194076e-06,
+      "loss": 0.0128,
+      "step": 2187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.911804571312746e-06,
+      "loss": 0.0106,
+      "step": 2188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.908578691641092e-06,
+      "loss": 0.0082,
+      "step": 2189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.905351881751372e-06,
+      "loss": 0.0088,
+      "step": 2190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9021241432162886e-06,
+      "loss": 0.0108,
+      "step": 2191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.898895477609007e-06,
+      "loss": 0.0157,
+      "step": 2192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.895665886503136e-06,
+      "loss": 0.0105,
+      "step": 2193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.892435371472741e-06,
+      "loss": 0.0109,
+      "step": 2194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.889203934092337e-06,
+      "loss": 0.0131,
+      "step": 2195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.885971575936884e-06,
+      "loss": 0.0141,
+      "step": 2196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.882738298581797e-06,
+      "loss": 0.0099,
+      "step": 2197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.879504103602934e-06,
+      "loss": 0.0155,
+      "step": 2198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.876268992576605e-06,
+      "loss": 0.0162,
+      "step": 2199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.873032967079562e-06,
+      "loss": 0.0162,
+      "step": 2200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.869796028689002e-06,
+      "loss": 0.0088,
+      "step": 2201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.866558178982575e-06,
+      "loss": 0.0109,
+      "step": 2202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.863319419538366e-06,
+      "loss": 0.0112,
+      "step": 2203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.860079751934908e-06,
+      "loss": 0.0087,
+      "step": 2204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.856839177751175e-06,
+      "loss": 0.0134,
+      "step": 2205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.853597698566583e-06,
+      "loss": 0.024,
+      "step": 2206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.850355315960992e-06,
+      "loss": 0.0152,
+      "step": 2207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.847112031514698e-06,
+      "loss": 0.0134,
+      "step": 2208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.843867846808438e-06,
+      "loss": 0.0109,
+      "step": 2209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.840622763423391e-06,
+      "loss": 0.0194,
+      "step": 2210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.837376782941168e-06,
+      "loss": 0.0115,
+      "step": 2211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.834129906943822e-06,
+      "loss": 0.0141,
+      "step": 2212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.830882137013839e-06,
+      "loss": 0.0152,
+      "step": 2213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.827633474734145e-06,
+      "loss": 0.0332,
+      "step": 2214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.824383921688098e-06,
+      "loss": 0.0122,
+      "step": 2215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.821133479459492e-06,
+      "loss": 0.0094,
+      "step": 2216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.81788214963255e-06,
+      "loss": 0.0094,
+      "step": 2217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.814629933791932e-06,
+      "loss": 0.0141,
+      "step": 2218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.811376833522729e-06,
+      "loss": 0.0807,
+      "step": 2219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.808122850410461e-06,
+      "loss": 0.0105,
+      "step": 2220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.804867986041084e-06,
+      "loss": 0.0127,
+      "step": 2221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.8016122420009745e-06,
+      "loss": 0.0158,
+      "step": 2222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.798355619876944e-06,
+      "loss": 0.013,
+      "step": 2223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.7950981212562315e-06,
+      "loss": 0.0117,
+      "step": 2224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.7918397477265e-06,
+      "loss": 0.011,
+      "step": 2225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.788580500875848e-06,
+      "loss": 0.011,
+      "step": 2226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.785320382292783e-06,
+      "loss": 0.0129,
+      "step": 2227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.782059393566254e-06,
+      "loss": 0.0119,
+      "step": 2228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.778797536285625e-06,
+      "loss": 0.009,
+      "step": 2229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.775534812040686e-06,
+      "loss": 0.0139,
+      "step": 2230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.772271222421649e-06,
+      "loss": 0.0103,
+      "step": 2231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.769006769019147e-06,
+      "loss": 0.017,
+      "step": 2232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.765741453424237e-06,
+      "loss": 0.015,
+      "step": 2233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.762475277228393e-06,
+      "loss": 0.0089,
+      "step": 2234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.759208242023509e-06,
+      "loss": 0.0171,
+      "step": 2235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.755940349401901e-06,
+      "loss": 0.0154,
+      "step": 2236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.752671600956295e-06,
+      "loss": 0.0224,
+      "step": 2237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.749401998279845e-06,
+      "loss": 0.0117,
+      "step": 2238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.746131542966112e-06,
+      "loss": 0.0198,
+      "step": 2239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.7428602366090764e-06,
+      "loss": 0.0095,
+      "step": 2240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.739588080803134e-06,
+      "loss": 0.0117,
+      "step": 2241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.736315077143095e-06,
+      "loss": 0.031,
+      "step": 2242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.733041227224182e-06,
+      "loss": 0.0102,
+      "step": 2243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.729766532642024e-06,
+      "loss": 0.0143,
+      "step": 2244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.7264909949926735e-06,
+      "loss": 0.0194,
+      "step": 2245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.723214615872585e-06,
+      "loss": 0.0113,
+      "step": 2246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.719937396878628e-06,
+      "loss": 0.0158,
+      "step": 2247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.716659339608077e-06,
+      "loss": 0.0156,
+      "step": 2248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.713380445658618e-06,
+      "loss": 0.0105,
+      "step": 2249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.710100716628345e-06,
+      "loss": 0.0092,
+      "step": 2250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.7068201541157555e-06,
+      "loss": 0.0108,
+      "step": 2251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.70353875971976e-06,
+      "loss": 0.0153,
+      "step": 2252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.700256535039665e-06,
+      "loss": 0.0114,
+      "step": 2253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.6969734816751906e-06,
+      "loss": 0.0104,
+      "step": 2254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.693689601226458e-06,
+      "loss": 0.0089,
+      "step": 2255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.690404895293987e-06,
+      "loss": 0.0153,
+      "step": 2256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.687119365478707e-06,
+      "loss": 0.012,
+      "step": 2257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.683833013381942e-06,
+      "loss": 0.0117,
+      "step": 2258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.680545840605423e-06,
+      "loss": 0.0185,
+      "step": 2259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.677257848751276e-06,
+      "loss": 0.0107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.673969039422029e-06,
+      "loss": 0.0334,
+      "step": 2261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.6706794142206085e-06,
+      "loss": 0.0097,
+      "step": 2262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.6673889747503364e-06,
+      "loss": 0.0124,
+      "step": 2263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.664097722614934e-06,
+      "loss": 0.0108,
+      "step": 2264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.6608056594185166e-06,
+      "loss": 0.0161,
+      "step": 2265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.657512786765599e-06,
+      "loss": 0.0148,
+      "step": 2266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.654219106261082e-06,
+      "loss": 0.0276,
+      "step": 2267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.6509246195102685e-06,
+      "loss": 0.0145,
+      "step": 2268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.647629328118852e-06,
+      "loss": 0.0158,
+      "step": 2269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.644333233692917e-06,
+      "loss": 0.0127,
+      "step": 2270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.64103633783894e-06,
+      "loss": 0.0125,
+      "step": 2271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.637738642163785e-06,
+      "loss": 0.0259,
+      "step": 2272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.634440148274712e-06,
+      "loss": 0.0157,
+      "step": 2273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.631140857779368e-06,
+      "loss": 0.0101,
+      "step": 2274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.627840772285784e-06,
+      "loss": 0.011,
+      "step": 2275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.624539893402383e-06,
+      "loss": 0.0129,
+      "step": 2276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.6212382227379726e-06,
+      "loss": 0.0109,
+      "step": 2277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.617935761901748e-06,
+      "loss": 0.0109,
+      "step": 2278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.614632512503289e-06,
+      "loss": 0.0118,
+      "step": 2279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.611328476152557e-06,
+      "loss": 0.0089,
+      "step": 2280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.6080236544599e-06,
+      "loss": 0.0113,
+      "step": 2281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.604718049036047e-06,
+      "loss": 0.0097,
+      "step": 2282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.601411661492114e-06,
+      "loss": 0.0119,
+      "step": 2283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.59810449343959e-06,
+      "loss": 0.0122,
+      "step": 2284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.594796546490351e-06,
+      "loss": 0.0199,
+      "step": 2285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.591487822256648e-06,
+      "loss": 0.0115,
+      "step": 2286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.588178322351113e-06,
+      "loss": 0.0122,
+      "step": 2287
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.58486804838676e-06,
+      "loss": 0.008,
+      "step": 2288
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.58155700197697e-06,
+      "loss": 0.0098,
+      "step": 2289
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.578245184735513e-06,
+      "loss": 0.013,
+      "step": 2290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.574932598276524e-06,
+      "loss": 0.009,
+      "step": 2291
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.571619244214521e-06,
+      "loss": 0.0086,
+      "step": 2292
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5683051241643894e-06,
+      "loss": 0.0177,
+      "step": 2293
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5649902397413915e-06,
+      "loss": 0.0299,
+      "step": 2294
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.561674592561164e-06,
+      "loss": 0.017,
+      "step": 2295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.558358184239709e-06,
+      "loss": 0.019,
+      "step": 2296
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.55504101639341e-06,
+      "loss": 0.015,
+      "step": 2297
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.551723090639008e-06,
+      "loss": 0.0187,
+      "step": 2298
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.548404408593622e-06,
+      "loss": 0.0076,
+      "step": 2299
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 0.0113,
+      "step": 2300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.541764782100208e-06,
+      "loss": 0.0096,
+      "step": 2301
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.538443840888254e-06,
+      "loss": 0.0143,
+      "step": 2302
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.53512214985746e-06,
+      "loss": 0.0113,
+      "step": 2303
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.53179971062678e-06,
+      "loss": 0.0107,
+      "step": 2304
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5284765248155295e-06,
+      "loss": 0.0257,
+      "step": 2305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.525152594043389e-06,
+      "loss": 0.0156,
+      "step": 2306
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5218279199304014e-06,
+      "loss": 0.0115,
+      "step": 2307
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.518502504096972e-06,
+      "loss": 0.0303,
+      "step": 2308
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5151763481638705e-06,
+      "loss": 0.0096,
+      "step": 2309
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5118494537522235e-06,
+      "loss": 0.0094,
+      "step": 2310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.508521822483518e-06,
+      "loss": 0.0107,
+      "step": 2311
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.505193455979603e-06,
+      "loss": 0.0201,
+      "step": 2312
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.501864355862682e-06,
+      "loss": 0.0118,
+      "step": 2313
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.49853452375532e-06,
+      "loss": 0.0137,
+      "step": 2314
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.495203961280434e-06,
+      "loss": 0.0173,
+      "step": 2315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.491872670061302e-06,
+      "loss": 0.011,
+      "step": 2316
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.4885406517215535e-06,
+      "loss": 0.0148,
+      "step": 2317
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.485207907885175e-06,
+      "loss": 0.0159,
+      "step": 2318
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.481874440176506e-06,
+      "loss": 0.0103,
+      "step": 2319
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.4785402502202345e-06,
+      "loss": 0.0123,
+      "step": 2320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.4752053396414075e-06,
+      "loss": 0.0126,
+      "step": 2321
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.471869710065418e-06,
+      "loss": 0.0133,
+      "step": 2322
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.4685333631180145e-06,
+      "loss": 0.0132,
+      "step": 2323
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.465196300425287e-06,
+      "loss": 0.0129,
+      "step": 2324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.461858523613684e-06,
+      "loss": 0.0147,
+      "step": 2325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.458520034309995e-06,
+      "loss": 0.0117,
+      "step": 2326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.455180834141359e-06,
+      "loss": 0.0173,
+      "step": 2327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.451840924735264e-06,
+      "loss": 0.0118,
+      "step": 2328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.448500307719537e-06,
+      "loss": 0.0125,
+      "step": 2329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.445158984722358e-06,
+      "loss": 0.0144,
+      "step": 2330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.441816957372247e-06,
+      "loss": 0.021,
+      "step": 2331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.438474227298065e-06,
+      "loss": 0.0094,
+      "step": 2332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.435130796129019e-06,
+      "loss": 0.0123,
+      "step": 2333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.431786665494657e-06,
+      "loss": 0.0142,
+      "step": 2334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.428441837024868e-06,
+      "loss": 0.0175,
+      "step": 2335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.425096312349881e-06,
+      "loss": 0.0115,
+      "step": 2336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.421750093100264e-06,
+      "loss": 0.0216,
+      "step": 2337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.418403180906923e-06,
+      "loss": 0.009,
+      "step": 2338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.415055577401101e-06,
+      "loss": 0.0125,
+      "step": 2339
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.411707284214384e-06,
+      "loss": 0.0114,
+      "step": 2340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.408358302978683e-06,
+      "loss": 0.0096,
+      "step": 2341
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.4050086353262565e-06,
+      "loss": 0.0114,
+      "step": 2342
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.401658282889689e-06,
+      "loss": 0.0198,
+      "step": 2343
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3983072473019e-06,
+      "loss": 0.0087,
+      "step": 2344
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3949555301961474e-06,
+      "loss": 0.0117,
+      "step": 2345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.391603133206015e-06,
+      "loss": 0.0455,
+      "step": 2346
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.388250057965421e-06,
+      "loss": 0.0118,
+      "step": 2347
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.384896306108612e-06,
+      "loss": 0.0189,
+      "step": 2348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3815418792701686e-06,
+      "loss": 0.011,
+      "step": 2349
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.378186779084996e-06,
+      "loss": 0.0401,
+      "step": 2350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.374831007188331e-06,
+      "loss": 0.0093,
+      "step": 2351
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.371474565215734e-06,
+      "loss": 0.0111,
+      "step": 2352
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.368117454803093e-06,
+      "loss": 0.0113,
+      "step": 2353
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.364759677586627e-06,
+      "loss": 0.0128,
+      "step": 2354
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.361401235202872e-06,
+      "loss": 0.0138,
+      "step": 2355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.358042129288694e-06,
+      "loss": 0.0121,
+      "step": 2356
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.35468236148128e-06,
+      "loss": 0.011,
+      "step": 2357
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.35132193341814e-06,
+      "loss": 0.015,
+      "step": 2358
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3479608467371055e-06,
+      "loss": 0.017,
+      "step": 2359
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.344599103076329e-06,
+      "loss": 0.0824,
+      "step": 2360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.341236704074285e-06,
+      "loss": 0.0102,
+      "step": 2361
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.337873651369764e-06,
+      "loss": 0.0123,
+      "step": 2362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.334509946601879e-06,
+      "loss": 0.0097,
+      "step": 2363
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.331145591410057e-06,
+      "loss": 0.0094,
+      "step": 2364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.327780587434045e-06,
+      "loss": 0.0164,
+      "step": 2365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.324414936313904e-06,
+      "loss": 0.0241,
+      "step": 2366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.321048639690013e-06,
+      "loss": 0.0125,
+      "step": 2367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.317681699203065e-06,
+      "loss": 0.0138,
+      "step": 2368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.314314116494061e-06,
+      "loss": 0.0164,
+      "step": 2369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.310945893204324e-06,
+      "loss": 0.0114,
+      "step": 2370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.307577030975485e-06,
+      "loss": 0.0077,
+      "step": 2371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.304207531449486e-06,
+      "loss": 0.0129,
+      "step": 2372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3008373962685785e-06,
+      "loss": 0.0085,
+      "step": 2373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.297466627075327e-06,
+      "loss": 0.013,
+      "step": 2374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.294095225512604e-06,
+      "loss": 0.0124,
+      "step": 2375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.290723193223589e-06,
+      "loss": 0.013,
+      "step": 2376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.28735053185177e-06,
+      "loss": 0.0091,
+      "step": 2377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.28397724304094e-06,
+      "loss": 0.0097,
+      "step": 2378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.280603328435199e-06,
+      "loss": 0.0143,
+      "step": 2379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.277228789678953e-06,
+      "loss": 0.0127,
+      "step": 2380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.273853628416911e-06,
+      "loss": 0.011,
+      "step": 2381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.270477846294086e-06,
+      "loss": 0.0146,
+      "step": 2382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.267101444955792e-06,
+      "loss": 0.0085,
+      "step": 2383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2637244260476474e-06,
+      "loss": 0.0129,
+      "step": 2384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.26034679121557e-06,
+      "loss": 0.012,
+      "step": 2385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.256968542105775e-06,
+      "loss": 0.0158,
+      "step": 2386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2535896803647845e-06,
+      "loss": 0.0206,
+      "step": 2387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.250210207639411e-06,
+      "loss": 0.0157,
+      "step": 2388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.24683012557677e-06,
+      "loss": 0.0103,
+      "step": 2389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.243449435824276e-06,
+      "loss": 0.0101,
+      "step": 2390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.240068140029628e-06,
+      "loss": 0.0098,
+      "step": 2391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.236686239840836e-06,
+      "loss": 0.008,
+      "step": 2392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.233303736906193e-06,
+      "loss": 0.0096,
+      "step": 2393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.229920632874291e-06,
+      "loss": 0.0125,
+      "step": 2394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2265369293940135e-06,
+      "loss": 0.0113,
+      "step": 2395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.223152628114537e-06,
+      "loss": 0.0112,
+      "step": 2396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.219767730685329e-06,
+      "loss": 0.0106,
+      "step": 2397
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.216382238756147e-06,
+      "loss": 0.0106,
+      "step": 2398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.212996153977038e-06,
+      "loss": 0.0125,
+      "step": 2399
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.209609477998339e-06,
+      "loss": 0.0139,
+      "step": 2400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.206222212470675e-06,
+      "loss": 0.0395,
+      "step": 2401
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.202834359044959e-06,
+      "loss": 0.0103,
+      "step": 2402
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.199445919372388e-06,
+      "loss": 0.0117,
+      "step": 2403
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.1960568951044475e-06,
+      "loss": 0.0113,
+      "step": 2404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.192667287892905e-06,
+      "loss": 0.0115,
+      "step": 2405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.189277099389816e-06,
+      "loss": 0.0095,
+      "step": 2406
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.185886331247516e-06,
+      "loss": 0.0139,
+      "step": 2407
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.182494985118625e-06,
+      "loss": 0.0096,
+      "step": 2408
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.179103062656042e-06,
+      "loss": 0.0104,
+      "step": 2409
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.17571056551295e-06,
+      "loss": 0.0137,
+      "step": 2410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.172317495342812e-06,
+      "loss": 0.0107,
+      "step": 2411
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.168923853799369e-06,
+      "loss": 0.0111,
+      "step": 2412
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.16552964253664e-06,
+      "loss": 0.0181,
+      "step": 2413
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.1621348632089205e-06,
+      "loss": 0.0141,
+      "step": 2414
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.158739517470786e-06,
+      "loss": 0.0107,
+      "step": 2415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.155343606977091e-06,
+      "loss": 0.0091,
+      "step": 2416
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.151947133382954e-06,
+      "loss": 0.0095,
+      "step": 2417
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.148550098343778e-06,
+      "loss": 0.012,
+      "step": 2418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.145152503515239e-06,
+      "loss": 0.0202,
+      "step": 2419
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.141754350553279e-06,
+      "loss": 0.008,
+      "step": 2420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.138355641114121e-06,
+      "loss": 0.0108,
+      "step": 2421
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.134956376854251e-06,
+      "loss": 0.0169,
+      "step": 2422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.13155655943043e-06,
+      "loss": 0.0126,
+      "step": 2423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.128156190499688e-06,
+      "loss": 0.0106,
+      "step": 2424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.124755271719326e-06,
+      "loss": 0.0143,
+      "step": 2425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.121353804746907e-06,
+      "loss": 0.0072,
+      "step": 2426
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.117951791240265e-06,
+      "loss": 0.0111,
+      "step": 2427
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.114549232857503e-06,
+      "loss": 0.0105,
+      "step": 2428
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.111146131256983e-06,
+      "loss": 0.0095,
+      "step": 2429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.107742488097338e-06,
+      "loss": 0.0322,
+      "step": 2430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.10433830503746e-06,
+      "loss": 0.0092,
+      "step": 2431
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.100933583736508e-06,
+      "loss": 0.0112,
+      "step": 2432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.097528325853903e-06,
+      "loss": 0.0414,
+      "step": 2433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.094122533049324e-06,
+      "loss": 0.0115,
+      "step": 2434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.090716206982714e-06,
+      "loss": 0.0132,
+      "step": 2435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.087309349314275e-06,
+      "loss": 0.0108,
+      "step": 2436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.083901961704467e-06,
+      "loss": 0.0102,
+      "step": 2437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.080494045814011e-06,
+      "loss": 0.0118,
+      "step": 2438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.077085603303883e-06,
+      "loss": 0.0103,
+      "step": 2439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.073676635835317e-06,
+      "loss": 0.0261,
+      "step": 2440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.0702671450698005e-06,
+      "loss": 0.0464,
+      "step": 2441
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.066857132669081e-06,
+      "loss": 0.0108,
+      "step": 2442
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.0634466002951545e-06,
+      "loss": 0.0182,
+      "step": 2443
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.060035549610275e-06,
+      "loss": 0.0126,
+      "step": 2444
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.056623982276945e-06,
+      "loss": 0.0109,
+      "step": 2445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.0532118999579206e-06,
+      "loss": 0.0091,
+      "step": 2446
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.049799304316214e-06,
+      "loss": 0.0118,
+      "step": 2447
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.046386197015076e-06,
+      "loss": 0.0179,
+      "step": 2448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.04297257971802e-06,
+      "loss": 0.0126,
+      "step": 2449
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.039558454088796e-06,
+      "loss": 0.0146,
+      "step": 2450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.036143821791413e-06,
+      "loss": 0.0094,
+      "step": 2451
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.032728684490118e-06,
+      "loss": 0.0097,
+      "step": 2452
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.029313043849407e-06,
+      "loss": 0.0124,
+      "step": 2453
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.025896901534023e-06,
+      "loss": 0.0135,
+      "step": 2454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.022480259208951e-06,
+      "loss": 0.013,
+      "step": 2455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.019063118539425e-06,
+      "loss": 0.0126,
+      "step": 2456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.015645481190912e-06,
+      "loss": 0.0096,
+      "step": 2457
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.01222734882913e-06,
+      "loss": 0.0098,
+      "step": 2458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.008808723120035e-06,
+      "loss": 0.0143,
+      "step": 2459
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.005389605729824e-06,
+      "loss": 0.0125,
+      "step": 2460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.001969998324932e-06,
+      "loss": 0.0119,
+      "step": 2461
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.9985499025720354e-06,
+      "loss": 0.0121,
+      "step": 2462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.995129320138047e-06,
+      "loss": 0.0138,
+      "step": 2463
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.991708252690117e-06,
+      "loss": 0.0124,
+      "step": 2464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.988286701895631e-06,
+      "loss": 0.0093,
+      "step": 2465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.984864669422214e-06,
+      "loss": 0.0105,
+      "step": 2466
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.98144215693772e-06,
+      "loss": 0.0088,
+      "step": 2467
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.978019166110242e-06,
+      "loss": 0.0121,
+      "step": 2468
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.974595698608103e-06,
+      "loss": 0.0081,
+      "step": 2469
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.97117175609986e-06,
+      "loss": 0.0792,
+      "step": 2470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.967747340254303e-06,
+      "loss": 0.0137,
+      "step": 2471
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.964322452740445e-06,
+      "loss": 0.0101,
+      "step": 2472
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.960897095227541e-06,
+      "loss": 0.0153,
+      "step": 2473
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.957471269385065e-06,
+      "loss": 0.0107,
+      "step": 2474
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.954044976882725e-06,
+      "loss": 0.0148,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.950618219390451e-06,
+      "loss": 0.0111,
+      "step": 2476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.947190998578407e-06,
+      "loss": 0.0254,
+      "step": 2477
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.943763316116977e-06,
+      "loss": 0.009,
+      "step": 2478
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.94033517367677e-06,
+      "loss": 0.0093,
+      "step": 2479
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.936906572928625e-06,
+      "loss": 0.0103,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.933477515543595e-06,
+      "loss": 0.014,
+      "step": 2481
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.930048003192965e-06,
+      "loss": 0.0239,
+      "step": 2482
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.926618037548237e-06,
+      "loss": 0.0211,
+      "step": 2483
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.923187620281135e-06,
+      "loss": 0.0113,
+      "step": 2484
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.919756753063601e-06,
+      "loss": 0.0149,
+      "step": 2485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9163254375677995e-06,
+      "loss": 0.0115,
+      "step": 2486
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.912893675466112e-06,
+      "loss": 0.0136,
+      "step": 2487
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.909461468431135e-06,
+      "loss": 0.0177,
+      "step": 2488
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.906028818135687e-06,
+      "loss": 0.0102,
+      "step": 2489
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.902595726252801e-06,
+      "loss": 0.0152,
+      "step": 2490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8991621944557224e-06,
+      "loss": 0.0124,
+      "step": 2491
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8957282244179125e-06,
+      "loss": 0.0215,
+      "step": 2492
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.892293817813048e-06,
+      "loss": 0.0219,
+      "step": 2493
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8888589763150165e-06,
+      "loss": 0.0118,
+      "step": 2494
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.885423701597918e-06,
+      "loss": 0.0138,
+      "step": 2495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.881987995336062e-06,
+      "loss": 0.0159,
+      "step": 2496
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.878551859203974e-06,
+      "loss": 0.0246,
+      "step": 2497
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8751152948763815e-06,
+      "loss": 0.0108,
+      "step": 2498
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.871678304028224e-06,
+      "loss": 0.0286,
+      "step": 2499
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8682408883346535e-06,
+      "loss": 0.0111,
+      "step": 2500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8648030494710195e-06,
+      "loss": 0.012,
+      "step": 2501
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8613647891128845e-06,
+      "loss": 0.0221,
+      "step": 2502
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.857926108936015e-06,
+      "loss": 0.0136,
+      "step": 2503
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.854487010616384e-06,
+      "loss": 0.0112,
+      "step": 2504
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.851047495830163e-06,
+      "loss": 0.0116,
+      "step": 2505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.847607566253732e-06,
+      "loss": 0.0129,
+      "step": 2506
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.844167223563669e-06,
+      "loss": 0.0119,
+      "step": 2507
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.840726469436758e-06,
+      "loss": 0.023,
+      "step": 2508
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.837285305549978e-06,
+      "loss": 0.0185,
+      "step": 2509
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8338437335805124e-06,
+      "loss": 0.0105,
+      "step": 2510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.83040175520574e-06,
+      "loss": 0.0098,
+      "step": 2511
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.826959372103239e-06,
+      "loss": 0.0104,
+      "step": 2512
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.823516585950787e-06,
+      "loss": 0.015,
+      "step": 2513
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8200733984263556e-06,
+      "loss": 0.0103,
+      "step": 2514
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.816629811208112e-06,
+      "loss": 0.0106,
+      "step": 2515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.813185825974419e-06,
+      "loss": 0.4315,
+      "step": 2516
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.809741444403831e-06,
+      "loss": 0.0318,
+      "step": 2517
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8062966681751046e-06,
+      "loss": 0.0147,
+      "step": 2518
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.802851498967173e-06,
+      "loss": 0.0123,
+      "step": 2519
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.799405938459175e-06,
+      "loss": 0.0092,
+      "step": 2520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.795959988330434e-06,
+      "loss": 0.0241,
+      "step": 2521
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.792513650260465e-06,
+      "loss": 0.0176,
+      "step": 2522
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.78906692592897e-06,
+      "loss": 0.0206,
+      "step": 2523
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.78561981701584e-06,
+      "loss": 0.013,
+      "step": 2524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.782172325201155e-06,
+      "loss": 0.0136,
+      "step": 2525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.778724452165181e-06,
+      "loss": 0.0083,
+      "step": 2526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.77527619958837e-06,
+      "loss": 0.0095,
+      "step": 2527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.771827569151357e-06,
+      "loss": 0.0099,
+      "step": 2528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.768378562534962e-06,
+      "loss": 0.0095,
+      "step": 2529
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764929181420191e-06,
+      "loss": 0.0152,
+      "step": 2530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.761479427488229e-06,
+      "loss": 0.0138,
+      "step": 2531
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7580293024204455e-06,
+      "loss": 0.0104,
+      "step": 2532
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7545788078983875e-06,
+      "loss": 0.026,
+      "step": 2533
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.751127945603786e-06,
+      "loss": 0.0152,
+      "step": 2534
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.747676717218549e-06,
+      "loss": 0.0133,
+      "step": 2535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.744225124424762e-06,
+      "loss": 0.0167,
+      "step": 2536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.740773168904691e-06,
+      "loss": 0.0106,
+      "step": 2537
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.737320852340776e-06,
+      "loss": 0.012,
+      "step": 2538
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.733868176415633e-06,
+      "loss": 0.0108,
+      "step": 2539
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.730415142812059e-06,
+      "loss": 0.0151,
+      "step": 2540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.726961753213016e-06,
+      "loss": 0.0113,
+      "step": 2541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.723508009301646e-06,
+      "loss": 0.019,
+      "step": 2542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.720053912761261e-06,
+      "loss": 0.0105,
+      "step": 2543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.716599465275347e-06,
+      "loss": 0.0168,
+      "step": 2544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7131446685275595e-06,
+      "loss": 0.0201,
+      "step": 2545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.709689524201723e-06,
+      "loss": 0.0189,
+      "step": 2546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.706234033981835e-06,
+      "loss": 0.0101,
+      "step": 2547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.702778199552055e-06,
+      "loss": 0.0192,
+      "step": 2548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6993220225967214e-06,
+      "loss": 0.0178,
+      "step": 2549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.695865504800328e-06,
+      "loss": 0.0125,
+      "step": 2550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.692408647847542e-06,
+      "loss": 0.0107,
+      "step": 2551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.68895145342319e-06,
+      "loss": 0.0113,
+      "step": 2552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.685493923212273e-06,
+      "loss": 0.0173,
+      "step": 2553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.682036058899942e-06,
+      "loss": 0.0091,
+      "step": 2554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.678577862171523e-06,
+      "loss": 0.0296,
+      "step": 2555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.675119334712496e-06,
+      "loss": 0.01,
+      "step": 2556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.671660478208508e-06,
+      "loss": 0.0303,
+      "step": 2557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.668201294345363e-06,
+      "loss": 0.0123,
+      "step": 2558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6647417848090225e-06,
+      "loss": 0.0134,
+      "step": 2559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.661281951285613e-06,
+      "loss": 0.0101,
+      "step": 2560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.657821795461413e-06,
+      "loss": 0.0123,
+      "step": 2561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.654361319022862e-06,
+      "loss": 0.0106,
+      "step": 2562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.650900523656553e-06,
+      "loss": 0.0138,
+      "step": 2563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.647439411049235e-06,
+      "loss": 0.015,
+      "step": 2564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.643977982887815e-06,
+      "loss": 0.0209,
+      "step": 2565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.640516240859348e-06,
+      "loss": 0.0264,
+      "step": 2566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6370541866510476e-06,
+      "loss": 0.0208,
+      "step": 2567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.633591821950274e-06,
+      "loss": 0.0078,
+      "step": 2568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.630129148444543e-06,
+      "loss": 0.0112,
+      "step": 2569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.626666167821522e-06,
+      "loss": 0.0149,
+      "step": 2570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.623202881769023e-06,
+      "loss": 0.0137,
+      "step": 2571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6197392919750095e-06,
+      "loss": 0.0185,
+      "step": 2572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.616275400127594e-06,
+      "loss": 0.0168,
+      "step": 2573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.612811207915034e-06,
+      "loss": 0.0204,
+      "step": 2574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.609346717025738e-06,
+      "loss": 0.011,
+      "step": 2575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.605881929148254e-06,
+      "loss": 0.015,
+      "step": 2576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.6024168459712765e-06,
+      "loss": 0.0141,
+      "step": 2577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.598951469183649e-06,
+      "loss": 0.01,
+      "step": 2578
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.59548580047435e-06,
+      "loss": 0.0174,
+      "step": 2579
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.592019841532507e-06,
+      "loss": 0.0112,
+      "step": 2580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.588553594047382e-06,
+      "loss": 0.0103,
+      "step": 2581
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.585087059708389e-06,
+      "loss": 0.0126,
+      "step": 2582
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.581620240205068e-06,
+      "loss": 0.0143,
+      "step": 2583
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.578153137227109e-06,
+      "loss": 0.0085,
+      "step": 2584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5746857524643335e-06,
+      "loss": 0.0084,
+      "step": 2585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5712180876067045e-06,
+      "loss": 0.0171,
+      "step": 2586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.567750144344318e-06,
+      "loss": 0.0089,
+      "step": 2587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5642819243674085e-06,
+      "loss": 0.0101,
+      "step": 2588
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.560813429366345e-06,
+      "loss": 0.0155,
+      "step": 2589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.557344661031628e-06,
+      "loss": 0.0116,
+      "step": 2590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.553875621053893e-06,
+      "loss": 0.0125,
+      "step": 2591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5504063111239116e-06,
+      "loss": 0.0138,
+      "step": 2592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.546936732932578e-06,
+      "loss": 0.0109,
+      "step": 2593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.543466888170927e-06,
+      "loss": 0.0093,
+      "step": 2594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.539996778530114e-06,
+      "loss": 0.0136,
+      "step": 2595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.536526405701433e-06,
+      "loss": 0.0115,
+      "step": 2596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5330557713763e-06,
+      "loss": 0.0094,
+      "step": 2597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.52958487724626e-06,
+      "loss": 0.012,
+      "step": 2598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.526113725002984e-06,
+      "loss": 0.0075,
+      "step": 2599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.522642316338268e-06,
+      "loss": 0.0103,
+      "step": 2600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.519170652944037e-06,
+      "loss": 0.01,
+      "step": 2601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.515698736512337e-06,
+      "loss": 0.0097,
+      "step": 2602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.512226568735338e-06,
+      "loss": 0.0084,
+      "step": 2603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.508754151305332e-06,
+      "loss": 0.0105,
+      "step": 2604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.505281485914732e-06,
+      "loss": 0.0128,
+      "step": 2605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5018085742560745e-06,
+      "loss": 0.0076,
+      "step": 2606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.498335418022015e-06,
+      "loss": 0.0098,
+      "step": 2607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.4948620189053255e-06,
+      "loss": 0.0116,
+      "step": 2608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.491388378598899e-06,
+      "loss": 0.013,
+      "step": 2609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.487914498795748e-06,
+      "loss": 0.0096,
+      "step": 2610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.484440381188997e-06,
+      "loss": 0.0141,
+      "step": 2611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.480966027471889e-06,
+      "loss": 0.0115,
+      "step": 2612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.477491439337782e-06,
+      "loss": 0.0136,
+      "step": 2613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.474016618480147e-06,
+      "loss": 0.0082,
+      "step": 2614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.470541566592573e-06,
+      "loss": 0.0108,
+      "step": 2615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.467066285368754e-06,
+      "loss": 0.0101,
+      "step": 2616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.463590776502501e-06,
+      "loss": 0.0099,
+      "step": 2617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.460115041687737e-06,
+      "loss": 0.0101,
+      "step": 2618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.456639082618489e-06,
+      "loss": 0.0106,
+      "step": 2619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.453162900988902e-06,
+      "loss": 0.0108,
+      "step": 2620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.449686498493219e-06,
+      "loss": 0.011,
+      "step": 2621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.446209876825803e-06,
+      "loss": 0.0103,
+      "step": 2622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.442733037681112e-06,
+      "loss": 0.0137,
+      "step": 2623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.439255982753717e-06,
+      "loss": 0.0117,
+      "step": 2624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.435778713738292e-06,
+      "loss": 0.0106,
+      "step": 2625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.432301232329615e-06,
+      "loss": 0.0106,
+      "step": 2626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.42882354022257e-06,
+      "loss": 0.0167,
+      "step": 2627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.425345639112141e-06,
+      "loss": 0.0162,
+      "step": 2628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4218675306934145e-06,
+      "loss": 0.0093,
+      "step": 2629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.41838921666158e-06,
+      "loss": 0.0277,
+      "step": 2630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.41491069871192e-06,
+      "loss": 0.0172,
+      "step": 2631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.411431978539829e-06,
+      "loss": 0.0144,
+      "step": 2632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4079530578407895e-06,
+      "loss": 0.012,
+      "step": 2633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.404473938310384e-06,
+      "loss": 0.011,
+      "step": 2634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.400994621644294e-06,
+      "loss": 0.0147,
+      "step": 2635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3975151095383e-06,
+      "loss": 0.0107,
+      "step": 2636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.394035403688268e-06,
+      "loss": 0.0108,
+      "step": 2637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.390555505790168e-06,
+      "loss": 0.0118,
+      "step": 2638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3870754175400595e-06,
+      "loss": 0.0072,
+      "step": 2639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.383595140634093e-06,
+      "loss": 0.0091,
+      "step": 2640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.380114676768516e-06,
+      "loss": 0.0102,
+      "step": 2641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.376634027639664e-06,
+      "loss": 0.0092,
+      "step": 2642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.373153194943962e-06,
+      "loss": 0.0156,
+      "step": 2643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3696721803779265e-06,
+      "loss": 0.0087,
+      "step": 2644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.366190985638159e-06,
+      "loss": 0.0088,
+      "step": 2645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.362709612421355e-06,
+      "loss": 0.0092,
+      "step": 2646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.359228062424292e-06,
+      "loss": 0.0106,
+      "step": 2647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.355746337343835e-06,
+      "loss": 0.0173,
+      "step": 2648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.352264438876935e-06,
+      "loss": 0.0172,
+      "step": 2649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.348782368720627e-06,
+      "loss": 0.0243,
+      "step": 2650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.345300128572031e-06,
+      "loss": 0.0246,
+      "step": 2651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.341817720128344e-06,
+      "loss": 0.0111,
+      "step": 2652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.338335145086855e-06,
+      "loss": 0.026,
+      "step": 2653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.334852405144926e-06,
+      "loss": 0.0123,
+      "step": 2654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3313695020000026e-06,
+      "loss": 0.0135,
+      "step": 2655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.327886437349609e-06,
+      "loss": 0.0147,
+      "step": 2656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3244032128913485e-06,
+      "loss": 0.0101,
+      "step": 2657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.320919830322903e-06,
+      "loss": 0.0137,
+      "step": 2658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.317436291342031e-06,
+      "loss": 0.0118,
+      "step": 2659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3139525976465675e-06,
+      "loss": 0.01,
+      "step": 2660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.310468750934421e-06,
+      "loss": 0.0272,
+      "step": 2661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.306984752903578e-06,
+      "loss": 0.0161,
+      "step": 2662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.303500605252095e-06,
+      "loss": 0.016,
+      "step": 2663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.300016309678104e-06,
+      "loss": 0.0116,
+      "step": 2664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.296531867879809e-06,
+      "loss": 0.0102,
+      "step": 2665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.293047281555482e-06,
+      "loss": 0.0098,
+      "step": 2666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.289562552403472e-06,
+      "loss": 0.012,
+      "step": 2667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2860776821221915e-06,
+      "loss": 0.0102,
+      "step": 2668
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.282592672410124e-06,
+      "loss": 0.0193,
+      "step": 2669
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.27910752496582e-06,
+      "loss": 0.0194,
+      "step": 2670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.275622241487899e-06,
+      "loss": 0.0174,
+      "step": 2671
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.272136823675046e-06,
+      "loss": 0.0098,
+      "step": 2672
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.268651273226011e-06,
+      "loss": 0.0104,
+      "step": 2673
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.26516559183961e-06,
+      "loss": 0.0213,
+      "step": 2674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2616797812147205e-06,
+      "loss": 0.0149,
+      "step": 2675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.258193843050283e-06,
+      "loss": 0.0121,
+      "step": 2676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.254707779045305e-06,
+      "loss": 0.0139,
+      "step": 2677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.251221590898848e-06,
+      "loss": 0.0106,
+      "step": 2678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.247735280310041e-06,
+      "loss": 0.0089,
+      "step": 2679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.244248848978067e-06,
+      "loss": 0.0096,
+      "step": 2680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.240762298602171e-06,
+      "loss": 0.0114,
+      "step": 2681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.237275630881658e-06,
+      "loss": 0.0127,
+      "step": 2682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.233788847515882e-06,
+      "loss": 0.009,
+      "step": 2683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.230301950204261e-06,
+      "loss": 0.0149,
+      "step": 2684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.226814940646268e-06,
+      "loss": 0.0122,
+      "step": 2685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.223327820541432e-06,
+      "loss": 0.009,
+      "step": 2686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.219840591589325e-06,
+      "loss": 0.0108,
+      "step": 2687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.216353255489586e-06,
+      "loss": 0.0802,
+      "step": 2688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.212865813941899e-06,
+      "loss": 0.0114,
+      "step": 2689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.209378268645998e-06,
+      "loss": 0.0099,
+      "step": 2690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.205890621301676e-06,
+      "loss": 0.0113,
+      "step": 2691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.202402873608763e-06,
+      "loss": 0.0109,
+      "step": 2692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.19891502726715e-06,
+      "loss": 0.0083,
+      "step": 2693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.195427083976768e-06,
+      "loss": 0.0105,
+      "step": 2694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1919390454376e-06,
+      "loss": 0.0099,
+      "step": 2695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.188450913349674e-06,
+      "loss": 0.0085,
+      "step": 2696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.18496268941306e-06,
+      "loss": 0.0233,
+      "step": 2697
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.18147437532788e-06,
+      "loss": 0.0091,
+      "step": 2698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.177985972794293e-06,
+      "loss": 0.0423,
+      "step": 2699
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.174497483512506e-06,
+      "loss": 0.0103,
+      "step": 2700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.171008909182765e-06,
+      "loss": 0.0156,
+      "step": 2701
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.167520251505358e-06,
+      "loss": 0.008,
+      "step": 2702
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.164031512180616e-06,
+      "loss": 0.0102,
+      "step": 2703
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.160542692908909e-06,
+      "loss": 0.0128,
+      "step": 2704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.157053795390642e-06,
+      "loss": 0.0091,
+      "step": 2705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.153564821326265e-06,
+      "loss": 0.0103,
+      "step": 2706
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.150075772416256e-06,
+      "loss": 0.0118,
+      "step": 2707
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.146586650361143e-06,
+      "loss": 0.0127,
+      "step": 2708
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.143097456861474e-06,
+      "loss": 0.0129,
+      "step": 2709
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.139608193617846e-06,
+      "loss": 0.014,
+      "step": 2710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.136118862330876e-06,
+      "loss": 0.0102,
+      "step": 2711
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.13262946470123e-06,
+      "loss": 0.0084,
+      "step": 2712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1291400024295946e-06,
+      "loss": 0.0091,
+      "step": 2713
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1256504772166885e-06,
+      "loss": 0.0097,
+      "step": 2714
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1221608907632665e-06,
+      "loss": 0.0108,
+      "step": 2715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.118671244770111e-06,
+      "loss": 0.0113,
+      "step": 2716
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.115181540938032e-06,
+      "loss": 0.0108,
+      "step": 2717
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.111691780967869e-06,
+      "loss": 0.0123,
+      "step": 2718
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1082019665604895e-06,
+      "loss": 0.0094,
+      "step": 2719
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1047120994167855e-06,
+      "loss": 0.0112,
+      "step": 2720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.101222181237676e-06,
+      "loss": 0.0091,
+      "step": 2721
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.097732213724107e-06,
+      "loss": 0.0089,
+      "step": 2722
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.0942421985770415e-06,
+      "loss": 0.0094,
+      "step": 2723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.090752137497474e-06,
+      "loss": 0.0092,
+      "step": 2724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.087262032186418e-06,
+      "loss": 0.0131,
+      "step": 2725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.083771884344908e-06,
+      "loss": 0.0081,
+      "step": 2726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.080281695673999e-06,
+      "loss": 0.01,
+      "step": 2727
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0767914678747655e-06,
+      "loss": 0.0101,
+      "step": 2728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.073301202648304e-06,
+      "loss": 0.0104,
+      "step": 2729
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.069810901695727e-06,
+      "loss": 0.0112,
+      "step": 2730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.066320566718165e-06,
+      "loss": 0.0096,
+      "step": 2731
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.062830199416764e-06,
+      "loss": 0.0118,
+      "step": 2732
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.059339801492687e-06,
+      "loss": 0.0078,
+      "step": 2733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.055849374647112e-06,
+      "loss": 0.0106,
+      "step": 2734
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.05235892058123e-06,
+      "loss": 0.0079,
+      "step": 2735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.048868440996246e-06,
+      "loss": 0.0084,
+      "step": 2736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.045377937593376e-06,
+      "loss": 0.0107,
+      "step": 2737
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.041887412073853e-06,
+      "loss": 0.0095,
+      "step": 2738
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.038396866138915e-06,
+      "loss": 0.0075,
+      "step": 2739
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.034906301489808e-06,
+      "loss": 0.0089,
+      "step": 2740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.031415719827796e-06,
+      "loss": 0.0114,
+      "step": 2741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.027925122854141e-06,
+      "loss": 0.0065,
+      "step": 2742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.024434512270123e-06,
+      "loss": 0.0101,
+      "step": 2743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0209438897770205e-06,
+      "loss": 0.0088,
+      "step": 2744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0174532570761194e-06,
+      "loss": 0.0112,
+      "step": 2745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.013962615868714e-06,
+      "loss": 0.0149,
+      "step": 2746
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.010471967856096e-06,
+      "loss": 0.0113,
+      "step": 2747
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.006981314739573e-06,
+      "loss": 0.0115,
+      "step": 2748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.003490658220438e-06,
+      "loss": 0.009,
+      "step": 2749
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 0.0084,
+      "step": 2750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.996509341779563e-06,
+      "loss": 0.0102,
+      "step": 2751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.993018685260428e-06,
+      "loss": 0.008,
+      "step": 2752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9895280321439036e-06,
+      "loss": 0.0093,
+      "step": 2753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.986037384131288e-06,
+      "loss": 0.0082,
+      "step": 2754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.982546742923883e-06,
+      "loss": 0.0183,
+      "step": 2755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.979056110222982e-06,
+      "loss": 0.0087,
+      "step": 2756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.975565487729879e-06,
+      "loss": 0.0128,
+      "step": 2757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9720748771458595e-06,
+      "loss": 0.0099,
+      "step": 2758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.968584280172206e-06,
+      "loss": 0.0115,
+      "step": 2759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.965093698510192e-06,
+      "loss": 0.0377,
+      "step": 2760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.961603133861086e-06,
+      "loss": 0.0087,
+      "step": 2761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9581125879261476e-06,
+      "loss": 0.0112,
+      "step": 2762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.954622062406623e-06,
+      "loss": 0.0102,
+      "step": 2763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.951131559003756e-06,
+      "loss": 0.0095,
+      "step": 2764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9476410794187726e-06,
+      "loss": 0.0105,
+      "step": 2765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.94415062535289e-06,
+      "loss": 0.0123,
+      "step": 2766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.940660198507315e-06,
+      "loss": 0.0128,
+      "step": 2767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.937169800583237e-06,
+      "loss": 0.0317,
+      "step": 2768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.933679433281837e-06,
+      "loss": 0.0598,
+      "step": 2769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9301890983042744e-06,
+      "loss": 0.0108,
+      "step": 2770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.926698797351697e-06,
+      "loss": 0.0097,
+      "step": 2771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.923208532125235e-06,
+      "loss": 0.0108,
+      "step": 2772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9197183043260035e-06,
+      "loss": 0.0398,
+      "step": 2773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9162281156550945e-06,
+      "loss": 0.0106,
+      "step": 2774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9127379678135825e-06,
+      "loss": 0.0077,
+      "step": 2775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9092478625025266e-06,
+      "loss": 0.0113,
+      "step": 2776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.90575780142296e-06,
+      "loss": 0.0145,
+      "step": 2777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.902267786275895e-06,
+      "loss": 0.0073,
+      "step": 2778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.898777818762325e-06,
+      "loss": 0.0141,
+      "step": 2779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.895287900583216e-06,
+      "loss": 0.013,
+      "step": 2780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.891798033439511e-06,
+      "loss": 0.0141,
+      "step": 2781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.888308219032133e-06,
+      "loss": 0.0113,
+      "step": 2782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.88481845906197e-06,
+      "loss": 0.0121,
+      "step": 2783
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.881328755229892e-06,
+      "loss": 0.0074,
+      "step": 2784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.877839109236735e-06,
+      "loss": 0.0106,
+      "step": 2785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.874349522783313e-06,
+      "loss": 0.0102,
+      "step": 2786
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.870859997570407e-06,
+      "loss": 0.0077,
+      "step": 2787
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.86737053529877e-06,
+      "loss": 0.009,
+      "step": 2788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.863881137669123e-06,
+      "loss": 0.009,
+      "step": 2789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.860391806382157e-06,
+      "loss": 0.0097,
+      "step": 2790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.856902543138528e-06,
+      "loss": 0.0079,
+      "step": 2791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.853413349638859e-06,
+      "loss": 0.0093,
+      "step": 2792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8499242275837444e-06,
+      "loss": 0.0098,
+      "step": 2793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.846435178673737e-06,
+      "loss": 0.0104,
+      "step": 2794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.842946204609359e-06,
+      "loss": 0.0115,
+      "step": 2795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.839457307091093e-06,
+      "loss": 0.0433,
+      "step": 2796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.835968487819384e-06,
+      "loss": 0.0103,
+      "step": 2797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.832479748494643e-06,
+      "loss": 0.0113,
+      "step": 2798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.828991090817238e-06,
+      "loss": 0.0207,
+      "step": 2799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.825502516487497e-06,
+      "loss": 0.0164,
+      "step": 2800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.822014027205708e-06,
+      "loss": 0.0115,
+      "step": 2801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.818525624672122e-06,
+      "loss": 0.0096,
+      "step": 2802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.815037310586941e-06,
+      "loss": 0.0081,
+      "step": 2803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.811549086650327e-06,
+      "loss": 0.01,
+      "step": 2804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8080609545624004e-06,
+      "loss": 0.0124,
+      "step": 2805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8045729160232326e-06,
+      "loss": 0.0103,
+      "step": 2806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.801084972732851e-06,
+      "loss": 0.009,
+      "step": 2807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.797597126391238e-06,
+      "loss": 0.0091,
+      "step": 2808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.794109378698327e-06,
+      "loss": 0.0117,
+      "step": 2809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7906217313540035e-06,
+      "loss": 0.0198,
+      "step": 2810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.787134186058103e-06,
+      "loss": 0.0112,
+      "step": 2811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.783646744510416e-06,
+      "loss": 0.0091,
+      "step": 2812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.780159408410677e-06,
+      "loss": 0.0138,
+      "step": 2813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7766721794585704e-06,
+      "loss": 0.0103,
+      "step": 2814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7731850593537316e-06,
+      "loss": 0.0095,
+      "step": 2815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.769698049795739e-06,
+      "loss": 0.0078,
+      "step": 2816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.766211152484122e-06,
+      "loss": 0.0084,
+      "step": 2817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.762724369118346e-06,
+      "loss": 0.016,
+      "step": 2818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.759237701397831e-06,
+      "loss": 0.017,
+      "step": 2819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.755751151021934e-06,
+      "loss": 0.0199,
+      "step": 2820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.752264719689961e-06,
+      "loss": 0.0097,
+      "step": 2821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.748778409101153e-06,
+      "loss": 0.0147,
+      "step": 2822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.745292220954696e-06,
+      "loss": 0.0224,
+      "step": 2823
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.741806156949718e-06,
+      "loss": 0.0876,
+      "step": 2824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.738320218785281e-06,
+      "loss": 0.1361,
+      "step": 2825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.734834408160393e-06,
+      "loss": 0.0099,
+      "step": 2826
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.73134872677399e-06,
+      "loss": 0.0103,
+      "step": 2827
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.727863176324955e-06,
+      "loss": 0.0101,
+      "step": 2828
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7243777585121034e-06,
+      "loss": 0.0197,
+      "step": 2829
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.720892475034181e-06,
+      "loss": 0.0105,
+      "step": 2830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.717407327589878e-06,
+      "loss": 0.0096,
+      "step": 2831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.71392231787781e-06,
+      "loss": 0.013,
+      "step": 2832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.710437447596528e-06,
+      "loss": 0.0114,
+      "step": 2833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.706952718444518e-06,
+      "loss": 0.0175,
+      "step": 2834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.703468132120193e-06,
+      "loss": 0.0095,
+      "step": 2835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.699983690321898e-06,
+      "loss": 0.0103,
+      "step": 2836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.696499394747906e-06,
+      "loss": 0.0086,
+      "step": 2837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.693015247096423e-06,
+      "loss": 0.0093,
+      "step": 2838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.689531249065581e-06,
+      "loss": 0.0104,
+      "step": 2839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.686047402353433e-06,
+      "loss": 0.0087,
+      "step": 2840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.68256370865797e-06,
+      "loss": 0.0163,
+      "step": 2841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.679080169677097e-06,
+      "loss": 0.01,
+      "step": 2842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.675596787108652e-06,
+      "loss": 0.0105,
+      "step": 2843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.672113562650394e-06,
+      "loss": 0.0119,
+      "step": 2844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.668630498000001e-06,
+      "loss": 0.0127,
+      "step": 2845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6651475948550765e-06,
+      "loss": 0.0112,
+      "step": 2846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.661664854913147e-06,
+      "loss": 0.0122,
+      "step": 2847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.658182279871657e-06,
+      "loss": 0.0127,
+      "step": 2848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.654699871427972e-06,
+      "loss": 0.0087,
+      "step": 2849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.651217631279374e-06,
+      "loss": 0.0114,
+      "step": 2850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6477355611230655e-06,
+      "loss": 0.0099,
+      "step": 2851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.644253662656167e-06,
+      "loss": 0.0096,
+      "step": 2852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6407719375757095e-06,
+      "loss": 0.0104,
+      "step": 2853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.637290387578647e-06,
+      "loss": 0.01,
+      "step": 2854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6338090143618435e-06,
+      "loss": 0.0104,
+      "step": 2855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.630327819622076e-06,
+      "loss": 0.0156,
+      "step": 2856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6268468050560394e-06,
+      "loss": 0.0101,
+      "step": 2857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6233659723603374e-06,
+      "loss": 0.0143,
+      "step": 2858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.619885323231484e-06,
+      "loss": 0.0094,
+      "step": 2859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6164048593659076e-06,
+      "loss": 0.0082,
+      "step": 2860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.612924582459943e-06,
+      "loss": 0.0103,
+      "step": 2861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.609444494209834e-06,
+      "loss": 0.011,
+      "step": 2862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.605964596311733e-06,
+      "loss": 0.014,
+      "step": 2863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.602484890461702e-06,
+      "loss": 0.0105,
+      "step": 2864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5990053783557066e-06,
+      "loss": 0.0128,
+      "step": 2865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.595526061689617e-06,
+      "loss": 0.0277,
+      "step": 2866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.592046942159213e-06,
+      "loss": 0.0488,
+      "step": 2867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.588568021460172e-06,
+      "loss": 0.0143,
+      "step": 2868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5850893012880806e-06,
+      "loss": 0.0115,
+      "step": 2869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.581610783338424e-06,
+      "loss": 0.0157,
+      "step": 2870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.578132469306588e-06,
+      "loss": 0.0117,
+      "step": 2871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.57465436088786e-06,
+      "loss": 0.0141,
+      "step": 2872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.571176459777431e-06,
+      "loss": 0.0141,
+      "step": 2873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5676987676703865e-06,
+      "loss": 0.0101,
+      "step": 2874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.564221286261709e-06,
+      "loss": 0.021,
+      "step": 2875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.560744017246284e-06,
+      "loss": 0.0112,
+      "step": 2876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.557266962318889e-06,
+      "loss": 0.0183,
+      "step": 2877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.553790123174198e-06,
+      "loss": 0.01,
+      "step": 2878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5503135015067815e-06,
+      "loss": 0.0213,
+      "step": 2879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.546837099011101e-06,
+      "loss": 0.0097,
+      "step": 2880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.543360917381512e-06,
+      "loss": 0.0099,
+      "step": 2881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.539884958312265e-06,
+      "loss": 0.0133,
+      "step": 2882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5364092234975e-06,
+      "loss": 0.0084,
+      "step": 2883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.532933714631248e-06,
+      "loss": 0.0106,
+      "step": 2884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.529458433407429e-06,
+      "loss": 0.0072,
+      "step": 2885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.525983381519853e-06,
+      "loss": 0.0123,
+      "step": 2886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.522508560662219e-06,
+      "loss": 0.0119,
+      "step": 2887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.519033972528114e-06,
+      "loss": 0.0109,
+      "step": 2888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5155596188110055e-06,
+      "loss": 0.0097,
+      "step": 2889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.512085501204254e-06,
+      "loss": 0.0114,
+      "step": 2890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.508611621401102e-06,
+      "loss": 0.0094,
+      "step": 2891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.505137981094675e-06,
+      "loss": 0.0093,
+      "step": 2892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5016645819779865e-06,
+      "loss": 0.0141,
+      "step": 2893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4981914257439254e-06,
+      "loss": 0.0604,
+      "step": 2894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.494718514085269e-06,
+      "loss": 0.0089,
+      "step": 2895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.491245848694669e-06,
+      "loss": 0.0157,
+      "step": 2896
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.487773431264664e-06,
+      "loss": 0.01,
+      "step": 2897
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.484301263487664e-06,
+      "loss": 0.0103,
+      "step": 2898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4808293470559645e-06,
+      "loss": 0.0105,
+      "step": 2899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.477357683661734e-06,
+      "loss": 0.0227,
+      "step": 2900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.473886274997018e-06,
+      "loss": 0.0084,
+      "step": 2901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.470415122753742e-06,
+      "loss": 0.0118,
+      "step": 2902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.466944228623701e-06,
+      "loss": 0.0162,
+      "step": 2903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.463473594298567e-06,
+      "loss": 0.0102,
+      "step": 2904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.460003221469886e-06,
+      "loss": 0.0131,
+      "step": 2905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.456533111829076e-06,
+      "loss": 0.008,
+      "step": 2906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.453063267067424e-06,
+      "loss": 0.0095,
+      "step": 2907
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.44959368887609e-06,
+      "loss": 0.0122,
+      "step": 2908
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.446124378946108e-06,
+      "loss": 0.0081,
+      "step": 2909
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.442655338968373e-06,
+      "loss": 0.0141,
+      "step": 2910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.439186570633656e-06,
+      "loss": 0.011,
+      "step": 2911
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4357180756325915e-06,
+      "loss": 0.0145,
+      "step": 2912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.432249855655681e-06,
+      "loss": 0.0086,
+      "step": 2913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.428781912393299e-06,
+      "loss": 0.0329,
+      "step": 2914
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.425314247535668e-06,
+      "loss": 0.0084,
+      "step": 2915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4218468627728935e-06,
+      "loss": 0.01,
+      "step": 2916
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.418379759794934e-06,
+      "loss": 0.0097,
+      "step": 2917
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.414912940291614e-06,
+      "loss": 0.0113,
+      "step": 2918
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4114464059526185e-06,
+      "loss": 0.0111,
+      "step": 2919
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4079801584674955e-06,
+      "loss": 0.012,
+      "step": 2920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.404514199525651e-06,
+      "loss": 0.0089,
+      "step": 2921
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.401048530816353e-06,
+      "loss": 0.0108,
+      "step": 2922
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.397583154028725e-06,
+      "loss": 0.0119,
+      "step": 2923
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.394118070851749e-06,
+      "loss": 0.009,
+      "step": 2924
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.390653282974264e-06,
+      "loss": 0.0117,
+      "step": 2925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.387188792084967e-06,
+      "loss": 0.0081,
+      "step": 2926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.383724599872407e-06,
+      "loss": 0.0075,
+      "step": 2927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.380260708024991e-06,
+      "loss": 0.0098,
+      "step": 2928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.376797118230978e-06,
+      "loss": 0.0088,
+      "step": 2929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.373333832178478e-06,
+      "loss": 0.0099,
+      "step": 2930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.369870851555457e-06,
+      "loss": 0.0091,
+      "step": 2931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.366408178049728e-06,
+      "loss": 0.0086,
+      "step": 2932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.362945813348956e-06,
+      "loss": 0.0102,
+      "step": 2933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.359483759140654e-06,
+      "loss": 0.0091,
+      "step": 2934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.356022017112187e-06,
+      "loss": 0.009,
+      "step": 2935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.352560588950766e-06,
+      "loss": 0.0125,
+      "step": 2936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.349099476343448e-06,
+      "loss": 0.01,
+      "step": 2937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.34563868097714e-06,
+      "loss": 0.0079,
+      "step": 2938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.342178204538588e-06,
+      "loss": 0.0087,
+      "step": 2939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3387180487143875e-06,
+      "loss": 0.0094,
+      "step": 2940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.335258215190979e-06,
+      "loss": 0.0145,
+      "step": 2941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.331798705654639e-06,
+      "loss": 0.0116,
+      "step": 2942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.328339521791493e-06,
+      "loss": 0.0125,
+      "step": 2943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3248806652875045e-06,
+      "loss": 0.0148,
+      "step": 2944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.321422137828479e-06,
+      "loss": 0.0115,
+      "step": 2945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.317963941100059e-06,
+      "loss": 0.0088,
+      "step": 2946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.314506076787729e-06,
+      "loss": 0.01,
+      "step": 2947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.31104854657681e-06,
+      "loss": 0.0091,
+      "step": 2948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.307591352152459e-06,
+      "loss": 0.0101,
+      "step": 2949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.304134495199675e-06,
+      "loss": 0.0069,
+      "step": 2950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.300677977403281e-06,
+      "loss": 0.0205,
+      "step": 2951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.297221800447946e-06,
+      "loss": 0.011,
+      "step": 2952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.293765966018167e-06,
+      "loss": 0.0088,
+      "step": 2953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.290310475798278e-06,
+      "loss": 0.013,
+      "step": 2954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.286855331472442e-06,
+      "loss": 0.0084,
+      "step": 2955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.283400534724654e-06,
+      "loss": 0.0151,
+      "step": 2956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.279946087238739e-06,
+      "loss": 0.0085,
+      "step": 2957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2764919906983545e-06,
+      "loss": 0.0096,
+      "step": 2958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.273038246786986e-06,
+      "loss": 0.0101,
+      "step": 2959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.269584857187942e-06,
+      "loss": 0.0114,
+      "step": 2960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.266131823584368e-06,
+      "loss": 0.0098,
+      "step": 2961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.262679147659227e-06,
+      "loss": 0.0353,
+      "step": 2962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.259226831095311e-06,
+      "loss": 0.0188,
+      "step": 2963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.255774875575239e-06,
+      "loss": 0.0103,
+      "step": 2964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2523232827814534e-06,
+      "loss": 0.0089,
+      "step": 2965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.248872054396215e-06,
+      "loss": 0.0076,
+      "step": 2966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.245421192101613e-06,
+      "loss": 0.0097,
+      "step": 2967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.241970697579557e-06,
+      "loss": 0.0118,
+      "step": 2968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.238520572511773e-06,
+      "loss": 0.009,
+      "step": 2969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.23507081857981e-06,
+      "loss": 0.0125,
+      "step": 2970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.23162143746504e-06,
+      "loss": 0.0097,
+      "step": 2971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.228172430848645e-06,
+      "loss": 0.0151,
+      "step": 2972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.224723800411631e-06,
+      "loss": 0.0089,
+      "step": 2973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.22127554783482e-06,
+      "loss": 0.0086,
+      "step": 2974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.217827674798845e-06,
+      "loss": 0.0114,
+      "step": 2975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2143801829841635e-06,
+      "loss": 0.0158,
+      "step": 2976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.210933074071033e-06,
+      "loss": 0.0094,
+      "step": 2977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.207486349739538e-06,
+      "loss": 0.0091,
+      "step": 2978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.204040011669567e-06,
+      "loss": 0.0113,
+      "step": 2979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.200594061540827e-06,
+      "loss": 0.0089,
+      "step": 2980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.197148501032829e-06,
+      "loss": 0.0102,
+      "step": 2981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.193703331824898e-06,
+      "loss": 0.0094,
+      "step": 2982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.190258555596168e-06,
+      "loss": 0.0094,
+      "step": 2983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.186814174025582e-06,
+      "loss": 0.009,
+      "step": 2984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.183370188791891e-06,
+      "loss": 0.0226,
+      "step": 2985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.179926601573645e-06,
+      "loss": 0.0104,
+      "step": 2986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.176483414049214e-06,
+      "loss": 0.0126,
+      "step": 2987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.173040627896762e-06,
+      "loss": 0.0143,
+      "step": 2988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.169598244794261e-06,
+      "loss": 0.0144,
+      "step": 2989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.166156266419489e-06,
+      "loss": 0.0176,
+      "step": 2990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.162714694450023e-06,
+      "loss": 0.0133,
+      "step": 2991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.159273530563243e-06,
+      "loss": 0.0142,
+      "step": 2992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.155832776436331e-06,
+      "loss": 0.0086,
+      "step": 2993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.15239243374627e-06,
+      "loss": 0.0115,
+      "step": 2994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.148952504169839e-06,
+      "loss": 0.0128,
+      "step": 2995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.145512989383618e-06,
+      "loss": 0.0069,
+      "step": 2996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.142073891063986e-06,
+      "loss": 0.0102,
+      "step": 2997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.138635210887117e-06,
+      "loss": 0.0122,
+      "step": 2998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.135196950528982e-06,
+      "loss": 0.0082,
+      "step": 2999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.131759111665349e-06,
+      "loss": 0.008,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.128321695971775e-06,
+      "loss": 0.0106,
+      "step": 3001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.124884705123619e-06,
+      "loss": 0.0095,
+      "step": 3002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.121448140796029e-06,
+      "loss": 0.0115,
+      "step": 3003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.118012004663939e-06,
+      "loss": 0.0122,
+      "step": 3004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.114576298402085e-06,
+      "loss": 0.0124,
+      "step": 3005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.111141023684986e-06,
+      "loss": 0.0079,
+      "step": 3006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.107706182186954e-06,
+      "loss": 0.0107,
+      "step": 3007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.104271775582089e-06,
+      "loss": 0.0101,
+      "step": 3008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.100837805544279e-06,
+      "loss": 0.008,
+      "step": 3009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0974042737472005e-06,
+      "loss": 0.0077,
+      "step": 3010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.093971181864313e-06,
+      "loss": 0.0092,
+      "step": 3011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.090538531568867e-06,
+      "loss": 0.0106,
+      "step": 3012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.087106324533891e-06,
+      "loss": 0.0094,
+      "step": 3013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.083674562432203e-06,
+      "loss": 0.0114,
+      "step": 3014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0802432469364e-06,
+      "loss": 0.0148,
+      "step": 3015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0768123797188665e-06,
+      "loss": 0.0139,
+      "step": 3016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.073381962451764e-06,
+      "loss": 0.0104,
+      "step": 3017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.069951996807034e-06,
+      "loss": 0.0139,
+      "step": 3018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.066522484456406e-06,
+      "loss": 0.0102,
+      "step": 3019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.063093427071376e-06,
+      "loss": 0.03,
+      "step": 3020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0596648263232315e-06,
+      "loss": 0.0106,
+      "step": 3021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0562366838830255e-06,
+      "loss": 0.0118,
+      "step": 3022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.052809001421595e-06,
+      "loss": 0.0106,
+      "step": 3023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0493817806095504e-06,
+      "loss": 0.0097,
+      "step": 3024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.045955023117276e-06,
+      "loss": 0.0117,
+      "step": 3025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.042528730614935e-06,
+      "loss": 0.0111,
+      "step": 3026
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.039102904772459e-06,
+      "loss": 0.0134,
+      "step": 3027
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.035677547259555e-06,
+      "loss": 0.009,
+      "step": 3028
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.032252659745699e-06,
+      "loss": 0.0148,
+      "step": 3029
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.028828243900141e-06,
+      "loss": 0.0115,
+      "step": 3030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.025404301391898e-06,
+      "loss": 0.01,
+      "step": 3031
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.02198083388976e-06,
+      "loss": 0.0095,
+      "step": 3032
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.018557843062282e-06,
+      "loss": 0.012,
+      "step": 3033
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.015135330577787e-06,
+      "loss": 0.008,
+      "step": 3034
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0117132981043695e-06,
+      "loss": 0.0143,
+      "step": 3035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0082917473098845e-06,
+      "loss": 0.0088,
+      "step": 3036
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.004870679861953e-06,
+      "loss": 0.0087,
+      "step": 3037
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.001450097427965e-06,
+      "loss": 0.0089,
+      "step": 3038
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9980300016750696e-06,
+      "loss": 0.0091,
+      "step": 3039
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.994610394270178e-06,
+      "loss": 0.01,
+      "step": 3040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991191276879966e-06,
+      "loss": 0.0097,
+      "step": 3041
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987772651170871e-06,
+      "loss": 0.0077,
+      "step": 3042
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98435451880909e-06,
+      "loss": 0.0091,
+      "step": 3043
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.980936881460576e-06,
+      "loss": 0.0125,
+      "step": 3044
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.977519740791049e-06,
+      "loss": 0.0088,
+      "step": 3045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.974103098465976e-06,
+      "loss": 0.0105,
+      "step": 3046
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.970686956150595e-06,
+      "loss": 0.0105,
+      "step": 3047
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.967271315509884e-06,
+      "loss": 0.0091,
+      "step": 3048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.963856178208588e-06,
+      "loss": 0.0127,
+      "step": 3049
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.960441545911205e-06,
+      "loss": 0.0128,
+      "step": 3050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.957027420281981e-06,
+      "loss": 0.0119,
+      "step": 3051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9536138029849244e-06,
+      "loss": 0.0174,
+      "step": 3052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.950200695683788e-06,
+      "loss": 0.0106,
+      "step": 3053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.94678810004208e-06,
+      "loss": 0.0126,
+      "step": 3054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.943376017723058e-06,
+      "loss": 0.0096,
+      "step": 3055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.939964450389728e-06,
+      "loss": 0.0068,
+      "step": 3056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.936553399704848e-06,
+      "loss": 0.0085,
+      "step": 3057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.933142867330921e-06,
+      "loss": 0.0098,
+      "step": 3058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9297328549302e-06,
+      "loss": 0.01,
+      "step": 3059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.926323364164684e-06,
+      "loss": 0.016,
+      "step": 3060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.922914396696118e-06,
+      "loss": 0.021,
+      "step": 3061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.91950595418599e-06,
+      "loss": 0.0088,
+      "step": 3062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9160980382955336e-06,
+      "loss": 0.0109,
+      "step": 3063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.912690650685726e-06,
+      "loss": 0.0098,
+      "step": 3064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.909283793017289e-06,
+      "loss": 0.0106,
+      "step": 3065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.905877466950679e-06,
+      "loss": 0.0081,
+      "step": 3066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.902471674146099e-06,
+      "loss": 0.0095,
+      "step": 3067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.899066416263493e-06,
+      "loss": 0.0099,
+      "step": 3068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.895661694962542e-06,
+      "loss": 0.0087,
+      "step": 3069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.892257511902664e-06,
+      "loss": 0.0104,
+      "step": 3070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.888853868743018e-06,
+      "loss": 0.0086,
+      "step": 3071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.885450767142498e-06,
+      "loss": 0.009,
+      "step": 3072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.882048208759735e-06,
+      "loss": 0.0105,
+      "step": 3073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8786461952530955e-06,
+      "loss": 0.028,
+      "step": 3074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.875244728280676e-06,
+      "loss": 0.0104,
+      "step": 3075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.871843809500313e-06,
+      "loss": 0.0137,
+      "step": 3076
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.868443440569571e-06,
+      "loss": 0.0108,
+      "step": 3077
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.865043623145751e-06,
+      "loss": 0.0078,
+      "step": 3078
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.86164435888588e-06,
+      "loss": 0.0098,
+      "step": 3079
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8582456494467214e-06,
+      "loss": 0.015,
+      "step": 3080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.854847496484762e-06,
+      "loss": 0.0086,
+      "step": 3081
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8514499016562216e-06,
+      "loss": 0.0096,
+      "step": 3082
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8480528666170495e-06,
+      "loss": 0.0308,
+      "step": 3083
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.844656393022912e-06,
+      "loss": 0.018,
+      "step": 3084
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.841260482529215e-06,
+      "loss": 0.0102,
+      "step": 3085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.83786513679108e-06,
+      "loss": 0.0091,
+      "step": 3086
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.834470357463362e-06,
+      "loss": 0.0102,
+      "step": 3087
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.831076146200633e-06,
+      "loss": 0.0076,
+      "step": 3088
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.827682504657187e-06,
+      "loss": 0.0089,
+      "step": 3089
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.82428943448705e-06,
+      "loss": 0.0099,
+      "step": 3090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.820896937343959e-06,
+      "loss": 0.012,
+      "step": 3091
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.817505014881378e-06,
+      "loss": 0.0102,
+      "step": 3092
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.814113668752486e-06,
+      "loss": 0.0881,
+      "step": 3093
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.810722900610186e-06,
+      "loss": 0.0095,
+      "step": 3094
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8073327121070968e-06,
+      "loss": 0.0111,
+      "step": 3095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8039431048955537e-06,
+      "loss": 0.0079,
+      "step": 3096
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8005540806276132e-06,
+      "loss": 0.0109,
+      "step": 3097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.797165640955041e-06,
+      "loss": 0.0092,
+      "step": 3098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.793777787529325e-06,
+      "loss": 0.0115,
+      "step": 3099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.790390522001662e-06,
+      "loss": 0.0097,
+      "step": 3100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.787003846022964e-06,
+      "loss": 0.0092,
+      "step": 3101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7836177612438557e-06,
+      "loss": 0.01,
+      "step": 3102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7802322693146726e-06,
+      "loss": 0.0103,
+      "step": 3103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.776847371885464e-06,
+      "loss": 0.009,
+      "step": 3104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7734630706059873e-06,
+      "loss": 0.0451,
+      "step": 3105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.77007936712571e-06,
+      "loss": 0.0102,
+      "step": 3106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7666962630938084e-06,
+      "loss": 0.0132,
+      "step": 3107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7633137601591647e-06,
+      "loss": 0.0081,
+      "step": 3108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.759931859970374e-06,
+      "loss": 0.0095,
+      "step": 3109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.756550564175727e-06,
+      "loss": 0.0096,
+      "step": 3110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7531698744232307e-06,
+      "loss": 0.0083,
+      "step": 3111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.74978979236059e-06,
+      "loss": 0.0163,
+      "step": 3112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7464103196352176e-06,
+      "loss": 0.0086,
+      "step": 3113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7430314578942263e-06,
+      "loss": 0.0072,
+      "step": 3114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7396532087844318e-06,
+      "loss": 0.007,
+      "step": 3115
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.736275573952354e-06,
+      "loss": 0.01,
+      "step": 3116
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7328985550442086e-06,
+      "loss": 0.0093,
+      "step": 3117
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7295221537059162e-06,
+      "loss": 0.0091,
+      "step": 3118
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7261463715830902e-06,
+      "loss": 0.0112,
+      "step": 3119
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7227712103210485e-06,
+      "loss": 0.013,
+      "step": 3120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7193966715648026e-06,
+      "loss": 0.0107,
+      "step": 3121
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.716022756959061e-06,
+      "loss": 0.0079,
+      "step": 3122
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7126494681482317e-06,
+      "loss": 0.014,
+      "step": 3123
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.709276806776412e-06,
+      "loss": 0.0104,
+      "step": 3124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.705904774487396e-06,
+      "loss": 0.0132,
+      "step": 3125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7025333729246733e-06,
+      "loss": 0.0219,
+      "step": 3126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.699162603731423e-06,
+      "loss": 0.0178,
+      "step": 3127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.695792468550517e-06,
+      "loss": 0.0079,
+      "step": 3128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6924229690245163e-06,
+      "loss": 0.0087,
+      "step": 3129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6890541067956775e-06,
+      "loss": 0.008,
+      "step": 3130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.68568588350594e-06,
+      "loss": 0.0113,
+      "step": 3131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6823183007969375e-06,
+      "loss": 0.0079,
+      "step": 3132
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.678951360309988e-06,
+      "loss": 0.0084,
+      "step": 3133
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6755850636860956e-06,
+      "loss": 0.0083,
+      "step": 3134
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.672219412565956e-06,
+      "loss": 0.0103,
+      "step": 3135
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.668854408589945e-06,
+      "loss": 0.0083,
+      "step": 3136
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6654900533981234e-06,
+      "loss": 0.0075,
+      "step": 3137
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6621263486302373e-06,
+      "loss": 0.0084,
+      "step": 3138
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6587632959257168e-06,
+      "loss": 0.0088,
+      "step": 3139
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.655400896923672e-06,
+      "loss": 0.0075,
+      "step": 3140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6520391532628953e-06,
+      "loss": 0.0157,
+      "step": 3141
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.648678066581861e-06,
+      "loss": 0.0211,
+      "step": 3142
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.645317638518721e-06,
+      "loss": 0.0145,
+      "step": 3143
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6419578707113055e-06,
+      "loss": 0.0153,
+      "step": 3144
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6385987647971287e-06,
+      "loss": 0.0125,
+      "step": 3145
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.635240322413375e-06,
+      "loss": 0.0084,
+      "step": 3146
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6318825451969085e-06,
+      "loss": 0.0082,
+      "step": 3147
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.628525434784268e-06,
+      "loss": 0.0121,
+      "step": 3148
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.625168992811671e-06,
+      "loss": 0.0098,
+      "step": 3149
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6218132209150047e-06,
+      "loss": 0.0084,
+      "step": 3150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.618458120729832e-06,
+      "loss": 0.0103,
+      "step": 3151
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6151036938913887e-06,
+      "loss": 0.0109,
+      "step": 3152
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.61174994203458e-06,
+      "loss": 0.0135,
+      "step": 3153
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.608396866793988e-06,
+      "loss": 0.0073,
+      "step": 3154
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6050444698038547e-06,
+      "loss": 0.0115,
+      "step": 3155
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6016927526981014e-06,
+      "loss": 0.0089,
+      "step": 3156
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.598341717110313e-06,
+      "loss": 0.0115,
+      "step": 3157
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5949913646737456e-06,
+      "loss": 0.0212,
+      "step": 3158
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5916416970213173e-06,
+      "loss": 0.0084,
+      "step": 3159
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5882927157856175e-06,
+      "loss": 0.0084,
+      "step": 3160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.584944422598899e-06,
+      "loss": 0.0128,
+      "step": 3161
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5815968190930793e-06,
+      "loss": 0.0112,
+      "step": 3162
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5782499068997386e-06,
+      "loss": 0.0095,
+      "step": 3163
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5749036876501196e-06,
+      "loss": 0.0094,
+      "step": 3164
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.571558162975133e-06,
+      "loss": 0.0105,
+      "step": 3165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.568213334505345e-06,
+      "loss": 0.009,
+      "step": 3166
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.564869203870982e-06,
+      "loss": 0.0137,
+      "step": 3167
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.561525772701937e-06,
+      "loss": 0.0137,
+      "step": 3168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5581830426277554e-06,
+      "loss": 0.0083,
+      "step": 3169
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5548410152776414e-06,
+      "loss": 0.0118,
+      "step": 3170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5514996922804636e-06,
+      "loss": 0.0078,
+      "step": 3171
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.548159075264738e-06,
+      "loss": 0.0129,
+      "step": 3172
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5448191658586423e-06,
+      "loss": 0.0078,
+      "step": 3173
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5414799656900057e-06,
+      "loss": 0.0125,
+      "step": 3174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.538141476386317e-06,
+      "loss": 0.0092,
+      "step": 3175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5348036995747135e-06,
+      "loss": 0.0102,
+      "step": 3176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.531466636881987e-06,
+      "loss": 0.0091,
+      "step": 3177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5281302899345825e-06,
+      "loss": 0.007,
+      "step": 3178
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.524794660358593e-06,
+      "loss": 0.0083,
+      "step": 3179
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.521459749779769e-06,
+      "loss": 0.0071,
+      "step": 3180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5181255598234963e-06,
+      "loss": 0.0112,
+      "step": 3181
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5147920921148267e-06,
+      "loss": 0.008,
+      "step": 3182
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.511459348278448e-06,
+      "loss": 0.0085,
+      "step": 3183
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.508127329938699e-06,
+      "loss": 0.0091,
+      "step": 3184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5047960387195673e-06,
+      "loss": 0.0105,
+      "step": 3185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.501465476244681e-06,
+      "loss": 0.0153,
+      "step": 3186
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.498135644137318e-06,
+      "loss": 0.0184,
+      "step": 3187
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4948065440203982e-06,
+      "loss": 0.0078,
+      "step": 3188
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.491478177516484e-06,
+      "loss": 0.0101,
+      "step": 3189
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.488150546247778e-06,
+      "loss": 0.0086,
+      "step": 3190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.484823651836131e-06,
+      "loss": 0.0094,
+      "step": 3191
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4814974959030294e-06,
+      "loss": 0.0099,
+      "step": 3192
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4781720800696006e-06,
+      "loss": 0.0084,
+      "step": 3193
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.474847405956613e-06,
+      "loss": 0.0103,
+      "step": 3194
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.471523475184472e-06,
+      "loss": 0.0077,
+      "step": 3195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4682002893732203e-06,
+      "loss": 0.0343,
+      "step": 3196
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.464877850142541e-06,
+      "loss": 0.0114,
+      "step": 3197
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4615561591117486e-06,
+      "loss": 0.0103,
+      "step": 3198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4582352178997937e-06,
+      "loss": 0.0102,
+      "step": 3199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 0.0102,
+      "step": 3200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4515955914063796e-06,
+      "loss": 0.0283,
+      "step": 3201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4482769093609945e-06,
+      "loss": 0.0084,
+      "step": 3202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.444958983606592e-06,
+      "loss": 0.0077,
+      "step": 3203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.441641815760291e-06,
+      "loss": 0.0102,
+      "step": 3204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4383254074388373e-06,
+      "loss": 0.0095,
+      "step": 3205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4350097602586085e-06,
+      "loss": 0.0096,
+      "step": 3206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4316948758356127e-06,
+      "loss": 0.0107,
+      "step": 3207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4283807557854814e-06,
+      "loss": 0.0142,
+      "step": 3208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4250674017234774e-06,
+      "loss": 0.0094,
+      "step": 3209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4217548152644887e-06,
+      "loss": 0.0091,
+      "step": 3210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4184429980230305e-06,
+      "loss": 0.0077,
+      "step": 3211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4151319516132414e-06,
+      "loss": 0.0179,
+      "step": 3212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.411821677648887e-06,
+      "loss": 0.0088,
+      "step": 3213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4085121777433532e-06,
+      "loss": 0.0097,
+      "step": 3214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.40520345350965e-06,
+      "loss": 0.0099,
+      "step": 3215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.401895506560411e-06,
+      "loss": 0.0085,
+      "step": 3216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.3985883385078875e-06,
+      "loss": 0.0088,
+      "step": 3217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.3952819509639534e-06,
+      "loss": 0.0083,
+      "step": 3218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.3919763455401016e-06,
+      "loss": 0.0118,
+      "step": 3219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.3886715238474454e-06,
+      "loss": 0.0136,
+      "step": 3220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.3853674874967134e-06,
+      "loss": 0.0093,
+      "step": 3221
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.3820642380982527e-06,
+      "loss": 0.0085,
+      "step": 3222
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.378761777262028e-06,
+      "loss": 0.0094,
+      "step": 3223
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.375460106597619e-06,
+      "loss": 0.0102,
+      "step": 3224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.372159227714218e-06,
+      "loss": 0.0133,
+      "step": 3225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3688591422206333e-06,
+      "loss": 0.0108,
+      "step": 3226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3655598517252886e-06,
+      "loss": 0.0083,
+      "step": 3227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3622613578362162e-06,
+      "loss": 0.0135,
+      "step": 3228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.358963662161062e-06,
+      "loss": 0.0068,
+      "step": 3229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.355666766307084e-06,
+      "loss": 0.0088,
+      "step": 3230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.352370671881148e-06,
+      "loss": 0.0114,
+      "step": 3231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3490753804897315e-06,
+      "loss": 0.0083,
+      "step": 3232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.34578089373892e-06,
+      "loss": 0.0078,
+      "step": 3233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3424872132344044e-06,
+      "loss": 0.0087,
+      "step": 3234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.339194340581485e-06,
+      "loss": 0.0088,
+      "step": 3235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3359022773850673e-06,
+      "loss": 0.0101,
+      "step": 3236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3326110252496652e-06,
+      "loss": 0.0089,
+      "step": 3237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3293205857793924e-06,
+      "loss": 0.0091,
+      "step": 3238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3260309605779717e-06,
+      "loss": 0.0097,
+      "step": 3239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.322742151248726e-06,
+      "loss": 0.0082,
+      "step": 3240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.319454159394578e-06,
+      "loss": 0.0087,
+      "step": 3241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.31616698661806e-06,
+      "loss": 0.0078,
+      "step": 3242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.312880634521295e-06,
+      "loss": 0.0083,
+      "step": 3243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3095951047060147e-06,
+      "loss": 0.0072,
+      "step": 3244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3063103987735433e-06,
+      "loss": 0.0078,
+      "step": 3245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.30302651832481e-06,
+      "loss": 0.0066,
+      "step": 3246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2997434649603368e-06,
+      "loss": 0.0085,
+      "step": 3247
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2964612402802422e-06,
+      "loss": 0.0098,
+      "step": 3248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.293179845884245e-06,
+      "loss": 0.0081,
+      "step": 3249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.289899283371657e-06,
+      "loss": 0.0094,
+      "step": 3250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2866195543413843e-06,
+      "loss": 0.0186,
+      "step": 3251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2833406603919243e-06,
+      "loss": 0.0148,
+      "step": 3252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.280062603121373e-06,
+      "loss": 0.0104,
+      "step": 3253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2767853841274154e-06,
+      "loss": 0.0094,
+      "step": 3254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.273509005007327e-06,
+      "loss": 0.0109,
+      "step": 3255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2702334673579765e-06,
+      "loss": 0.0136,
+      "step": 3256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.26695877277582e-06,
+      "loss": 0.0077,
+      "step": 3257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.263684922856905e-06,
+      "loss": 0.0096,
+      "step": 3258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.260411919196866e-06,
+      "loss": 0.0073,
+      "step": 3259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2571397633909252e-06,
+      "loss": 0.007,
+      "step": 3260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2538684570338908e-06,
+      "loss": 0.0086,
+      "step": 3261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2505980017201564e-06,
+      "loss": 0.0104,
+      "step": 3262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.247328399043706e-06,
+      "loss": 0.0091,
+      "step": 3263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2440596505981005e-06,
+      "loss": 0.0094,
+      "step": 3264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2407917579764914e-06,
+      "loss": 0.0069,
+      "step": 3265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2375247227716077e-06,
+      "loss": 0.0122,
+      "step": 3266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2342585465757625e-06,
+      "loss": 0.0079,
+      "step": 3267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.230993230980853e-06,
+      "loss": 0.0107,
+      "step": 3268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.227728777578353e-06,
+      "loss": 0.0398,
+      "step": 3269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.224465187959316e-06,
+      "loss": 0.0114,
+      "step": 3270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2212024637143756e-06,
+      "loss": 0.0092,
+      "step": 3271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.217940606433747e-06,
+      "loss": 0.0105,
+      "step": 3272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2146796177072183e-06,
+      "loss": 0.0303,
+      "step": 3273
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.211419499124154e-06,
+      "loss": 0.0093,
+      "step": 3274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2081602522734987e-06,
+      "loss": 0.01,
+      "step": 3275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2049018787437693e-06,
+      "loss": 0.0088,
+      "step": 3276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.201644380123056e-06,
+      "loss": 0.0088,
+      "step": 3277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1983877579990276e-06,
+      "loss": 0.0186,
+      "step": 3278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.195132013958918e-06,
+      "loss": 0.0101,
+      "step": 3279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1918771495895395e-06,
+      "loss": 0.0125,
+      "step": 3280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.188623166477272e-06,
+      "loss": 0.0092,
+      "step": 3281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.185370066208069e-06,
+      "loss": 0.0089,
+      "step": 3282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1821178503674515e-06,
+      "loss": 0.0102,
+      "step": 3283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.178866520540509e-06,
+      "loss": 0.0072,
+      "step": 3284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1756160783119015e-06,
+      "loss": 0.0072,
+      "step": 3285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1723665252658564e-06,
+      "loss": 0.0115,
+      "step": 3286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.169117862986163e-06,
+      "loss": 0.0073,
+      "step": 3287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.16587009305618e-06,
+      "loss": 0.0081,
+      "step": 3288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1626232170588343e-06,
+      "loss": 0.0081,
+      "step": 3289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1593772365766107e-06,
+      "loss": 0.0084,
+      "step": 3290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1561321531915622e-06,
+      "loss": 0.0091,
+      "step": 3291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.152887968485303e-06,
+      "loss": 0.0079,
+      "step": 3292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.149644684039008e-06,
+      "loss": 0.0303,
+      "step": 3293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1464023014334164e-06,
+      "loss": 0.0072,
+      "step": 3294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1431608222488276e-06,
+      "loss": 0.009,
+      "step": 3295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.139920248065095e-06,
+      "loss": 0.0076,
+      "step": 3296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1366805804616353e-06,
+      "loss": 0.0103,
+      "step": 3297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1334418210174268e-06,
+      "loss": 0.0089,
+      "step": 3298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.130203971310999e-06,
+      "loss": 0.0142,
+      "step": 3299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.12696703292044e-06,
+      "loss": 0.0083,
+      "step": 3300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1237310074233964e-06,
+      "loss": 0.0125,
+      "step": 3301
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1204958963970666e-06,
+      "loss": 0.0196,
+      "step": 3302
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.117261701418204e-06,
+      "loss": 0.0178,
+      "step": 3303
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.114028424063118e-06,
+      "loss": 0.0135,
+      "step": 3304
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.110796065907665e-06,
+      "loss": 0.0086,
+      "step": 3305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1075646285272608e-06,
+      "loss": 0.0091,
+      "step": 3306
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1043341134968653e-06,
+      "loss": 0.0077,
+      "step": 3307
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1011045223909954e-06,
+      "loss": 0.0071,
+      "step": 3308
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.097875856783713e-06,
+      "loss": 0.0223,
+      "step": 3309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.09464811824863e-06,
+      "loss": 0.0388,
+      "step": 3310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0914213083589086e-06,
+      "loss": 0.0131,
+      "step": 3311
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.088195428687254e-06,
+      "loss": 0.008,
+      "step": 3312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0849704808059266e-06,
+      "loss": 0.0082,
+      "step": 3313
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0817464662867192e-06,
+      "loss": 0.0083,
+      "step": 3314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.078523386700982e-06,
+      "loss": 0.0075,
+      "step": 3315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0753012436196033e-06,
+      "loss": 0.0092,
+      "step": 3316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0720800386130176e-06,
+      "loss": 0.0083,
+      "step": 3317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0688597732512004e-06,
+      "loss": 0.011,
+      "step": 3318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0656404491036696e-06,
+      "loss": 0.0312,
+      "step": 3319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0624220677394854e-06,
+      "loss": 0.0109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.059204630727247e-06,
+      "loss": 0.0087,
+      "step": 3321
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0559881396350967e-06,
+      "loss": 0.0129,
+      "step": 3322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0527725960307083e-06,
+      "loss": 0.0076,
+      "step": 3323
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.049558001481302e-06,
+      "loss": 0.0087,
+      "step": 3324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0463443575536324e-06,
+      "loss": 0.0075,
+      "step": 3325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.043131665813988e-06,
+      "loss": 0.0105,
+      "step": 3326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0399199278281986e-06,
+      "loss": 0.0077,
+      "step": 3327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0367091451616254e-06,
+      "loss": 0.0073,
+      "step": 3328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.033499319379163e-06,
+      "loss": 0.0074,
+      "step": 3329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.030290452045245e-06,
+      "loss": 0.0102,
+      "step": 3330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0270825447238316e-06,
+      "loss": 0.0111,
+      "step": 3331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.023875598978419e-06,
+      "loss": 0.0069,
+      "step": 3332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0206696163720317e-06,
+      "loss": 0.0088,
+      "step": 3333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0174645984672298e-06,
+      "loss": 0.0071,
+      "step": 3334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0142605468260976e-06,
+      "loss": 0.0329,
+      "step": 3335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.011057463010252e-06,
+      "loss": 0.0087,
+      "step": 3336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.007855348580837e-06,
+      "loss": 0.0074,
+      "step": 3337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.004654205098524e-06,
+      "loss": 0.0084,
+      "step": 3338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.001454034123512e-06,
+      "loss": 0.0098,
+      "step": 3339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9982548372155264e-06,
+      "loss": 0.0075,
+      "step": 3340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9950566159338146e-06,
+      "loss": 0.0112,
+      "step": 3341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.991859371837151e-06,
+      "loss": 0.0101,
+      "step": 3342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9886631064838355e-06,
+      "loss": 0.0113,
+      "step": 3343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9854678214316875e-06,
+      "loss": 0.0069,
+      "step": 3344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.98227351823805e-06,
+      "loss": 0.0087,
+      "step": 3345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9790801984597885e-06,
+      "loss": 0.0096,
+      "step": 3346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9758878636532884e-06,
+      "loss": 0.0101,
+      "step": 3347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.972696515374455e-06,
+      "loss": 0.0083,
+      "step": 3348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.969506155178711e-06,
+      "loss": 0.0106,
+      "step": 3349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.966316784621e-06,
+      "loss": 0.0095,
+      "step": 3350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.963128405255783e-06,
+      "loss": 0.0107,
+      "step": 3351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9599410186370363e-06,
+      "loss": 0.025,
+      "step": 3352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9567546263182554e-06,
+      "loss": 0.0093,
+      "step": 3353
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9535692298524477e-06,
+      "loss": 0.0086,
+      "step": 3354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9503848307921363e-06,
+      "loss": 0.0078,
+      "step": 3355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9472014306893605e-06,
+      "loss": 0.008,
+      "step": 3356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.94401903109567e-06,
+      "loss": 0.012,
+      "step": 3357
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.940837633562127e-06,
+      "loss": 0.0226,
+      "step": 3358
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9376572396393047e-06,
+      "loss": 0.0086,
+      "step": 3359
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.934477850877292e-06,
+      "loss": 0.016,
+      "step": 3360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.931299468825682e-06,
+      "loss": 0.0069,
+      "step": 3361
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.92812209503358e-06,
+      "loss": 0.0178,
+      "step": 3362
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9249457310495994e-06,
+      "loss": 0.0082,
+      "step": 3363
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.921770378421861e-06,
+      "loss": 0.0076,
+      "step": 3364
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.918596038697995e-06,
+      "loss": 0.009,
+      "step": 3365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.915422713425134e-06,
+      "loss": 0.0087,
+      "step": 3366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.912250404149918e-06,
+      "loss": 0.0093,
+      "step": 3367
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9090791124184934e-06,
+      "loss": 0.0123,
+      "step": 3368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.905908839776509e-06,
+      "loss": 0.0077,
+      "step": 3369
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9027395877691143e-06,
+      "loss": 0.0071,
+      "step": 3370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.899571357940969e-06,
+      "loss": 0.011,
+      "step": 3371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.896404151836227e-06,
+      "loss": 0.0077,
+      "step": 3372
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.893237970998547e-06,
+      "loss": 0.0095,
+      "step": 3373
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8900728169710866e-06,
+      "loss": 0.0085,
+      "step": 3374
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.886908691296504e-06,
+      "loss": 0.009,
+      "step": 3375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8837455955169547e-06,
+      "loss": 0.0086,
+      "step": 3376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8805835311740933e-06,
+      "loss": 0.0088,
+      "step": 3377
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.877422499809072e-06,
+      "loss": 0.008,
+      "step": 3378
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.874262502962537e-06,
+      "loss": 0.0069,
+      "step": 3379
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.871103542174637e-06,
+      "loss": 0.0085,
+      "step": 3380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8679456189850076e-06,
+      "loss": 0.0136,
+      "step": 3381
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.864788734932783e-06,
+      "loss": 0.0104,
+      "step": 3382
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8616328915565907e-06,
+      "loss": 0.0126,
+      "step": 3383
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.858478090394549e-06,
+      "loss": 0.0075,
+      "step": 3384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8553243329842715e-06,
+      "loss": 0.0089,
+      "step": 3385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8521716208628597e-06,
+      "loss": 0.0079,
+      "step": 3386
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.849019955566908e-06,
+      "loss": 0.0097,
+      "step": 3387
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8458693386325e-06,
+      "loss": 0.0122,
+      "step": 3388
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8427197715952047e-06,
+      "loss": 0.0085,
+      "step": 3389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.839571255990088e-06,
+      "loss": 0.0105,
+      "step": 3390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8364237933516964e-06,
+      "loss": 0.0089,
+      "step": 3391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8332773852140644e-06,
+      "loss": 0.0076,
+      "step": 3392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.830132033110713e-06,
+      "loss": 0.0077,
+      "step": 3393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.826987738574649e-06,
+      "loss": 0.0081,
+      "step": 3394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8238445031383634e-06,
+      "loss": 0.0078,
+      "step": 3395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8207023283338304e-06,
+      "loss": 0.0098,
+      "step": 3396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8175612156925082e-06,
+      "loss": 0.0084,
+      "step": 3397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.814421166745337e-06,
+      "loss": 0.0176,
+      "step": 3398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.811282183022736e-06,
+      "loss": 0.0113,
+      "step": 3399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8081442660546126e-06,
+      "loss": 0.0092,
+      "step": 3400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.805007417370347e-06,
+      "loss": 0.01,
+      "step": 3401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8018716384988034e-06,
+      "loss": 0.0106,
+      "step": 3402
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.798736930968315e-06,
+      "loss": 0.0156,
+      "step": 3403
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.795603296306708e-06,
+      "loss": 0.0077,
+      "step": 3404
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7924707360412743e-06,
+      "loss": 0.0129,
+      "step": 3405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7893392516987873e-06,
+      "loss": 0.0068,
+      "step": 3406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7862088448054936e-06,
+      "loss": 0.0109,
+      "step": 3407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7830795168871127e-06,
+      "loss": 0.011,
+      "step": 3408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.779951269468847e-06,
+      "loss": 0.0312,
+      "step": 3409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.776824104075364e-06,
+      "loss": 0.0109,
+      "step": 3410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7736980222308042e-06,
+      "loss": 0.008,
+      "step": 3411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7705730254587802e-06,
+      "loss": 0.0185,
+      "step": 3412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7674491152823825e-06,
+      "loss": 0.0094,
+      "step": 3413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7643262932241642e-06,
+      "loss": 0.0092,
+      "step": 3414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.761204560806152e-06,
+      "loss": 0.009,
+      "step": 3415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7580839195498397e-06,
+      "loss": 0.0086,
+      "step": 3416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.75496437097619e-06,
+      "loss": 0.0086,
+      "step": 3417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.75184591660563e-06,
+      "loss": 0.0083,
+      "step": 3418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7487285579580635e-06,
+      "loss": 0.0087,
+      "step": 3419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7456122965528475e-06,
+      "loss": 0.0083,
+      "step": 3420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.742497133908812e-06,
+      "loss": 0.0194,
+      "step": 3421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.739383071544246e-06,
+      "loss": 0.0085,
+      "step": 3422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.736270110976912e-06,
+      "loss": 0.0102,
+      "step": 3423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7331582537240243e-06,
+      "loss": 0.0099,
+      "step": 3424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7300475013022666e-06,
+      "loss": 0.0071,
+      "step": 3425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.726937855227781e-06,
+      "loss": 0.0068,
+      "step": 3426
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.723829317016169e-06,
+      "loss": 0.0097,
+      "step": 3427
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7207218881825016e-06,
+      "loss": 0.0083,
+      "step": 3428
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.717615570241294e-06,
+      "loss": 0.008,
+      "step": 3429
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.714510364706531e-06,
+      "loss": 0.0109,
+      "step": 3430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7114062730916513e-06,
+      "loss": 0.0065,
+      "step": 3431
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.708303296909551e-06,
+      "loss": 0.008,
+      "step": 3432
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.705201437672585e-06,
+      "loss": 0.007,
+      "step": 3433
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7021006968925613e-06,
+      "loss": 0.0072,
+      "step": 3434
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.699001076080742e-06,
+      "loss": 0.024,
+      "step": 3435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6959025767478466e-06,
+      "loss": 0.0069,
+      "step": 3436
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.692805200404044e-06,
+      "loss": 0.0079,
+      "step": 3437
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6897089485589584e-06,
+      "loss": 0.0091,
+      "step": 3438
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.686613822721666e-06,
+      "loss": 0.0085,
+      "step": 3439
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.683519824400693e-06,
+      "loss": 0.0115,
+      "step": 3440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.680426955104014e-06,
+      "loss": 0.0082,
+      "step": 3441
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.677335216339062e-06,
+      "loss": 0.0296,
+      "step": 3442
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6742446096127086e-06,
+      "loss": 0.0079,
+      "step": 3443
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.671155136431279e-06,
+      "loss": 0.0111,
+      "step": 3444
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6680667983005446e-06,
+      "loss": 0.0371,
+      "step": 3445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6649795967257243e-06,
+      "loss": 0.0071,
+      "step": 3446
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.661893533211482e-06,
+      "loss": 0.0204,
+      "step": 3447
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.658808609261928e-06,
+      "loss": 0.0088,
+      "step": 3448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6557248263806175e-06,
+      "loss": 0.0157,
+      "step": 3449
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6526421860705474e-06,
+      "loss": 0.0067,
+      "step": 3450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.649560689834158e-06,
+      "loss": 0.0157,
+      "step": 3451
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.646480339173337e-06,
+      "loss": 0.017,
+      "step": 3452
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6434011355894074e-06,
+      "loss": 0.0235,
+      "step": 3453
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.640323080583137e-06,
+      "loss": 0.0128,
+      "step": 3454
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.637246175654731e-06,
+      "loss": 0.0078,
+      "step": 3455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.634170422303835e-06,
+      "loss": 0.0069,
+      "step": 3456
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6310958220295356e-06,
+      "loss": 0.008,
+      "step": 3457
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6280223763303546e-06,
+      "loss": 0.0079,
+      "step": 3458
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6249500867042523e-06,
+      "loss": 0.0084,
+      "step": 3459
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6218789546486235e-06,
+      "loss": 0.0302,
+      "step": 3460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.618808981660304e-06,
+      "loss": 0.0101,
+      "step": 3461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.61574016923556e-06,
+      "loss": 0.0085,
+      "step": 3462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.612672518870093e-06,
+      "loss": 0.0072,
+      "step": 3463
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6096060320590393e-06,
+      "loss": 0.0126,
+      "step": 3464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6065407102969664e-06,
+      "loss": 0.0072,
+      "step": 3465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6034765550778753e-06,
+      "loss": 0.0108,
+      "step": 3466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.600413567895198e-06,
+      "loss": 0.007,
+      "step": 3467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5973517502417966e-06,
+      "loss": 0.0116,
+      "step": 3468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5942911036099657e-06,
+      "loss": 0.0086,
+      "step": 3469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5912316294914232e-06,
+      "loss": 0.0107,
+      "step": 3470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.588173329377324e-06,
+      "loss": 0.0086,
+      "step": 3471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5851162047582477e-06,
+      "loss": 0.008,
+      "step": 3472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.582060257124195e-06,
+      "loss": 0.0079,
+      "step": 3473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5790054879645964e-06,
+      "loss": 0.0098,
+      "step": 3474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5759518987683154e-06,
+      "loss": 0.0078,
+      "step": 3475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5728994910236304e-06,
+      "loss": 0.0102,
+      "step": 3476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5698482662182494e-06,
+      "loss": 0.0152,
+      "step": 3477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5667982258393016e-06,
+      "loss": 0.0198,
+      "step": 3478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5637493713733376e-06,
+      "loss": 0.0112,
+      "step": 3479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.560701704306336e-06,
+      "loss": 0.0061,
+      "step": 3480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.557655226123693e-06,
+      "loss": 0.009,
+      "step": 3481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5546099383102206e-06,
+      "loss": 0.0159,
+      "step": 3482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5515658423501573e-06,
+      "loss": 0.0076,
+      "step": 3483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5485229397271567e-06,
+      "loss": 0.0111,
+      "step": 3484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.545481231924296e-06,
+      "loss": 0.0086,
+      "step": 3485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5424407204240653e-06,
+      "loss": 0.0084,
+      "step": 3486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.539401406708373e-06,
+      "loss": 0.0083,
+      "step": 3487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.536363292258543e-06,
+      "loss": 0.0065,
+      "step": 3488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.533326378555314e-06,
+      "loss": 0.0077,
+      "step": 3489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5302906670788463e-06,
+      "loss": 0.0085,
+      "step": 3490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.527256159308703e-06,
+      "loss": 0.0065,
+      "step": 3491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.524222856723869e-06,
+      "loss": 0.0078,
+      "step": 3492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5211907608027366e-06,
+      "loss": 0.0141,
+      "step": 3493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.518159873023116e-06,
+      "loss": 0.0085,
+      "step": 3494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5151301948622235e-06,
+      "loss": 0.0069,
+      "step": 3495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5121017277966875e-06,
+      "loss": 0.0066,
+      "step": 3496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.509074473302546e-06,
+      "loss": 0.0066,
+      "step": 3497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.506048432855247e-06,
+      "loss": 0.0108,
+      "step": 3498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5030236079296443e-06,
+      "loss": 0.006,
+      "step": 3499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 0.0099,
+      "step": 3500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.496977610539988e-06,
+      "loss": 0.0099,
+      "step": 3501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.49395644102268e-06,
+      "loss": 0.0087,
+      "step": 3502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4909364929205575e-06,
+      "loss": 0.0077,
+      "step": 3503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.48791776770551e-06,
+      "loss": 0.0085,
+      "step": 3504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.484900266848825e-06,
+      "loss": 0.0066,
+      "step": 3505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4818839918211963e-06,
+      "loss": 0.0098,
+      "step": 3506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4788689440927193e-06,
+      "loss": 0.0077,
+      "step": 3507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4758551251328923e-06,
+      "loss": 0.0076,
+      "step": 3508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4728425364106136e-06,
+      "loss": 0.0082,
+      "step": 3509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.469831179394182e-06,
+      "loss": 0.0064,
+      "step": 3510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4668210555512974e-06,
+      "loss": 0.0085,
+      "step": 3511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4638121663490546e-06,
+      "loss": 0.008,
+      "step": 3512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4608045132539536e-06,
+      "loss": 0.0107,
+      "step": 3513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4577980977318866e-06,
+      "loss": 0.009,
+      "step": 3514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4547929212481436e-06,
+      "loss": 0.0095,
+      "step": 3515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4517889852674114e-06,
+      "loss": 0.0084,
+      "step": 3516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.448786291253772e-06,
+      "loss": 0.0098,
+      "step": 3517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4457848406707014e-06,
+      "loss": 0.0064,
+      "step": 3518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.442784634981071e-06,
+      "loss": 0.0069,
+      "step": 3519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4397856756471435e-06,
+      "loss": 0.0076,
+      "step": 3520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4367879641305757e-06,
+      "loss": 0.0171,
+      "step": 3521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4337915018924147e-06,
+      "loss": 0.008,
+      "step": 3522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4307962903931025e-06,
+      "loss": 0.0213,
+      "step": 3523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4278023310924676e-06,
+      "loss": 0.0076,
+      "step": 3524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.424809625449729e-06,
+      "loss": 0.0082,
+      "step": 3525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4218181749234954e-06,
+      "loss": 0.0067,
+      "step": 3526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.418827980971763e-06,
+      "loss": 0.0062,
+      "step": 3527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.415839045051916e-06,
+      "loss": 0.0079,
+      "step": 3528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.412851368620726e-06,
+      "loss": 0.0091,
+      "step": 3529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.40986495313435e-06,
+      "loss": 0.0106,
+      "step": 3530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4068798000483306e-06,
+      "loss": 0.0078,
+      "step": 3531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.403895910817593e-06,
+      "loss": 0.0094,
+      "step": 3532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4009132868964525e-06,
+      "loss": 0.0076,
+      "step": 3533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3979319297386035e-06,
+      "loss": 0.0088,
+      "step": 3534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.39495184079712e-06,
+      "loss": 0.0071,
+      "step": 3535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.391973021524461e-06,
+      "loss": 0.0089,
+      "step": 3536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3889954733724708e-06,
+      "loss": 0.0093,
+      "step": 3537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3860191977923673e-06,
+      "loss": 0.0078,
+      "step": 3538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3830441962347528e-06,
+      "loss": 0.0066,
+      "step": 3539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.380070470149605e-06,
+      "loss": 0.0066,
+      "step": 3540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3770980209862814e-06,
+      "loss": 0.0077,
+      "step": 3541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3741268501935212e-06,
+      "loss": 0.009,
+      "step": 3542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3711569592194363e-06,
+      "loss": 0.011,
+      "step": 3543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3681883495115114e-06,
+      "loss": 0.0072,
+      "step": 3544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3652210225166122e-06,
+      "loss": 0.0067,
+      "step": 3545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3622549796809807e-06,
+      "loss": 0.009,
+      "step": 3546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3592902224502284e-06,
+      "loss": 0.0078,
+      "step": 3547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.356326752269342e-06,
+      "loss": 0.0086,
+      "step": 3548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.353364570582681e-06,
+      "loss": 0.0114,
+      "step": 3549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3504036788339763e-06,
+      "loss": 0.0182,
+      "step": 3550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3474440784663287e-06,
+      "loss": 0.0082,
+      "step": 3551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.344485770922218e-06,
+      "loss": 0.0074,
+      "step": 3552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3415287576434807e-06,
+      "loss": 0.0097,
+      "step": 3553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.338573040071332e-06,
+      "loss": 0.0084,
+      "step": 3554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3356186196463497e-06,
+      "loss": 0.0075,
+      "step": 3555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3326654978084872e-06,
+      "loss": 0.0085,
+      "step": 3556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.329713675997058e-06,
+      "loss": 0.0095,
+      "step": 3557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3267631556507443e-06,
+      "loss": 0.007,
+      "step": 3558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.323813938207593e-06,
+      "loss": 0.0114,
+      "step": 3559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.320866025105016e-06,
+      "loss": 0.0095,
+      "step": 3560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3179194177797954e-06,
+      "loss": 0.0085,
+      "step": 3561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3149741176680666e-06,
+      "loss": 0.0116,
+      "step": 3562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.312030126205335e-06,
+      "loss": 0.0131,
+      "step": 3563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.309087444826464e-06,
+      "loss": 0.0101,
+      "step": 3564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3061460749656844e-06,
+      "loss": 0.0086,
+      "step": 3565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.303206018056583e-06,
+      "loss": 0.0096,
+      "step": 3566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3002672755321076e-06,
+      "loss": 0.008,
+      "step": 3567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.297329848824565e-06,
+      "loss": 0.0161,
+      "step": 3568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.294393739365621e-06,
+      "loss": 0.0091,
+      "step": 3569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2914589485863015e-06,
+      "loss": 0.0066,
+      "step": 3570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.288525477916986e-06,
+      "loss": 0.009,
+      "step": 3571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.285593328787414e-06,
+      "loss": 0.0107,
+      "step": 3572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.282662502626678e-06,
+      "loss": 0.0126,
+      "step": 3573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2797330008632255e-06,
+      "loss": 0.0087,
+      "step": 3574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2768048249248648e-06,
+      "loss": 0.0081,
+      "step": 3575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.27387797623875e-06,
+      "loss": 0.0193,
+      "step": 3576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2709524562313923e-06,
+      "loss": 0.0114,
+      "step": 3577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.268028266328655e-06,
+      "loss": 0.009,
+      "step": 3578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.265105407955752e-06,
+      "loss": 0.0061,
+      "step": 3579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2621838825372496e-06,
+      "loss": 0.0069,
+      "step": 3580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2592636914970633e-06,
+      "loss": 0.0082,
+      "step": 3581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.256344836258459e-06,
+      "loss": 0.0103,
+      "step": 3582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2534273182440515e-06,
+      "loss": 0.0172,
+      "step": 3583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.250511138875801e-06,
+      "loss": 0.0151,
+      "step": 3584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2475962995750224e-06,
+      "loss": 0.0063,
+      "step": 3585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.24468280176237e-06,
+      "loss": 0.0095,
+      "step": 3586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2417706468578495e-06,
+      "loss": 0.0082,
+      "step": 3587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2388598362808074e-06,
+      "loss": 0.008,
+      "step": 3588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.235950371449938e-06,
+      "loss": 0.0084,
+      "step": 3589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.23304225378328e-06,
+      "loss": 0.0069,
+      "step": 3590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2301354846982148e-06,
+      "loss": 0.0071,
+      "step": 3591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2272300656114648e-06,
+      "loss": 0.0066,
+      "step": 3592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.224325997939095e-06,
+      "loss": 0.0072,
+      "step": 3593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.221423283096517e-06,
+      "loss": 0.0094,
+      "step": 3594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.218521922498476e-06,
+      "loss": 0.0076,
+      "step": 3595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2156219175590623e-06,
+      "loss": 0.0077,
+      "step": 3596
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.212723269691697e-06,
+      "loss": 0.0167,
+      "step": 3597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.209825980309151e-06,
+      "loss": 0.0075,
+      "step": 3598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2069300508235273e-06,
+      "loss": 0.0075,
+      "step": 3599
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.204035482646267e-06,
+      "loss": 0.0061,
+      "step": 3600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.201142277188146e-06,
+      "loss": 0.014,
+      "step": 3601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1982504358592777e-06,
+      "loss": 0.0063,
+      "step": 3602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.19535996006911e-06,
+      "loss": 0.0087,
+      "step": 3603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.192470851226428e-06,
+      "loss": 0.0065,
+      "step": 3604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1895831107393485e-06,
+      "loss": 0.0073,
+      "step": 3605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1866967400153184e-06,
+      "loss": 0.0157,
+      "step": 3606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.183811740461118e-06,
+      "loss": 0.0068,
+      "step": 3607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1809281134828663e-06,
+      "loss": 0.011,
+      "step": 3608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1780458604860056e-06,
+      "loss": 0.01,
+      "step": 3609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.175164982875311e-06,
+      "loss": 0.0087,
+      "step": 3610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1722854820548873e-06,
+      "loss": 0.0065,
+      "step": 3611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1694073594281663e-06,
+      "loss": 0.0115,
+      "step": 3612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1665306163979132e-06,
+      "loss": 0.0074,
+      "step": 3613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1636552543662187e-06,
+      "loss": 0.0075,
+      "step": 3614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1607812747344955e-06,
+      "loss": 0.009,
+      "step": 3615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.157908678903487e-06,
+      "loss": 0.0067,
+      "step": 3616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1550374682732605e-06,
+      "loss": 0.088,
+      "step": 3617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.152167644243213e-06,
+      "loss": 0.0077,
+      "step": 3618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.14929920821206e-06,
+      "loss": 0.0138,
+      "step": 3619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.146432161577842e-06,
+      "loss": 0.0263,
+      "step": 3620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1435665057379233e-06,
+      "loss": 0.0093,
+      "step": 3621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.140702242088987e-06,
+      "loss": 0.0072,
+      "step": 3622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.137839372027047e-06,
+      "loss": 0.0074,
+      "step": 3623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.134977896947425e-06,
+      "loss": 0.0089,
+      "step": 3624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.132117818244771e-06,
+      "loss": 0.0093,
+      "step": 3625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1292591373130515e-06,
+      "loss": 0.0088,
+      "step": 3626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1264018555455563e-06,
+      "loss": 0.0079,
+      "step": 3627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1235459743348874e-06,
+      "loss": 0.006,
+      "step": 3628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1206914950729673e-06,
+      "loss": 0.0057,
+      "step": 3629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1178384191510344e-06,
+      "loss": 0.0066,
+      "step": 3630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.114986747959643e-06,
+      "loss": 0.0088,
+      "step": 3631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.112136482888663e-06,
+      "loss": 0.0142,
+      "step": 3632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1092876253272793e-06,
+      "loss": 0.0075,
+      "step": 3633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.10644017666399e-06,
+      "loss": 0.0072,
+      "step": 3634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.103594138286607e-06,
+      "loss": 0.0083,
+      "step": 3635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.100749511582254e-06,
+      "loss": 0.0067,
+      "step": 3636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.09790629793737e-06,
+      "loss": 0.0055,
+      "step": 3637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.095064498737701e-06,
+      "loss": 0.0076,
+      "step": 3638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0922241153683064e-06,
+      "loss": 0.007,
+      "step": 3639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0893851492135536e-06,
+      "loss": 0.0073,
+      "step": 3640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0865476016571206e-06,
+      "loss": 0.019,
+      "step": 3641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.083711474081993e-06,
+      "loss": 0.0087,
+      "step": 3642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.080876767870466e-06,
+      "loss": 0.0096,
+      "step": 3643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.07804348440414e-06,
+      "loss": 0.0073,
+      "step": 3644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.075211625063923e-06,
+      "loss": 0.0075,
+      "step": 3645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0723811912300295e-06,
+      "loss": 0.0086,
+      "step": 3646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0695521842819788e-06,
+      "loss": 0.0093,
+      "step": 3647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.066724605598594e-06,
+      "loss": 0.0087,
+      "step": 3648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.063898456558002e-06,
+      "loss": 0.0066,
+      "step": 3649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.061073738537635e-06,
+      "loss": 0.0121,
+      "step": 3650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0582504529142248e-06,
+      "loss": 0.0105,
+      "step": 3651
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0554286010638076e-06,
+      "loss": 0.0114,
+      "step": 3652
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0526081843617183e-06,
+      "loss": 0.0069,
+      "step": 3653
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.049789204182596e-06,
+      "loss": 0.0058,
+      "step": 3654
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.046971661900373e-06,
+      "loss": 0.0108,
+      "step": 3655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.04415555888829e-06,
+      "loss": 0.0113,
+      "step": 3656
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.04134089651888e-06,
+      "loss": 0.0097,
+      "step": 3657
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0385276761639768e-06,
+      "loss": 0.0062,
+      "step": 3658
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.035715899194704e-06,
+      "loss": 0.0158,
+      "step": 3659
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0329055669814936e-06,
+      "loss": 0.0083,
+      "step": 3660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.030096680894065e-06,
+      "loss": 0.0065,
+      "step": 3661
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.027289242301435e-06,
+      "loss": 0.0101,
+      "step": 3662
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0244832525719155e-06,
+      "loss": 0.0164,
+      "step": 3663
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.02167871307311e-06,
+      "loss": 0.009,
+      "step": 3664
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0188756251719204e-06,
+      "loss": 0.0085,
+      "step": 3665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.016073990234536e-06,
+      "loss": 0.0086,
+      "step": 3666
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0132738096264415e-06,
+      "loss": 0.0061,
+      "step": 3667
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0104750847124075e-06,
+      "loss": 0.0085,
+      "step": 3668
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.007677816856498e-06,
+      "loss": 0.0081,
+      "step": 3669
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0048820074220716e-06,
+      "loss": 0.0103,
+      "step": 3670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.002087657771769e-06,
+      "loss": 0.0084,
+      "step": 3671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.999294769267523e-06,
+      "loss": 0.007,
+      "step": 3672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.996503343270554e-06,
+      "loss": 0.0076,
+      "step": 3673
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9937133811413666e-06,
+      "loss": 0.0092,
+      "step": 3674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.990924884239758e-06,
+      "loss": 0.0075,
+      "step": 3675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.988137853924808e-06,
+      "loss": 0.006,
+      "step": 3676
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9853522915548777e-06,
+      "loss": 0.0068,
+      "step": 3677
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9825681984876173e-06,
+      "loss": 0.0241,
+      "step": 3678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.979785576079961e-06,
+      "loss": 0.0077,
+      "step": 3679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.977004425688126e-06,
+      "loss": 0.0107,
+      "step": 3680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.97422474866761e-06,
+      "loss": 0.0103,
+      "step": 3681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9714465463731934e-06,
+      "loss": 0.0068,
+      "step": 3682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9686698201589395e-06,
+      "loss": 0.0068,
+      "step": 3683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9658945713781883e-06,
+      "loss": 0.0156,
+      "step": 3684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9631208013835677e-06,
+      "loss": 0.0085,
+      "step": 3685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9603485115269743e-06,
+      "loss": 0.0113,
+      "step": 3686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9575777031595906e-06,
+      "loss": 0.0082,
+      "step": 3687
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9548083776318727e-06,
+      "loss": 0.0099,
+      "step": 3688
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.95204053629356e-06,
+      "loss": 0.0238,
+      "step": 3689
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9492741804936623e-06,
+      "loss": 0.0089,
+      "step": 3690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.946509311580469e-06,
+      "loss": 0.0079,
+      "step": 3691
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9437459309015426e-06,
+      "loss": 0.0067,
+      "step": 3692
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.94098403980372e-06,
+      "loss": 0.0159,
+      "step": 3693
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.938223639633119e-06,
+      "loss": 0.0104,
+      "step": 3694
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9354647317351187e-06,
+      "loss": 0.0091,
+      "step": 3695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.93270731745438e-06,
+      "loss": 0.0086,
+      "step": 3696
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.929951398134832e-06,
+      "loss": 0.0078,
+      "step": 3697
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927196975119678e-06,
+      "loss": 0.0071,
+      "step": 3698
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9244440497513895e-06,
+      "loss": 0.009,
+      "step": 3699
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9216926233717087e-06,
+      "loss": 0.0085,
+      "step": 3700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9189426973216478e-06,
+      "loss": 0.0085,
+      "step": 3701
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9161942729414876e-06,
+      "loss": 0.0117,
+      "step": 3702
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.913447351570776e-06,
+      "loss": 0.0087,
+      "step": 3703
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.910701934548329e-06,
+      "loss": 0.0076,
+      "step": 3704
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.90795802321223e-06,
+      "loss": 0.0097,
+      "step": 3705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9052156188998284e-06,
+      "loss": 0.0131,
+      "step": 3706
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9024747229477365e-06,
+      "loss": 0.008,
+      "step": 3707
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8997353366918369e-06,
+      "loss": 0.016,
+      "step": 3708
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.896997461467272e-06,
+      "loss": 0.0057,
+      "step": 3709
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8942610986084487e-06,
+      "loss": 0.009,
+      "step": 3710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8915262494490366e-06,
+      "loss": 0.0106,
+      "step": 3711
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8887929153219687e-06,
+      "loss": 0.0082,
+      "step": 3712
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8860610975594384e-06,
+      "loss": 0.0096,
+      "step": 3713
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8833307974929006e-06,
+      "loss": 0.0064,
+      "step": 3714
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8806020164530702e-06,
+      "loss": 0.0073,
+      "step": 3715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8778747557699223e-06,
+      "loss": 0.0089,
+      "step": 3716
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8751490167726888e-06,
+      "loss": 0.0072,
+      "step": 3717
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8724248007898648e-06,
+      "loss": 0.0223,
+      "step": 3718
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8697021091491991e-06,
+      "loss": 0.0069,
+      "step": 3719
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8669809431776991e-06,
+      "loss": 0.0062,
+      "step": 3720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8642613042016245e-06,
+      "loss": 0.006,
+      "step": 3721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8615431935464984e-06,
+      "loss": 0.0092,
+      "step": 3722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8588266125370929e-06,
+      "loss": 0.0168,
+      "step": 3723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8561115624974374e-06,
+      "loss": 0.0084,
+      "step": 3724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8533980447508138e-06,
+      "loss": 0.0101,
+      "step": 3725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8506860606197564e-06,
+      "loss": 0.0079,
+      "step": 3726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8479756114260562e-06,
+      "loss": 0.0077,
+      "step": 3727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8452666984907519e-06,
+      "loss": 0.0092,
+      "step": 3728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.842559323134136e-06,
+      "loss": 0.0139,
+      "step": 3729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8398534866757455e-06,
+      "loss": 0.011,
+      "step": 3730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.837149190434378e-06,
+      "loss": 0.0087,
+      "step": 3731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8344464357280722e-06,
+      "loss": 0.011,
+      "step": 3732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.831745223874118e-06,
+      "loss": 0.0081,
+      "step": 3733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.829045556189053e-06,
+      "loss": 0.0088,
+      "step": 3734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8263474339886628e-06,
+      "loss": 0.0071,
+      "step": 3735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8236508585879781e-06,
+      "loss": 0.0098,
+      "step": 3736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8209558313012792e-06,
+      "loss": 0.0081,
+      "step": 3737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8182623534420906e-06,
+      "loss": 0.0095,
+      "step": 3738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8155704263231777e-06,
+      "loss": 0.0065,
+      "step": 3739
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8128800512565514e-06,
+      "loss": 0.0066,
+      "step": 3740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.810191229553473e-06,
+      "loss": 0.0077,
+      "step": 3741
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.807503962524439e-06,
+      "loss": 0.0126,
+      "step": 3742
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8048182514791901e-06,
+      "loss": 0.0094,
+      "step": 3743
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8021340977267104e-06,
+      "loss": 0.0111,
+      "step": 3744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.799451502575222e-06,
+      "loss": 0.0127,
+      "step": 3745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7967704673321917e-06,
+      "loss": 0.0089,
+      "step": 3746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7940909933043243e-06,
+      "loss": 0.0104,
+      "step": 3747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7914130817975595e-06,
+      "loss": 0.0102,
+      "step": 3748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7887367341170781e-06,
+      "loss": 0.0073,
+      "step": 3749
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7860619515673034e-06,
+      "loss": 0.0072,
+      "step": 3750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7833887354518902e-06,
+      "loss": 0.0079,
+      "step": 3751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7807170870737317e-06,
+      "loss": 0.0258,
+      "step": 3752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7780470077349566e-06,
+      "loss": 0.0077,
+      "step": 3753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7753784987369287e-06,
+      "loss": 0.0087,
+      "step": 3754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7727115613802465e-06,
+      "loss": 0.0096,
+      "step": 3755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.770046196964747e-06,
+      "loss": 0.0071,
+      "step": 3756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7673824067894912e-06,
+      "loss": 0.0064,
+      "step": 3757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7647201921527802e-06,
+      "loss": 0.0112,
+      "step": 3758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.762059554352143e-06,
+      "loss": 0.0073,
+      "step": 3759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7594004946843458e-06,
+      "loss": 0.0113,
+      "step": 3760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7567430144453801e-06,
+      "loss": 0.0074,
+      "step": 3761
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75408711493047e-06,
+      "loss": 0.0085,
+      "step": 3762
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.751432797434068e-06,
+      "loss": 0.0119,
+      "step": 3763
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7487800632498547e-06,
+      "loss": 0.0094,
+      "step": 3764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.746128913670746e-06,
+      "loss": 0.0061,
+      "step": 3765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7434793499888746e-06,
+      "loss": 0.0069,
+      "step": 3766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7408313734956074e-06,
+      "loss": 0.0165,
+      "step": 3767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.738184985481536e-06,
+      "loss": 0.0058,
+      "step": 3768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7355401872364759e-06,
+      "loss": 0.0085,
+      "step": 3769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7328969800494727e-06,
+      "loss": 0.0077,
+      "step": 3770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7302553652087927e-06,
+      "loss": 0.0065,
+      "step": 3771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.727615344001926e-06,
+      "loss": 0.0457,
+      "step": 3772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7249769177155879e-06,
+      "loss": 0.0077,
+      "step": 3773
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7223400876357144e-06,
+      "loss": 0.0082,
+      "step": 3774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7197048550474643e-06,
+      "loss": 0.0072,
+      "step": 3775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7170712212352187e-06,
+      "loss": 0.0094,
+      "step": 3776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7144391874825784e-06,
+      "loss": 0.0119,
+      "step": 3777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7118087550723633e-06,
+      "loss": 0.007,
+      "step": 3778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.709179925286617e-06,
+      "loss": 0.0094,
+      "step": 3779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7065526994065973e-06,
+      "loss": 0.0075,
+      "step": 3780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7039270787127832e-06,
+      "loss": 0.0076,
+      "step": 3781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7013030644848698e-06,
+      "loss": 0.0081,
+      "step": 3782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6986806580017695e-06,
+      "loss": 0.0098,
+      "step": 3783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6960598605416117e-06,
+      "loss": 0.0089,
+      "step": 3784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6934406733817417e-06,
+      "loss": 0.0117,
+      "step": 3785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6908230977987184e-06,
+      "loss": 0.013,
+      "step": 3786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6882071350683165e-06,
+      "loss": 0.0135,
+      "step": 3787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6855927864655241e-06,
+      "loss": 0.0131,
+      "step": 3788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6829800532645447e-06,
+      "loss": 0.0076,
+      "step": 3789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.680368936738792e-06,
+      "loss": 0.0083,
+      "step": 3790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6777594381608936e-06,
+      "loss": 0.0086,
+      "step": 3791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6751515588026828e-06,
+      "loss": 0.0089,
+      "step": 3792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6725452999352137e-06,
+      "loss": 0.0129,
+      "step": 3793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6699406628287423e-06,
+      "loss": 0.0086,
+      "step": 3794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6673376487527382e-06,
+      "loss": 0.007,
+      "step": 3795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6647362589758787e-06,
+      "loss": 0.0153,
+      "step": 3796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6621364947660472e-06,
+      "loss": 0.0088,
+      "step": 3797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6595383573903412e-06,
+      "loss": 0.0089,
+      "step": 3798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6569418481150596e-06,
+      "loss": 0.0057,
+      "step": 3799
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6543469682057105e-06,
+      "loss": 0.0074,
+      "step": 3800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6517537189270043e-06,
+      "loss": 0.0105,
+      "step": 3801
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6491621015428588e-06,
+      "loss": 0.0073,
+      "step": 3802
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6465721173164e-06,
+      "loss": 0.0054,
+      "step": 3803
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.643983767509954e-06,
+      "loss": 0.0131,
+      "step": 3804
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6413970533850498e-06,
+      "loss": 0.0077,
+      "step": 3805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6388119762024213e-06,
+      "loss": 0.0065,
+      "step": 3806
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6362285372220016e-06,
+      "loss": 0.0079,
+      "step": 3807
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6336467377029308e-06,
+      "loss": 0.0088,
+      "step": 3808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6310665789035468e-06,
+      "loss": 0.0087,
+      "step": 3809
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6284880620813847e-06,
+      "loss": 0.0094,
+      "step": 3810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6259111884931817e-06,
+      "loss": 0.0083,
+      "step": 3811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6233359593948777e-06,
+      "loss": 0.0063,
+      "step": 3812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6207623760416074e-06,
+      "loss": 0.0111,
+      "step": 3813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6181904396877041e-06,
+      "loss": 0.009,
+      "step": 3814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6156201515866971e-06,
+      "loss": 0.0102,
+      "step": 3815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6130515129913144e-06,
+      "loss": 0.0058,
+      "step": 3816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6104845251534772e-06,
+      "loss": 0.0083,
+      "step": 3817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6079191893243102e-06,
+      "loss": 0.018,
+      "step": 3818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.605355506754121e-06,
+      "loss": 0.0111,
+      "step": 3819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6027934786924187e-06,
+      "loss": 0.0061,
+      "step": 3820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.600233106387904e-06,
+      "loss": 0.0105,
+      "step": 3821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.597674391088474e-06,
+      "loss": 0.0078,
+      "step": 3822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5951173340412134e-06,
+      "loss": 0.0074,
+      "step": 3823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5925619364924016e-06,
+      "loss": 0.008,
+      "step": 3824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5900081996875083e-06,
+      "loss": 0.0105,
+      "step": 3825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.587456124871191e-06,
+      "loss": 0.0084,
+      "step": 3826
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5849057132873063e-06,
+      "loss": 0.0102,
+      "step": 3827
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.582356966178888e-06,
+      "loss": 0.0079,
+      "step": 3828
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5798098847881664e-06,
+      "loss": 0.0064,
+      "step": 3829
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5772644703565564e-06,
+      "loss": 0.0064,
+      "step": 3830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5747207241246654e-06,
+      "loss": 0.008,
+      "step": 3831
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5721786473322825e-06,
+      "loss": 0.0087,
+      "step": 3832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5696382412183853e-06,
+      "loss": 0.008,
+      "step": 3833
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.567099507021137e-06,
+      "loss": 0.0073,
+      "step": 3834
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5645624459778858e-06,
+      "loss": 0.0069,
+      "step": 3835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5620270593251635e-06,
+      "loss": 0.0066,
+      "step": 3836
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5594933482986885e-06,
+      "loss": 0.0105,
+      "step": 3837
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.556961314133359e-06,
+      "loss": 0.0066,
+      "step": 3838
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.554430958063259e-06,
+      "loss": 0.0075,
+      "step": 3839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.551902281321651e-06,
+      "loss": 0.0067,
+      "step": 3840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5493752851409844e-06,
+      "loss": 0.0086,
+      "step": 3841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5468499707528856e-06,
+      "loss": 0.0071,
+      "step": 3842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5443263393881619e-06,
+      "loss": 0.0251,
+      "step": 3843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5418043922768e-06,
+      "loss": 0.0107,
+      "step": 3844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5392841306479667e-06,
+      "loss": 0.0064,
+      "step": 3845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5367655557300066e-06,
+      "loss": 0.0065,
+      "step": 3846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5342486687504432e-06,
+      "loss": 0.0062,
+      "step": 3847
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.531733470935976e-06,
+      "loss": 0.0154,
+      "step": 3848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.529219963512481e-06,
+      "loss": 0.0078,
+      "step": 3849
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5267081477050132e-06,
+      "loss": 0.0141,
+      "step": 3850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5241980247378008e-06,
+      "loss": 0.0066,
+      "step": 3851
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.521689595834246e-06,
+      "loss": 0.0159,
+      "step": 3852
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.519182862216929e-06,
+      "loss": 0.0085,
+      "step": 3853
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5166778251075964e-06,
+      "loss": 0.0078,
+      "step": 3854
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.514174485727178e-06,
+      "loss": 0.0082,
+      "step": 3855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5116728452957686e-06,
+      "loss": 0.0076,
+      "step": 3856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5091729050326376e-06,
+      "loss": 0.0075,
+      "step": 3857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5066746661562254e-06,
+      "loss": 0.0148,
+      "step": 3858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5041781298841424e-06,
+      "loss": 0.0151,
+      "step": 3859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5016832974331725e-06,
+      "loss": 0.0077,
+      "step": 3860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4991901700192657e-06,
+      "loss": 0.0083,
+      "step": 3861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.496698748857543e-06,
+      "loss": 0.0061,
+      "step": 3862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4942090351622884e-06,
+      "loss": 0.0103,
+      "step": 3863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.491721030146963e-06,
+      "loss": 0.0085,
+      "step": 3864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.489234735024188e-06,
+      "loss": 0.0105,
+      "step": 3865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4867501510057548e-06,
+      "loss": 0.0086,
+      "step": 3866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.484267279302618e-06,
+      "loss": 0.009,
+      "step": 3867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4817861211248996e-06,
+      "loss": 0.0066,
+      "step": 3868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4793066776818843e-06,
+      "loss": 0.0058,
+      "step": 3869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4768289501820265e-06,
+      "loss": 0.0063,
+      "step": 3870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4743529398329393e-06,
+      "loss": 0.0069,
+      "step": 3871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4718786478413983e-06,
+      "loss": 0.0076,
+      "step": 3872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.469406075413342e-06,
+      "loss": 0.0063,
+      "step": 3873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4669352237538763e-06,
+      "loss": 0.0082,
+      "step": 3874
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.0079,
+      "step": 3875
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4619986875569247e-06,
+      "loss": 0.0082,
+      "step": 3876
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.459533005425446e-06,
+      "loss": 0.0093,
+      "step": 3877
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4570690488745687e-06,
+      "loss": 0.0068,
+      "step": 3878
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4546068191051988e-06,
+      "loss": 0.0085,
+      "step": 3879
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4521463173173966e-06,
+      "loss": 0.0072,
+      "step": 3880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4496875447103781e-06,
+      "loss": 0.0061,
+      "step": 3881
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4472305024825189e-06,
+      "loss": 0.0092,
+      "step": 3882
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4447751918313552e-06,
+      "loss": 0.0289,
+      "step": 3883
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4423216139535735e-06,
+      "loss": 0.0075,
+      "step": 3884
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4398697700450181e-06,
+      "loss": 0.0073,
+      "step": 3885
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4374196613006874e-06,
+      "loss": 0.01,
+      "step": 3886
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4349712889147355e-06,
+      "loss": 0.0097,
+      "step": 3887
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4325246540804672e-06,
+      "loss": 0.0094,
+      "step": 3888
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4300797579903476e-06,
+      "loss": 0.009,
+      "step": 3889
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4276366018359845e-06,
+      "loss": 0.0133,
+      "step": 3890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4251951868081438e-06,
+      "loss": 0.0083,
+      "step": 3891
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4227555140967402e-06,
+      "loss": 0.0084,
+      "step": 3892
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.420317584890844e-06,
+      "loss": 0.0085,
+      "step": 3893
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4178814003786706e-06,
+      "loss": 0.0059,
+      "step": 3894
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4154469617475864e-06,
+      "loss": 0.0077,
+      "step": 3895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4130142701841076e-06,
+      "loss": 0.0082,
+      "step": 3896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4105833268738966e-06,
+      "loss": 0.0115,
+      "step": 3897
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4081541330017706e-06,
+      "loss": 0.0065,
+      "step": 3898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4057266897516842e-06,
+      "loss": 0.0078,
+      "step": 3899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4033009983067454e-06,
+      "loss": 0.0125,
+      "step": 3900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4008770598492072e-06,
+      "loss": 0.0281,
+      "step": 3901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3984548755604655e-06,
+      "loss": 0.0084,
+      "step": 3902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3960344466210669e-06,
+      "loss": 0.0066,
+      "step": 3903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3936157742106977e-06,
+      "loss": 0.0158,
+      "step": 3904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3911988595081894e-06,
+      "loss": 0.011,
+      "step": 3905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3887837036915169e-06,
+      "loss": 0.0065,
+      "step": 3906
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3863703079377971e-06,
+      "loss": 0.0118,
+      "step": 3907
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3839586734232907e-06,
+      "loss": 0.0072,
+      "step": 3908
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3815488013233986e-06,
+      "loss": 0.0081,
+      "step": 3909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3791406928126638e-06,
+      "loss": 0.0089,
+      "step": 3910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3767343490647668e-06,
+      "loss": 0.0106,
+      "step": 3911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3743297712525334e-06,
+      "loss": 0.0091,
+      "step": 3912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3719269605479241e-06,
+      "loss": 0.0139,
+      "step": 3913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3695259181220405e-06,
+      "loss": 0.0074,
+      "step": 3914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3671266451451209e-06,
+      "loss": 0.0098,
+      "step": 3915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3647291427865417e-06,
+      "loss": 0.0056,
+      "step": 3916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3623334122148164e-06,
+      "loss": 0.0083,
+      "step": 3917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3599394545975952e-06,
+      "loss": 0.0118,
+      "step": 3918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3575472711016634e-06,
+      "loss": 0.0059,
+      "step": 3919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3551568628929434e-06,
+      "loss": 0.0081,
+      "step": 3920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3527682311364886e-06,
+      "loss": 0.0079,
+      "step": 3921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3503813769964923e-06,
+      "loss": 0.011,
+      "step": 3922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3479963016362768e-06,
+      "loss": 0.0066,
+      "step": 3923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3456130062183003e-06,
+      "loss": 0.0096,
+      "step": 3924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3432314919041478e-06,
+      "loss": 0.0102,
+      "step": 3925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3408517598545446e-06,
+      "loss": 0.0081,
+      "step": 3926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3384738112293415e-06,
+      "loss": 0.0071,
+      "step": 3927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3360976471875226e-06,
+      "loss": 0.0171,
+      "step": 3928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.333723268887201e-06,
+      "loss": 0.0075,
+      "step": 3929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3313506774856177e-06,
+      "loss": 0.0073,
+      "step": 3930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3289798741391486e-06,
+      "loss": 0.0069,
+      "step": 3931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3266108600032928e-06,
+      "loss": 0.0065,
+      "step": 3932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3242436362326804e-06,
+      "loss": 0.0057,
+      "step": 3933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3218782039810634e-06,
+      "loss": 0.0059,
+      "step": 3934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3195145644013286e-06,
+      "loss": 0.0056,
+      "step": 3935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.317152718645484e-06,
+      "loss": 0.0109,
+      "step": 3936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.314792667864665e-06,
+      "loss": 0.0059,
+      "step": 3937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.312434413209131e-06,
+      "loss": 0.008,
+      "step": 3938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3100779558282673e-06,
+      "loss": 0.0063,
+      "step": 3939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3077232968705805e-06,
+      "loss": 0.0068,
+      "step": 3940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3053704374837063e-06,
+      "loss": 0.0068,
+      "step": 3941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3030193788143991e-06,
+      "loss": 0.0073,
+      "step": 3942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3006701220085338e-06,
+      "loss": 0.006,
+      "step": 3943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2983226682111094e-06,
+      "loss": 0.0077,
+      "step": 3944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2959770185662502e-06,
+      "loss": 0.0147,
+      "step": 3945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2936331742171943e-06,
+      "loss": 0.0069,
+      "step": 3946
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2912911363063048e-06,
+      "loss": 0.0269,
+      "step": 3947
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2889509059750605e-06,
+      "loss": 0.0286,
+      "step": 3948
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2866124843640614e-06,
+      "loss": 0.0094,
+      "step": 3949
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2842758726130283e-06,
+      "loss": 0.0089,
+      "step": 3950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2819410718607972e-06,
+      "loss": 0.0064,
+      "step": 3951
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2796080832453183e-06,
+      "loss": 0.0045,
+      "step": 3952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2772769079036639e-06,
+      "loss": 0.0058,
+      "step": 3953
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2749475469720196e-06,
+      "loss": 0.0062,
+      "step": 3954
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2726200015856893e-06,
+      "loss": 0.0069,
+      "step": 3955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2702942728790897e-06,
+      "loss": 0.0076,
+      "step": 3956
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2679703619857525e-06,
+      "loss": 0.0055,
+      "step": 3957
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2656482700383238e-06,
+      "loss": 0.0115,
+      "step": 3958
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2633279981685608e-06,
+      "loss": 0.0071,
+      "step": 3959
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2610095475073415e-06,
+      "loss": 0.0141,
+      "step": 3960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2586929191846453e-06,
+      "loss": 0.0084,
+      "step": 3961
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2563781143295705e-06,
+      "loss": 0.0666,
+      "step": 3962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2540651340703231e-06,
+      "loss": 0.0079,
+      "step": 3963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2517539795342248e-06,
+      "loss": 0.0064,
+      "step": 3964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2494446518477022e-06,
+      "loss": 0.0058,
+      "step": 3965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2471371521362946e-06,
+      "loss": 0.0081,
+      "step": 3966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2448314815246487e-06,
+      "loss": 0.0109,
+      "step": 3967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.24252764113652e-06,
+      "loss": 0.0092,
+      "step": 3968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.240225632094773e-06,
+      "loss": 0.0086,
+      "step": 3969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2379254555213788e-06,
+      "loss": 0.0068,
+      "step": 3970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2356271125374153e-06,
+      "loss": 0.0086,
+      "step": 3971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2333306042630672e-06,
+      "loss": 0.0074,
+      "step": 3972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2310359318176229e-06,
+      "loss": 0.0067,
+      "step": 3973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2287430963194807e-06,
+      "loss": 0.0057,
+      "step": 3974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.22645209888614e-06,
+      "loss": 0.0089,
+      "step": 3975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2241629406342048e-06,
+      "loss": 0.0084,
+      "step": 3976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2218756226793827e-06,
+      "loss": 0.0083,
+      "step": 3977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2195901461364851e-06,
+      "loss": 0.0069,
+      "step": 3978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.217306512119425e-06,
+      "loss": 0.0155,
+      "step": 3979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2150247217412186e-06,
+      "loss": 0.0067,
+      "step": 3980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2127447761139821e-06,
+      "loss": 0.0105,
+      "step": 3981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2104666763489326e-06,
+      "loss": 0.0082,
+      "step": 3982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2081904235563908e-06,
+      "loss": 0.007,
+      "step": 3983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2059160188457724e-06,
+      "loss": 0.0074,
+      "step": 3984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.203643463325596e-06,
+      "loss": 0.0078,
+      "step": 3985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2013727581034783e-06,
+      "loss": 0.0083,
+      "step": 3986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.199103904286129e-06,
+      "loss": 0.0101,
+      "step": 3987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1968369029793642e-06,
+      "loss": 0.0086,
+      "step": 3988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1945717552880919e-06,
+      "loss": 0.008,
+      "step": 3989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1923084623163172e-06,
+      "loss": 0.0076,
+      "step": 3990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1900470251671415e-06,
+      "loss": 0.0106,
+      "step": 3991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.18778744494276e-06,
+      "loss": 0.005,
+      "step": 3992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.185529722744469e-06,
+      "loss": 0.0058,
+      "step": 3993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1832738596726518e-06,
+      "loss": 0.0073,
+      "step": 3994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1810198568267906e-06,
+      "loss": 0.0058,
+      "step": 3995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.178767715305455e-06,
+      "loss": 0.0062,
+      "step": 3996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1765174362063152e-06,
+      "loss": 0.0071,
+      "step": 3997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1742690206261293e-06,
+      "loss": 0.0065,
+      "step": 3998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1720224696607474e-06,
+      "loss": 0.0081,
+      "step": 3999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1697777844051105e-06,
+      "loss": 0.0078,
+      "step": 4000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1675349659532514e-06,
+      "loss": 0.0071,
+      "step": 4001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1652940153982917e-06,
+      "loss": 0.0062,
+      "step": 4002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1630549338324454e-06,
+      "loss": 0.0098,
+      "step": 4003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.160817722347014e-06,
+      "loss": 0.0069,
+      "step": 4004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1585823820323845e-06,
+      "loss": 0.0073,
+      "step": 4005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1563489139780344e-06,
+      "loss": 0.007,
+      "step": 4006
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.154117319272532e-06,
+      "loss": 0.0092,
+      "step": 4007
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1518875990035278e-06,
+      "loss": 0.0164,
+      "step": 4008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1496597542577603e-06,
+      "loss": 0.0075,
+      "step": 4009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1474337861210543e-06,
+      "loss": 0.0061,
+      "step": 4010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1452096956783181e-06,
+      "loss": 0.0221,
+      "step": 4011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1429874840135492e-06,
+      "loss": 0.0075,
+      "step": 4012
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1407671522098262e-06,
+      "loss": 0.0076,
+      "step": 4013
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1385487013493095e-06,
+      "loss": 0.0096,
+      "step": 4014
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.136332132513245e-06,
+      "loss": 0.0119,
+      "step": 4015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1341174467819637e-06,
+      "loss": 0.0061,
+      "step": 4016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1319046452348758e-06,
+      "loss": 0.0058,
+      "step": 4017
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.129693728950474e-06,
+      "loss": 0.006,
+      "step": 4018
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1274846990063314e-06,
+      "loss": 0.0085,
+      "step": 4019
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1252775564791023e-06,
+      "loss": 0.0084,
+      "step": 4020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1230723024445212e-06,
+      "loss": 0.0085,
+      "step": 4021
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.120868937977404e-06,
+      "loss": 0.0064,
+      "step": 4022
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1186674641516415e-06,
+      "loss": 0.0066,
+      "step": 4023
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1164678820402059e-06,
+      "loss": 0.0061,
+      "step": 4024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1142701927151456e-06,
+      "loss": 0.0128,
+      "step": 4025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.11207439724759e-06,
+      "loss": 0.0242,
+      "step": 4026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1098804967077425e-06,
+      "loss": 0.0098,
+      "step": 4027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1076884921648834e-06,
+      "loss": 0.0092,
+      "step": 4028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1054983846873684e-06,
+      "loss": 0.0073,
+      "step": 4029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1033101753426285e-06,
+      "loss": 0.0077,
+      "step": 4030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1011238651971744e-06,
+      "loss": 0.0075,
+      "step": 4031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0989394553165833e-06,
+      "loss": 0.0189,
+      "step": 4032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0967569467655104e-06,
+      "loss": 0.0092,
+      "step": 4033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0945763406076837e-06,
+      "loss": 0.0071,
+      "step": 4034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0923976379059059e-06,
+      "loss": 0.0064,
+      "step": 4035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.09022083972205e-06,
+      "loss": 0.0085,
+      "step": 4036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0880459471170597e-06,
+      "loss": 0.0074,
+      "step": 4037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0858729611509516e-06,
+      "loss": 0.0059,
+      "step": 4038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0837018828828133e-06,
+      "loss": 0.0083,
+      "step": 4039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0815327133708015e-06,
+      "loss": 0.0071,
+      "step": 4040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0793654536721432e-06,
+      "loss": 0.0069,
+      "step": 4041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.077200104843134e-06,
+      "loss": 0.0054,
+      "step": 4042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0750366679391393e-06,
+      "loss": 0.0064,
+      "step": 4043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0728751440145907e-06,
+      "loss": 0.0093,
+      "step": 4044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0707155341229902e-06,
+      "loss": 0.0075,
+      "step": 4045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0685578393169054e-06,
+      "loss": 0.0057,
+      "step": 4046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0664020606479702e-06,
+      "loss": 0.0081,
+      "step": 4047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.064248199166884e-06,
+      "loss": 0.0071,
+      "step": 4048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0620962559234144e-06,
+      "loss": 0.006,
+      "step": 4049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0599462319663906e-06,
+      "loss": 0.0095,
+      "step": 4050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0577981283437095e-06,
+      "loss": 0.0068,
+      "step": 4051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0556519461023301e-06,
+      "loss": 0.0071,
+      "step": 4052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.053507686288276e-06,
+      "loss": 0.0065,
+      "step": 4053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0513653499466315e-06,
+      "loss": 0.0101,
+      "step": 4054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.049224938121548e-06,
+      "loss": 0.0095,
+      "step": 4055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.047086451856235e-06,
+      "loss": 0.0101,
+      "step": 4056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0449498921929669e-06,
+      "loss": 0.007,
+      "step": 4057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0428152601730718e-06,
+      "loss": 0.0247,
+      "step": 4058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0406825568369478e-06,
+      "loss": 0.0065,
+      "step": 4059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0385517832240472e-06,
+      "loss": 0.0063,
+      "step": 4060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.036422940372883e-06,
+      "loss": 0.0096,
+      "step": 4061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0342960293210281e-06,
+      "loss": 0.0099,
+      "step": 4062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0321710511051108e-06,
+      "loss": 0.0086,
+      "step": 4063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0300480067608232e-06,
+      "loss": 0.0066,
+      "step": 4064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0279268973229089e-06,
+      "loss": 0.0125,
+      "step": 4065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0258077238251735e-06,
+      "loss": 0.0092,
+      "step": 4066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0236904873004722e-06,
+      "loss": 0.0097,
+      "step": 4067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0215751887807228e-06,
+      "loss": 0.0096,
+      "step": 4068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0194618292968972e-06,
+      "loss": 0.0083,
+      "step": 4069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0173504098790188e-06,
+      "loss": 0.0088,
+      "step": 4070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0152409315561696e-06,
+      "loss": 0.0074,
+      "step": 4071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0131333953564825e-06,
+      "loss": 0.0075,
+      "step": 4072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0110278023071445e-06,
+      "loss": 0.0079,
+      "step": 4073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0089241534343986e-06,
+      "loss": 0.0079,
+      "step": 4074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.006822449763537e-06,
+      "loss": 0.0134,
+      "step": 4075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0047226923189024e-06,
+      "loss": 0.009,
+      "step": 4076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0026248821238915e-06,
+      "loss": 0.0077,
+      "step": 4077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0005290202009533e-06,
+      "loss": 0.0095,
+      "step": 4078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.984351075715848e-07,
+      "loss": 0.0082,
+      "step": 4079
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.963431452563331e-07,
+      "loss": 0.0067,
+      "step": 4080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.942531342747953e-07,
+      "loss": 0.0072,
+      "step": 4081
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.921650756456164e-07,
+      "loss": 0.0115,
+      "step": 4082
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.900789703864933e-07,
+      "loss": 0.0065,
+      "step": 4083
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.879948195141681e-07,
+      "loss": 0.0062,
+      "step": 4084
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.859126240444284e-07,
+      "loss": 0.0056,
+      "step": 4085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.838323849921123e-07,
+      "loss": 0.0064,
+      "step": 4086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.81754103371101e-07,
+      "loss": 0.0061,
+      "step": 4087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.79677780194327e-07,
+      "loss": 0.0068,
+      "step": 4088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.77603416473763e-07,
+      "loss": 0.0118,
+      "step": 4089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.7553101322043e-07,
+      "loss": 0.008,
+      "step": 4090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.734605714443906e-07,
+      "loss": 0.0067,
+      "step": 4091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.713920921547532e-07,
+      "loss": 0.0069,
+      "step": 4092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.69325576359672e-07,
+      "loss": 0.0081,
+      "step": 4093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.67261025066339e-07,
+      "loss": 0.0151,
+      "step": 4094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.651984392809916e-07,
+      "loss": 0.0058,
+      "step": 4095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.631378200089082e-07,
+      "loss": 0.0067,
+      "step": 4096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.610791682544123e-07,
+      "loss": 0.0151,
+      "step": 4097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.590224850208645e-07,
+      "loss": 0.0236,
+      "step": 4098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.569677713106673e-07,
+      "loss": 0.0055,
+      "step": 4099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.549150281252633e-07,
+      "loss": 0.0091,
+      "step": 4100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.528642564651341e-07,
+      "loss": 0.0051,
+      "step": 4101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.508154573298012e-07,
+      "loss": 0.0082,
+      "step": 4102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.487686317178241e-07,
+      "loss": 0.0066,
+      "step": 4103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.467237806268009e-07,
+      "loss": 0.007,
+      "step": 4104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.446809050533679e-07,
+      "loss": 0.0105,
+      "step": 4105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.426400059931956e-07,
+      "loss": 0.0161,
+      "step": 4106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.406010844409957e-07,
+      "loss": 0.0103,
+      "step": 4107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.385641413905139e-07,
+      "loss": 0.006,
+      "step": 4108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.365291778345303e-07,
+      "loss": 0.0096,
+      "step": 4109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.344961947648624e-07,
+      "loss": 0.0059,
+      "step": 4110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.3246519317236e-07,
+      "loss": 0.0064,
+      "step": 4111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.304361740469103e-07,
+      "loss": 0.0059,
+      "step": 4112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.284091383774313e-07,
+      "loss": 0.0056,
+      "step": 4113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.263840871518759e-07,
+      "loss": 0.0081,
+      "step": 4114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.243610213572285e-07,
+      "loss": 0.0089,
+      "step": 4115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.223399419795093e-07,
+      "loss": 0.0068,
+      "step": 4116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.203208500037664e-07,
+      "loss": 0.0065,
+      "step": 4117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.183037464140804e-07,
+      "loss": 0.0067,
+      "step": 4118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.162886321935632e-07,
+      "loss": 0.0074,
+      "step": 4119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.142755083243577e-07,
+      "loss": 0.0061,
+      "step": 4120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.122643757876354e-07,
+      "loss": 0.0071,
+      "step": 4121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.10255235563598e-07,
+      "loss": 0.0082,
+      "step": 4122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.08248088631476e-07,
+      "loss": 0.0069,
+      "step": 4123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.06242935969528e-07,
+      "loss": 0.0156,
+      "step": 4124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.042397785550405e-07,
+      "loss": 0.0115,
+      "step": 4125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.022386173643305e-07,
+      "loss": 0.0063,
+      "step": 4126
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.002394533727382e-07,
+      "loss": 0.0092,
+      "step": 4127
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.982422875546332e-07,
+      "loss": 0.008,
+      "step": 4128
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.962471208834056e-07,
+      "loss": 0.0047,
+      "step": 4129
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.942539543314799e-07,
+      "loss": 0.0111,
+      "step": 4130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.922627888703e-07,
+      "loss": 0.0064,
+      "step": 4131
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.902736254703347e-07,
+      "loss": 0.0075,
+      "step": 4132
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.882864651010798e-07,
+      "loss": 0.008,
+      "step": 4133
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.863013087310502e-07,
+      "loss": 0.0074,
+      "step": 4134
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.843181573277904e-07,
+      "loss": 0.0057,
+      "step": 4135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.823370118578628e-07,
+      "loss": 0.0068,
+      "step": 4136
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.803578732868545e-07,
+      "loss": 0.0055,
+      "step": 4137
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.783807425793722e-07,
+      "loss": 0.0068,
+      "step": 4138
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.764056206990446e-07,
+      "loss": 0.0057,
+      "step": 4139
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.744325086085248e-07,
+      "loss": 0.0086,
+      "step": 4140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.72461407269482e-07,
+      "loss": 0.0097,
+      "step": 4141
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.704923176426072e-07,
+      "loss": 0.0131,
+      "step": 4142
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.685252406876116e-07,
+      "loss": 0.0081,
+      "step": 4143
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.665601773632226e-07,
+      "loss": 0.0062,
+      "step": 4144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.645971286271903e-07,
+      "loss": 0.008,
+      "step": 4145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.626360954362817e-07,
+      "loss": 0.0062,
+      "step": 4146
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.606770787462776e-07,
+      "loss": 0.0076,
+      "step": 4147
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.587200795119793e-07,
+      "loss": 0.0109,
+      "step": 4148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.567650986872061e-07,
+      "loss": 0.0064,
+      "step": 4149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.54812137224792e-07,
+      "loss": 0.0063,
+      "step": 4150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.528611960765853e-07,
+      "loss": 0.0054,
+      "step": 4151
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.509122761934519e-07,
+      "loss": 0.0081,
+      "step": 4152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.489653785252711e-07,
+      "loss": 0.0176,
+      "step": 4153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.470205040209362e-07,
+      "loss": 0.0075,
+      "step": 4154
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.450776536283594e-07,
+      "loss": 0.0067,
+      "step": 4155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.431368282944585e-07,
+      "loss": 0.0074,
+      "step": 4156
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.411980289651689e-07,
+      "loss": 0.0073,
+      "step": 4157
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.392612565854374e-07,
+      "loss": 0.0072,
+      "step": 4158
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.373265120992252e-07,
+      "loss": 0.0082,
+      "step": 4159
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.353937964495029e-07,
+      "loss": 0.0829,
+      "step": 4160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.334631105782515e-07,
+      "loss": 0.0087,
+      "step": 4161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.315344554264643e-07,
+      "loss": 0.0063,
+      "step": 4162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.296078319341444e-07,
+      "loss": 0.0101,
+      "step": 4163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.276832410403051e-07,
+      "loss": 0.0116,
+      "step": 4164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.25760683682968e-07,
+      "loss": 0.011,
+      "step": 4165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.238401607991647e-07,
+      "loss": 0.0084,
+      "step": 4166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.21921673324933e-07,
+      "loss": 0.0074,
+      "step": 4167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.200052221953231e-07,
+      "loss": 0.0075,
+      "step": 4168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.180908083443884e-07,
+      "loss": 0.009,
+      "step": 4169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.161784327051919e-07,
+      "loss": 0.0087,
+      "step": 4170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.142680962098016e-07,
+      "loss": 0.0171,
+      "step": 4171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.123597997892918e-07,
+      "loss": 0.0072,
+      "step": 4172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.104535443737438e-07,
+      "loss": 0.0067,
+      "step": 4173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.085493308922432e-07,
+      "loss": 0.0109,
+      "step": 4174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.066471602728804e-07,
+      "loss": 0.0059,
+      "step": 4175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.047470334427504e-07,
+      "loss": 0.0089,
+      "step": 4176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.028489513279503e-07,
+      "loss": 0.008,
+      "step": 4177
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.009529148535855e-07,
+      "loss": 0.0063,
+      "step": 4178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.990589249437591e-07,
+      "loss": 0.0066,
+      "step": 4179
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.971669825215789e-07,
+      "loss": 0.0069,
+      "step": 4180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.952770885091548e-07,
+      "loss": 0.0125,
+      "step": 4181
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.933892438275987e-07,
+      "loss": 0.009,
+      "step": 4182
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.91503449397022e-07,
+      "loss": 0.0165,
+      "step": 4183
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.89619706136539e-07,
+      "loss": 0.0074,
+      "step": 4184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.877380149642628e-07,
+      "loss": 0.0084,
+      "step": 4185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.858583767973071e-07,
+      "loss": 0.008,
+      "step": 4186
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.839807925517834e-07,
+      "loss": 0.0062,
+      "step": 4187
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.821052631428061e-07,
+      "loss": 0.0062,
+      "step": 4188
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.802317894844835e-07,
+      "loss": 0.0059,
+      "step": 4189
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.783603724899258e-07,
+      "loss": 0.0059,
+      "step": 4190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.76491013071235e-07,
+      "loss": 0.0089,
+      "step": 4191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.746237121395184e-07,
+      "loss": 0.0124,
+      "step": 4192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.727584706048735e-07,
+      "loss": 0.0115,
+      "step": 4193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.708952893763972e-07,
+      "loss": 0.0065,
+      "step": 4194
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.690341693621805e-07,
+      "loss": 0.0089,
+      "step": 4195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.671751114693104e-07,
+      "loss": 0.0079,
+      "step": 4196
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.653181166038715e-07,
+      "loss": 0.0072,
+      "step": 4197
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.63463185670939e-07,
+      "loss": 0.0063,
+      "step": 4198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.61610319574585e-07,
+      "loss": 0.0068,
+      "step": 4199
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.597595192178702e-07,
+      "loss": 0.0061,
+      "step": 4200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.579107855028562e-07,
+      "loss": 0.0091,
+      "step": 4201
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.560641193305912e-07,
+      "loss": 0.0194,
+      "step": 4202
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.542195216011188e-07,
+      "loss": 0.01,
+      "step": 4203
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.523769932134739e-07,
+      "loss": 0.0058,
+      "step": 4204
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.505365350656813e-07,
+      "loss": 0.0064,
+      "step": 4205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.486981480547567e-07,
+      "loss": 0.0078,
+      "step": 4206
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.468618330767114e-07,
+      "loss": 0.0087,
+      "step": 4207
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.450275910265415e-07,
+      "loss": 0.0072,
+      "step": 4208
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.43195422798233e-07,
+      "loss": 0.0059,
+      "step": 4209
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.413653292847617e-07,
+      "loss": 0.0086,
+      "step": 4210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.395373113780962e-07,
+      "loss": 0.0064,
+      "step": 4211
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.377113699691879e-07,
+      "loss": 0.0081,
+      "step": 4212
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.358875059479792e-07,
+      "loss": 0.0059,
+      "step": 4213
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.34065720203399e-07,
+      "loss": 0.0069,
+      "step": 4214
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.322460136233622e-07,
+      "loss": 0.0073,
+      "step": 4215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.304283870947748e-07,
+      "loss": 0.0053,
+      "step": 4216
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.286128415035249e-07,
+      "loss": 0.0428,
+      "step": 4217
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.267993777344856e-07,
+      "loss": 0.0098,
+      "step": 4218
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.249879966715174e-07,
+      "loss": 0.0065,
+      "step": 4219
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.23178699197467e-07,
+      "loss": 0.0079,
+      "step": 4220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.213714861941628e-07,
+      "loss": 0.0064,
+      "step": 4221
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.195663585424195e-07,
+      "loss": 0.0125,
+      "step": 4222
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.177633171220339e-07,
+      "loss": 0.0066,
+      "step": 4223
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.159623628117856e-07,
+      "loss": 0.0076,
+      "step": 4224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.141634964894389e-07,
+      "loss": 0.0072,
+      "step": 4225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.123667190317396e-07,
+      "loss": 0.0088,
+      "step": 4226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.105720313144143e-07,
+      "loss": 0.0095,
+      "step": 4227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.087794342121724e-07,
+      "loss": 0.0065,
+      "step": 4228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.069889285987025e-07,
+      "loss": 0.0069,
+      "step": 4229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.052005153466779e-07,
+      "loss": 0.0064,
+      "step": 4230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.034141953277484e-07,
+      "loss": 0.0077,
+      "step": 4231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.01629969412545e-07,
+      "loss": 0.0065,
+      "step": 4232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.99847838470677e-07,
+      "loss": 0.0061,
+      "step": 4233
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.980678033707333e-07,
+      "loss": 0.0095,
+      "step": 4234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.962898649802824e-07,
+      "loss": 0.0084,
+      "step": 4235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.945140241658688e-07,
+      "loss": 0.0065,
+      "step": 4236
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.927402817930168e-07,
+      "loss": 0.0148,
+      "step": 4237
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.909686387262255e-07,
+      "loss": 0.0072,
+      "step": 4238
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.891990958289724e-07,
+      "loss": 0.0087,
+      "step": 4239
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.874316539637127e-07,
+      "loss": 0.0186,
+      "step": 4240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.856663139918751e-07,
+      "loss": 0.0114,
+      "step": 4241
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.839030767738653e-07,
+      "loss": 0.0075,
+      "step": 4242
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.821419431690629e-07,
+      "loss": 0.006,
+      "step": 4243
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.803829140358237e-07,
+      "loss": 0.0064,
+      "step": 4244
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.786259902314768e-07,
+      "loss": 0.0153,
+      "step": 4245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.768711726123261e-07,
+      "loss": 0.0087,
+      "step": 4246
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.751184620336471e-07,
+      "loss": 0.0117,
+      "step": 4247
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.733678593496901e-07,
+      "loss": 0.0054,
+      "step": 4248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.716193654136788e-07,
+      "loss": 0.0062,
+      "step": 4249
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.698729810778065e-07,
+      "loss": 0.0071,
+      "step": 4250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.681287071932408e-07,
+      "loss": 0.0078,
+      "step": 4251
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.663865446101192e-07,
+      "loss": 0.0092,
+      "step": 4252
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.646464941775499e-07,
+      "loss": 0.0085,
+      "step": 4253
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.629085567436133e-07,
+      "loss": 0.0071,
+      "step": 4254
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.611727331553585e-07,
+      "loss": 0.0076,
+      "step": 4255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.594390242588044e-07,
+      "loss": 0.0082,
+      "step": 4256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.577074308989406e-07,
+      "loss": 0.0069,
+      "step": 4257
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.559779539197231e-07,
+      "loss": 0.0068,
+      "step": 4258
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.542505941640803e-07,
+      "loss": 0.0062,
+      "step": 4259
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.52525352473905e-07,
+      "loss": 0.0074,
+      "step": 4260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.508022296900601e-07,
+      "loss": 0.006,
+      "step": 4261
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.490812266523716e-07,
+      "loss": 0.0065,
+      "step": 4262
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.47362344199639e-07,
+      "loss": 0.0073,
+      "step": 4263
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.456455831696234e-07,
+      "loss": 0.0059,
+      "step": 4264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.439309443990532e-07,
+      "loss": 0.0087,
+      "step": 4265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.422184287236227e-07,
+      "loss": 0.0075,
+      "step": 4266
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.405080369779898e-07,
+      "loss": 0.0057,
+      "step": 4267
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.387997699957815e-07,
+      "loss": 0.0067,
+      "step": 4268
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.370936286095842e-07,
+      "loss": 0.0088,
+      "step": 4269
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.353896136509524e-07,
+      "loss": 0.0071,
+      "step": 4270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.336877259504004e-07,
+      "loss": 0.0085,
+      "step": 4271
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.319879663374068e-07,
+      "loss": 0.0162,
+      "step": 4272
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.302903356404161e-07,
+      "loss": 0.0158,
+      "step": 4273
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.28594834686832e-07,
+      "loss": 0.0077,
+      "step": 4274
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.269014643030214e-07,
+      "loss": 0.0088,
+      "step": 4275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.252102253143122e-07,
+      "loss": 0.006,
+      "step": 4276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.235211185449919e-07,
+      "loss": 0.0085,
+      "step": 4277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.218341448183141e-07,
+      "loss": 0.0081,
+      "step": 4278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.201493049564883e-07,
+      "loss": 0.0086,
+      "step": 4279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.184665997806832e-07,
+      "loss": 0.0075,
+      "step": 4280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.167860301110284e-07,
+      "loss": 0.0116,
+      "step": 4281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.151075967666165e-07,
+      "loss": 0.0062,
+      "step": 4282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.134313005654929e-07,
+      "loss": 0.0071,
+      "step": 4283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.117571423246655e-07,
+      "loss": 0.0091,
+      "step": 4284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.100851228600974e-07,
+      "loss": 0.0114,
+      "step": 4285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.084152429867113e-07,
+      "loss": 0.0115,
+      "step": 4286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.067475035183862e-07,
+      "loss": 0.006,
+      "step": 4287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.050819052679585e-07,
+      "loss": 0.0065,
+      "step": 4288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.034184490472195e-07,
+      "loss": 0.007,
+      "step": 4289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.017571356669183e-07,
+      "loss": 0.0071,
+      "step": 4290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.000979659367579e-07,
+      "loss": 0.0078,
+      "step": 4291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.98440940665399e-07,
+      "loss": 0.0065,
+      "step": 4292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.967860606604553e-07,
+      "loss": 0.014,
+      "step": 4293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.951333267284942e-07,
+      "loss": 0.0089,
+      "step": 4294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.934827396750392e-07,
+      "loss": 0.0071,
+      "step": 4295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.918343003045656e-07,
+      "loss": 0.0084,
+      "step": 4296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.901880094205037e-07,
+      "loss": 0.0072,
+      "step": 4297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.885438678252342e-07,
+      "loss": 0.0066,
+      "step": 4298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.869018763200929e-07,
+      "loss": 0.0061,
+      "step": 4299
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.852620357053651e-07,
+      "loss": 0.0079,
+      "step": 4300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.836243467802915e-07,
+      "loss": 0.01,
+      "step": 4301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.819888103430598e-07,
+      "loss": 0.0191,
+      "step": 4302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.803554271908124e-07,
+      "loss": 0.006,
+      "step": 4303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.787241981196384e-07,
+      "loss": 0.0069,
+      "step": 4304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.770951239245803e-07,
+      "loss": 0.0093,
+      "step": 4305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.754682053996291e-07,
+      "loss": 0.0064,
+      "step": 4306
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.738434433377244e-07,
+      "loss": 0.0053,
+      "step": 4307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.722208385307559e-07,
+      "loss": 0.01,
+      "step": 4308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.706003917695619e-07,
+      "loss": 0.0051,
+      "step": 4309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.689821038439264e-07,
+      "loss": 0.0086,
+      "step": 4310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.673659755425859e-07,
+      "loss": 0.009,
+      "step": 4311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.657520076532208e-07,
+      "loss": 0.0073,
+      "step": 4312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.641402009624591e-07,
+      "loss": 0.0071,
+      "step": 4313
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.625305562558764e-07,
+      "loss": 0.0102,
+      "step": 4314
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.609230743179939e-07,
+      "loss": 0.0074,
+      "step": 4315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.593177559322776e-07,
+      "loss": 0.1036,
+      "step": 4316
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.577146018811419e-07,
+      "loss": 0.0739,
+      "step": 4317
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.561136129459432e-07,
+      "loss": 0.0057,
+      "step": 4318
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.545147899069836e-07,
+      "loss": 0.0095,
+      "step": 4319
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.529181335435124e-07,
+      "loss": 0.0087,
+      "step": 4320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.51323644633719e-07,
+      "loss": 0.02,
+      "step": 4321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.497313239547374e-07,
+      "loss": 0.079,
+      "step": 4322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.48141172282648e-07,
+      "loss": 0.0068,
+      "step": 4323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.46553190392467e-07,
+      "loss": 0.0077,
+      "step": 4324
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.449673790581611e-07,
+      "loss": 0.0102,
+      "step": 4325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.433837390526341e-07,
+      "loss": 0.0083,
+      "step": 4326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.418022711477333e-07,
+      "loss": 0.0099,
+      "step": 4327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.402229761142464e-07,
+      "loss": 0.0091,
+      "step": 4328
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.386458547219026e-07,
+      "loss": 0.0065,
+      "step": 4329
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.370709077393721e-07,
+      "loss": 0.0075,
+      "step": 4330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.354981359342659e-07,
+      "loss": 0.0085,
+      "step": 4331
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.339275400731331e-07,
+      "loss": 0.0078,
+      "step": 4332
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.323591209214612e-07,
+      "loss": 0.015,
+      "step": 4333
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.307928792436812e-07,
+      "loss": 0.0136,
+      "step": 4334
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.292288158031595e-07,
+      "loss": 0.0066,
+      "step": 4335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.276669313622013e-07,
+      "loss": 0.0097,
+      "step": 4336
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.2610722668205e-07,
+      "loss": 0.0091,
+      "step": 4337
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.245497025228874e-07,
+      "loss": 0.009,
+      "step": 4338
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.229943596438297e-07,
+      "loss": 0.0112,
+      "step": 4339
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.214411988029355e-07,
+      "loss": 0.0069,
+      "step": 4340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.198902207571955e-07,
+      "loss": 0.0059,
+      "step": 4341
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.183414262625364e-07,
+      "loss": 0.0092,
+      "step": 4342
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.167948160738206e-07,
+      "loss": 0.006,
+      "step": 4343
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.152503909448503e-07,
+      "loss": 0.0086,
+      "step": 4344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.137081516283582e-07,
+      "loss": 0.0202,
+      "step": 4345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.121680988760125e-07,
+      "loss": 0.0072,
+      "step": 4346
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.106302334384172e-07,
+      "loss": 0.0087,
+      "step": 4347
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.090945560651073e-07,
+      "loss": 0.0084,
+      "step": 4348
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.075610675045567e-07,
+      "loss": 0.009,
+      "step": 4349
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.06029768504166e-07,
+      "loss": 0.0106,
+      "step": 4350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.045006598102725e-07,
+      "loss": 0.0109,
+      "step": 4351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.029737421681446e-07,
+      "loss": 0.0076,
+      "step": 4352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.014490163219854e-07,
+      "loss": 0.007,
+      "step": 4353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.99926483014927e-07,
+      "loss": 0.0076,
+      "step": 4354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.984061429890324e-07,
+      "loss": 0.0066,
+      "step": 4355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.968879969852985e-07,
+      "loss": 0.0067,
+      "step": 4356
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9537204574365e-07,
+      "loss": 0.0255,
+      "step": 4357
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.938582900029437e-07,
+      "loss": 0.0079,
+      "step": 4358
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.92346730500966e-07,
+      "loss": 0.0088,
+      "step": 4359
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.908373679744316e-07,
+      "loss": 0.0082,
+      "step": 4360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.893302031589864e-07,
+      "loss": 0.0082,
+      "step": 4361
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.878252367892033e-07,
+      "loss": 0.0062,
+      "step": 4362
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.863224695985858e-07,
+      "loss": 0.0063,
+      "step": 4363
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.848219023195644e-07,
+      "loss": 0.0064,
+      "step": 4364
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.833235356834959e-07,
+      "loss": 0.0055,
+      "step": 4365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.818273704206678e-07,
+      "loss": 0.0087,
+      "step": 4366
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.803334072602917e-07,
+      "loss": 0.0083,
+      "step": 4367
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.788416469305068e-07,
+      "loss": 0.009,
+      "step": 4368
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.773520901583801e-07,
+      "loss": 0.0071,
+      "step": 4369
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.758647376699033e-07,
+      "loss": 0.0057,
+      "step": 4370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.743795901899928e-07,
+      "loss": 0.0066,
+      "step": 4371
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.728966484424913e-07,
+      "loss": 0.0065,
+      "step": 4372
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.714159131501689e-07,
+      "loss": 0.0072,
+      "step": 4373
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.699373850347161e-07,
+      "loss": 0.0086,
+      "step": 4374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6846106481675035e-07,
+      "loss": 0.0073,
+      "step": 4375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6698695321581165e-07,
+      "loss": 0.0073,
+      "step": 4376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.655150509503642e-07,
+      "loss": 0.0083,
+      "step": 4377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.640453587377958e-07,
+      "loss": 0.0078,
+      "step": 4378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.625778772944156e-07,
+      "loss": 0.0156,
+      "step": 4379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6111260733545714e-07,
+      "loss": 0.0156,
+      "step": 4380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5964954957507414e-07,
+      "loss": 0.156,
+      "step": 4381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.581887047263445e-07,
+      "loss": 0.0076,
+      "step": 4382
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.567300735012653e-07,
+      "loss": 0.0083,
+      "step": 4383
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.552736566107563e-07,
+      "loss": 0.0079,
+      "step": 4384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.538194547646574e-07,
+      "loss": 0.0081,
+      "step": 4385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.523674686717283e-07,
+      "loss": 0.0102,
+      "step": 4386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5091769903964965e-07,
+      "loss": 0.0065,
+      "step": 4387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.494701465750217e-07,
+      "loss": 0.0056,
+      "step": 4388
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.480248119833641e-07,
+      "loss": 0.0048,
+      "step": 4389
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4658169596911493e-07,
+      "loss": 0.0072,
+      "step": 4390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4514079923563103e-07,
+      "loss": 0.0098,
+      "step": 4391
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4370212248518895e-07,
+      "loss": 0.0083,
+      "step": 4392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4226566641898173e-07,
+      "loss": 0.0063,
+      "step": 4393
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4083143173712207e-07,
+      "loss": 0.0092,
+      "step": 4394
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3939941913863525e-07,
+      "loss": 0.0056,
+      "step": 4395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.379696293214697e-07,
+      "loss": 0.0073,
+      "step": 4396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3654206298248625e-07,
+      "loss": 0.0071,
+      "step": 4397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3511672081746393e-07,
+      "loss": 0.0072,
+      "step": 4398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.33693603521097e-07,
+      "loss": 0.0061,
+      "step": 4399
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.322727117869951e-07,
+      "loss": 0.0096,
+      "step": 4400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.308540463076849e-07,
+      "loss": 0.0063,
+      "step": 4401
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.29437607774606e-07,
+      "loss": 0.0063,
+      "step": 4402
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.280233968781139e-07,
+      "loss": 0.0159,
+      "step": 4403
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.266114143074751e-07,
+      "loss": 0.0071,
+      "step": 4404
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2520166075087635e-07,
+      "loss": 0.0069,
+      "step": 4405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.237941368954124e-07,
+      "loss": 0.0065,
+      "step": 4406
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2238884342709397e-07,
+      "loss": 0.007,
+      "step": 4407
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2098578103084376e-07,
+      "loss": 0.0085,
+      "step": 4408
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.195849503904975e-07,
+      "loss": 0.0069,
+      "step": 4409
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1818635218880186e-07,
+      "loss": 0.0163,
+      "step": 4410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1678998710741936e-07,
+      "loss": 0.0067,
+      "step": 4411
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.153958558269189e-07,
+      "loss": 0.0071,
+      "step": 4412
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.140039590267836e-07,
+      "loss": 0.006,
+      "step": 4413
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1261429738540694e-07,
+      "loss": 0.009,
+      "step": 4414
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.112268715800943e-07,
+      "loss": 0.0206,
+      "step": 4415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0984168228705934e-07,
+      "loss": 0.0091,
+      "step": 4416
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.084587301814269e-07,
+      "loss": 0.0077,
+      "step": 4417
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0707801593723006e-07,
+      "loss": 0.0112,
+      "step": 4418
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.056995402274122e-07,
+      "loss": 0.0103,
+      "step": 4419
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.043233037238281e-07,
+      "loss": 0.0092,
+      "step": 4420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.029493070972362e-07,
+      "loss": 0.0065,
+      "step": 4421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0157755101730645e-07,
+      "loss": 0.0064,
+      "step": 4422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.002080361526156e-07,
+      "loss": 0.0059,
+      "step": 4423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9884076317064813e-07,
+      "loss": 0.0086,
+      "step": 4424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9747573273779816e-07,
+      "loss": 0.0112,
+      "step": 4425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.961129455193641e-07,
+      "loss": 0.0115,
+      "step": 4426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.947524021795518e-07,
+      "loss": 0.0061,
+      "step": 4427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9339410338147363e-07,
+      "loss": 0.0068,
+      "step": 4428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.920380497871473e-07,
+      "loss": 0.0086,
+      "step": 4429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.90684242057498e-07,
+      "loss": 0.0076,
+      "step": 4430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.89332680852354e-07,
+      "loss": 0.0081,
+      "step": 4431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.879833668304506e-07,
+      "loss": 0.0049,
+      "step": 4432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.866363006494256e-07,
+      "loss": 0.0089,
+      "step": 4433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.85291482965825e-07,
+      "loss": 0.0138,
+      "step": 4434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8394891443509554e-07,
+      "loss": 0.0077,
+      "step": 4435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8260859571158883e-07,
+      "loss": 0.0149,
+      "step": 4436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.812705274485595e-07,
+      "loss": 0.0079,
+      "step": 4437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.7993471029816653e-07,
+      "loss": 0.0077,
+      "step": 4438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.7860114491147017e-07,
+      "loss": 0.0117,
+      "step": 4439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.772698319384349e-07,
+      "loss": 0.0091,
+      "step": 4440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.759407720279257e-07,
+      "loss": 0.0061,
+      "step": 4441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.7461396582771035e-07,
+      "loss": 0.0079,
+      "step": 4442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.732894139844578e-07,
+      "loss": 0.0055,
+      "step": 4443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.7196711714373947e-07,
+      "loss": 0.0092,
+      "step": 4444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.7064707595002636e-07,
+      "loss": 0.0082,
+      "step": 4445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.693292910466906e-07,
+      "loss": 0.0088,
+      "step": 4446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.680137630760039e-07,
+      "loss": 0.0069,
+      "step": 4447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6670049267913954e-07,
+      "loss": 0.0078,
+      "step": 4448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6538948049616886e-07,
+      "loss": 0.006,
+      "step": 4449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6408072716606346e-07,
+      "loss": 0.0063,
+      "step": 4450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.627742333266937e-07,
+      "loss": 0.0071,
+      "step": 4451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.614699996148285e-07,
+      "loss": 0.007,
+      "step": 4452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.601680266661367e-07,
+      "loss": 0.0061,
+      "step": 4453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5886831511518336e-07,
+      "loss": 0.0075,
+      "step": 4454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.575708655954324e-07,
+      "loss": 0.0066,
+      "step": 4455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.562756787392452e-07,
+      "loss": 0.0078,
+      "step": 4456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5498275517787783e-07,
+      "loss": 0.0075,
+      "step": 4457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5369209554148854e-07,
+      "loss": 0.0061,
+      "step": 4458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524037004591274e-07,
+      "loss": 0.0072,
+      "step": 4459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511175705587433e-07,
+      "loss": 0.0059,
+      "step": 4460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.498337064671803e-07,
+      "loss": 0.0085,
+      "step": 4461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4855210881017675e-07,
+      "loss": 0.0057,
+      "step": 4462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.472727782123697e-07,
+      "loss": 0.007,
+      "step": 4463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.459957152972887e-07,
+      "loss": 0.0057,
+      "step": 4464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4472092068735917e-07,
+      "loss": 0.0074,
+      "step": 4465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.434483950038986e-07,
+      "loss": 0.0092,
+      "step": 4466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.421781388671225e-07,
+      "loss": 0.0083,
+      "step": 4467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.409101528961378e-07,
+      "loss": 0.0102,
+      "step": 4468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.396444377089453e-07,
+      "loss": 0.0077,
+      "step": 4469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3838099392243915e-07,
+      "loss": 0.0062,
+      "step": 4470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.371198221524069e-07,
+      "loss": 0.0063,
+      "step": 4471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.358609230135268e-07,
+      "loss": 0.008,
+      "step": 4472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3460429711937417e-07,
+      "loss": 0.008,
+      "step": 4473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3334994508241013e-07,
+      "loss": 0.0066,
+      "step": 4474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.320978675139919e-07,
+      "loss": 0.0075,
+      "step": 4475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.3084806502436617e-07,
+      "loss": 0.0085,
+      "step": 4476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2960053822267245e-07,
+      "loss": 0.0133,
+      "step": 4477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.283552877169399e-07,
+      "loss": 0.0082,
+      "step": 4478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.271123141140886e-07,
+      "loss": 0.0071,
+      "step": 4479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.258716180199278e-07,
+      "loss": 0.0063,
+      "step": 4480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.246332000391583e-07,
+      "loss": 0.0872,
+      "step": 4481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.233970607753717e-07,
+      "loss": 0.0079,
+      "step": 4482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2216320083104434e-07,
+      "loss": 0.033,
+      "step": 4483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2093162080754634e-07,
+      "loss": 0.0069,
+      "step": 4484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1970232130513365e-07,
+      "loss": 0.016,
+      "step": 4485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1847530292295313e-07,
+      "loss": 0.0112,
+      "step": 4486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.172505662590386e-07,
+      "loss": 0.0153,
+      "step": 4487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.160281119103109e-07,
+      "loss": 0.0114,
+      "step": 4488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.148079404725801e-07,
+      "loss": 0.0061,
+      "step": 4489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.135900525405428e-07,
+      "loss": 0.0061,
+      "step": 4490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.123744487077829e-07,
+      "loss": 0.0083,
+      "step": 4491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1116112956677045e-07,
+      "loss": 0.0065,
+      "step": 4492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0995009570886305e-07,
+      "loss": 0.0055,
+      "step": 4493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0874134772430344e-07,
+      "loss": 0.0138,
+      "step": 4494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0753488620222037e-07,
+      "loss": 0.0061,
+      "step": 4495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0633071173062966e-07,
+      "loss": 0.0116,
+      "step": 4496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.051288248964307e-07,
+      "loss": 0.0078,
+      "step": 4497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0392922628540875e-07,
+      "loss": 0.0057,
+      "step": 4498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.027319164822329e-07,
+      "loss": 0.0069,
+      "step": 4499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.015368960704584e-07,
+      "loss": 0.0078,
+      "step": 4500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.003441656325229e-07,
+      "loss": 0.0098,
+      "step": 4501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.99153725749749e-07,
+      "loss": 0.0083,
+      "step": 4502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9796557700234317e-07,
+      "loss": 0.0064,
+      "step": 4503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.967797199693928e-07,
+      "loss": 0.0092,
+      "step": 4504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9559615522887275e-07,
+      "loss": 0.0067,
+      "step": 4505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9441488335763656e-07,
+      "loss": 0.0074,
+      "step": 4506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9323590493142206e-07,
+      "loss": 0.0136,
+      "step": 4507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.920592205248496e-07,
+      "loss": 0.0077,
+      "step": 4508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.908848307114198e-07,
+      "loss": 0.0072,
+      "step": 4509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8971273606351656e-07,
+      "loss": 0.0068,
+      "step": 4510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8854293715240455e-07,
+      "loss": 0.0075,
+      "step": 4511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8737543454822993e-07,
+      "loss": 0.0065,
+      "step": 4512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.862102288200186e-07,
+      "loss": 0.0063,
+      "step": 4513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.850473205356774e-07,
+      "loss": 0.0088,
+      "step": 4514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.838867102619952e-07,
+      "loss": 0.0088,
+      "step": 4515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8272839856463783e-07,
+      "loss": 0.0079,
+      "step": 4516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.815723860081537e-07,
+      "loss": 0.0081,
+      "step": 4517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.804186731559677e-07,
+      "loss": 0.0228,
+      "step": 4518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.792672605703867e-07,
+      "loss": 0.0071,
+      "step": 4519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7811814881259503e-07,
+      "loss": 0.0046,
+      "step": 4520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7697133844265535e-07,
+      "loss": 0.008,
+      "step": 4521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.758268300195094e-07,
+      "loss": 0.0058,
+      "step": 4522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.746846241009765e-07,
+      "loss": 0.0082,
+      "step": 4523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.735447212437531e-07,
+      "loss": 0.0123,
+      "step": 4524
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.724071220034158e-07,
+      "loss": 0.0064,
+      "step": 4525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.712718269344161e-07,
+      "loss": 0.0058,
+      "step": 4526
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.701388365900831e-07,
+      "loss": 0.0062,
+      "step": 4527
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.690081515226206e-07,
+      "loss": 0.0171,
+      "step": 4528
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6787977228311336e-07,
+      "loss": 0.0059,
+      "step": 4529
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6675369942151864e-07,
+      "loss": 0.014,
+      "step": 4530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.656299334866702e-07,
+      "loss": 0.0055,
+      "step": 4531
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6450847502627883e-07,
+      "loss": 0.0048,
+      "step": 4532
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6338932458692847e-07,
+      "loss": 0.0114,
+      "step": 4533
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.622724827140816e-07,
+      "loss": 0.0051,
+      "step": 4534
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.611579499520722e-07,
+      "loss": 0.0054,
+      "step": 4535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.600457268441092e-07,
+      "loss": 0.0056,
+      "step": 4536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.589358139322767e-07,
+      "loss": 0.0069,
+      "step": 4537
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.578282117575343e-07,
+      "loss": 0.0073,
+      "step": 4538
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5672292085971276e-07,
+      "loss": 0.0082,
+      "step": 4539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.556199417775174e-07,
+      "loss": 0.0066,
+      "step": 4540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5451927504852757e-07,
+      "loss": 0.0091,
+      "step": 4541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.534209212091937e-07,
+      "loss": 0.0065,
+      "step": 4542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.523248807948403e-07,
+      "loss": 0.007,
+      "step": 4543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5123115433966615e-07,
+      "loss": 0.0072,
+      "step": 4544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5013974237673824e-07,
+      "loss": 0.0066,
+      "step": 4545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4905064543799706e-07,
+      "loss": 0.0072,
+      "step": 4546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.479638640542564e-07,
+      "loss": 0.0066,
+      "step": 4547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4687939875519984e-07,
+      "loss": 0.0068,
+      "step": 4548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.457972500693834e-07,
+      "loss": 0.0075,
+      "step": 4549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.447174185242324e-07,
+      "loss": 0.0108,
+      "step": 4550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4363990464604357e-07,
+      "loss": 0.0072,
+      "step": 4551
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4256470895998363e-07,
+      "loss": 0.006,
+      "step": 4552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.414918319900922e-07,
+      "loss": 0.0068,
+      "step": 4553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.404212742592743e-07,
+      "loss": 0.0059,
+      "step": 4554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3935303628930705e-07,
+      "loss": 0.0071,
+      "step": 4555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3828711860083676e-07,
+      "loss": 0.0084,
+      "step": 4556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3722352171337836e-07,
+      "loss": 0.0142,
+      "step": 4557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.361622461453178e-07,
+      "loss": 0.0071,
+      "step": 4558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.351032924139063e-07,
+      "loss": 0.0097,
+      "step": 4559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3404666103526542e-07,
+      "loss": 0.006,
+      "step": 4560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3299235252438434e-07,
+      "loss": 0.005,
+      "step": 4561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.319403673951204e-07,
+      "loss": 0.0048,
+      "step": 4562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3089070616019838e-07,
+      "loss": 0.005,
+      "step": 4563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2984336933121076e-07,
+      "loss": 0.0084,
+      "step": 4564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.287983574186159e-07,
+      "loss": 0.0074,
+      "step": 4565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2775567093174022e-07,
+      "loss": 0.0054,
+      "step": 4566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2671531037877724e-07,
+      "loss": 0.0057,
+      "step": 4567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2567727626678527e-07,
+      "loss": 0.0059,
+      "step": 4568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2464156910168954e-07,
+      "loss": 0.0066,
+      "step": 4569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2360818938828189e-07,
+      "loss": 0.0073,
+      "step": 4570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2257713763021826e-07,
+      "loss": 0.0055,
+      "step": 4571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2154841433002062e-07,
+      "loss": 0.006,
+      "step": 4572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2052201998907673e-07,
+      "loss": 0.0082,
+      "step": 4573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1949795510763872e-07,
+      "loss": 0.0069,
+      "step": 4574
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1847622018482283e-07,
+      "loss": 0.0063,
+      "step": 4575
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.174568157186102e-07,
+      "loss": 0.0066,
+      "step": 4576
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1643974220584729e-07,
+      "loss": 0.0074,
+      "step": 4577
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.154250001422431e-07,
+      "loss": 0.0075,
+      "step": 4578
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1441259002236924e-07,
+      "loss": 0.0072,
+      "step": 4579
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.134025123396638e-07,
+      "loss": 0.0069,
+      "step": 4580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.123947675864252e-07,
+      "loss": 0.0066,
+      "step": 4581
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1138935625381663e-07,
+      "loss": 0.0064,
+      "step": 4582
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.103862788318628e-07,
+      "loss": 0.0105,
+      "step": 4583
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0938553580945208e-07,
+      "loss": 0.012,
+      "step": 4584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.083871276743338e-07,
+      "loss": 0.0057,
+      "step": 4585
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0739105491312028e-07,
+      "loss": 0.008,
+      "step": 4586
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0639731801128603e-07,
+      "loss": 0.0056,
+      "step": 4587
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.054059174531653e-07,
+      "loss": 0.0058,
+      "step": 4588
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0441685372195487e-07,
+      "loss": 0.0095,
+      "step": 4589
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0343012729971244e-07,
+      "loss": 0.007,
+      "step": 4590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0244573866735673e-07,
+      "loss": 0.0065,
+      "step": 4591
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0146368830466668e-07,
+      "loss": 0.007,
+      "step": 4592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0048397669028164e-07,
+      "loss": 0.0071,
+      "step": 4593
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.995066043017013e-07,
+      "loss": 0.006,
+      "step": 4594
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9853157161528468e-07,
+      "loss": 0.0082,
+      "step": 4595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9755887910625103e-07,
+      "loss": 0.0085,
+      "step": 4596
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9658852724868005e-07,
+      "loss": 0.0067,
+      "step": 4597
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9562051651550784e-07,
+      "loss": 0.0065,
+      "step": 4598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9465484737853092e-07,
+      "loss": 0.006,
+      "step": 4599
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9369152030840553e-07,
+      "loss": 0.0102,
+      "step": 4600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.927305357746462e-07,
+      "loss": 0.0048,
+      "step": 4601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.917718942456237e-07,
+      "loss": 0.0079,
+      "step": 4602
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9081559618856938e-07,
+      "loss": 0.0094,
+      "step": 4603
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8986164206957037e-07,
+      "loss": 0.009,
+      "step": 4604
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8891003235357307e-07,
+      "loss": 0.0078,
+      "step": 4605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8796076750438096e-07,
+      "loss": 0.0115,
+      "step": 4606
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8701384798465284e-07,
+      "loss": 0.0073,
+      "step": 4607
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8606927425590616e-07,
+      "loss": 0.0071,
+      "step": 4608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8512704677851489e-07,
+      "loss": 0.0055,
+      "step": 4609
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.841871660117095e-07,
+      "loss": 0.0074,
+      "step": 4610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.832496324135763e-07,
+      "loss": 0.0091,
+      "step": 4611
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8231444644105755e-07,
+      "loss": 0.01,
+      "step": 4612
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8138160854995145e-07,
+      "loss": 0.007,
+      "step": 4613
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.804511191949121e-07,
+      "loss": 0.0069,
+      "step": 4614
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7952297882945e-07,
+      "loss": 0.0084,
+      "step": 4615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.785971879059273e-07,
+      "loss": 0.0197,
+      "step": 4616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7767374687556405e-07,
+      "loss": 0.0189,
+      "step": 4617
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7675265618843361e-07,
+      "loss": 0.0554,
+      "step": 4618
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.758339162934658e-07,
+      "loss": 0.0113,
+      "step": 4619
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7491752763844294e-07,
+      "loss": 0.0103,
+      "step": 4620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.740034906700011e-07,
+      "loss": 0.0087,
+      "step": 4621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7309180583363062e-07,
+      "loss": 0.0089,
+      "step": 4622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7218247357367656e-07,
+      "loss": 0.0068,
+      "step": 4623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7127549433333557e-07,
+      "loss": 0.0069,
+      "step": 4624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7037086855465902e-07,
+      "loss": 0.0118,
+      "step": 4625
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6946859667854977e-07,
+      "loss": 0.0103,
+      "step": 4626
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6856867914476492e-07,
+      "loss": 0.0073,
+      "step": 4627
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6767111639191202e-07,
+      "loss": 0.0065,
+      "step": 4628
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6677590885745388e-07,
+      "loss": 0.0078,
+      "step": 4629
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6588305697770313e-07,
+      "loss": 0.0145,
+      "step": 4630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6499256118782503e-07,
+      "loss": 0.0075,
+      "step": 4631
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6410442192183574e-07,
+      "loss": 0.0054,
+      "step": 4632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6321863961260452e-07,
+      "loss": 0.006,
+      "step": 4633
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6233521469185054e-07,
+      "loss": 0.0108,
+      "step": 4634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6145414759014433e-07,
+      "loss": 0.0135,
+      "step": 4635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6057543873690685e-07,
+      "loss": 0.0108,
+      "step": 4636
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.596990885604105e-07,
+      "loss": 0.0095,
+      "step": 4637
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5882509748777809e-07,
+      "loss": 0.0052,
+      "step": 4638
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5795346594498162e-07,
+      "loss": 0.007,
+      "step": 4639
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5708419435684463e-07,
+      "loss": 0.0068,
+      "step": 4640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5621728314703822e-07,
+      "loss": 0.011,
+      "step": 4641
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.553527327380855e-07,
+      "loss": 0.0214,
+      "step": 4642
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5449054355135718e-07,
+      "loss": 0.0145,
+      "step": 4643
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5363071600707435e-07,
+      "loss": 0.0116,
+      "step": 4644
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5277325052430569e-07,
+      "loss": 0.008,
+      "step": 4645
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5191814752097024e-07,
+      "loss": 0.0062,
+      "step": 4646
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5106540741383402e-07,
+      "loss": 0.006,
+      "step": 4647
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.502150306185135e-07,
+      "loss": 0.0093,
+      "step": 4648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4936701754947104e-07,
+      "loss": 0.0082,
+      "step": 4649
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4852136862001766e-07,
+      "loss": 0.0062,
+      "step": 4650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4767808424231312e-07,
+      "loss": 0.0063,
+      "step": 4651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4683716482736364e-07,
+      "loss": 0.0073,
+      "step": 4652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.459986107850231e-07,
+      "loss": 0.0093,
+      "step": 4653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4516242252399227e-07,
+      "loss": 0.008,
+      "step": 4654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4432860045182019e-07,
+      "loss": 0.0055,
+      "step": 4655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4349714497490009e-07,
+      "loss": 0.0085,
+      "step": 4656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4266805649847392e-07,
+      "loss": 0.0077,
+      "step": 4657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4184133542663014e-07,
+      "loss": 0.0069,
+      "step": 4658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4101698216230254e-07,
+      "loss": 0.0056,
+      "step": 4659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4019499710726913e-07,
+      "loss": 0.0176,
+      "step": 4660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3937538066215672e-07,
+      "loss": 0.0073,
+      "step": 4661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.385581332264363e-07,
+      "loss": 0.0079,
+      "step": 4662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3774325519842423e-07,
+      "loss": 0.0075,
+      "step": 4663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3693074697528231e-07,
+      "loss": 0.0073,
+      "step": 4664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3612060895301759e-07,
+      "loss": 0.0063,
+      "step": 4665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3531284152647983e-07,
+      "loss": 0.0155,
+      "step": 4666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3450744508936687e-07,
+      "loss": 0.0055,
+      "step": 4667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3370442003421913e-07,
+      "loss": 0.0087,
+      "step": 4668
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3290376675242022e-07,
+      "loss": 0.0068,
+      "step": 4669
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3210548563419857e-07,
+      "loss": 0.0074,
+      "step": 4670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.313095770686279e-07,
+      "loss": 0.0096,
+      "step": 4671
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3051604144362407e-07,
+      "loss": 0.0058,
+      "step": 4672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.29724879145946e-07,
+      "loss": 0.0059,
+      "step": 4673
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.289360905611975e-07,
+      "loss": 0.0058,
+      "step": 4674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2814967607382433e-07,
+      "loss": 0.0065,
+      "step": 4675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2736563606711384e-07,
+      "loss": 0.0078,
+      "step": 4676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2658397092320028e-07,
+      "loss": 0.0055,
+      "step": 4677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.258046810230562e-07,
+      "loss": 0.0094,
+      "step": 4678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2502776674649776e-07,
+      "loss": 0.018,
+      "step": 4679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2425322847218368e-07,
+      "loss": 0.0068,
+      "step": 4680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2348106657761537e-07,
+      "loss": 0.0052,
+      "step": 4681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2271128143913458e-07,
+      "loss": 0.0086,
+      "step": 4682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2194387343192504e-07,
+      "loss": 0.0053,
+      "step": 4683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.211788429300126e-07,
+      "loss": 0.0055,
+      "step": 4684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2041619030626283e-07,
+      "loss": 0.0609,
+      "step": 4685
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1965591593238513e-07,
+      "loss": 0.0075,
+      "step": 4686
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1889802017892638e-07,
+      "loss": 0.0083,
+      "step": 4687
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1814250341527611e-07,
+      "loss": 0.0057,
+      "step": 4688
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1738936600966366e-07,
+      "loss": 0.0065,
+      "step": 4689
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.166386083291604e-07,
+      "loss": 0.0059,
+      "step": 4690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1589023073967586e-07,
+      "loss": 0.0058,
+      "step": 4691
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1514423360595939e-07,
+      "loss": 0.0053,
+      "step": 4692
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1440061729160235e-07,
+      "loss": 0.0064,
+      "step": 4693
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.136593821590326e-07,
+      "loss": 0.0076,
+      "step": 4694
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1292052856952063e-07,
+      "loss": 0.0092,
+      "step": 4695
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1218405688317447e-07,
+      "loss": 0.0057,
+      "step": 4696
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1144996745894033e-07,
+      "loss": 0.0064,
+      "step": 4697
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.107182606546059e-07,
+      "loss": 0.009,
+      "step": 4698
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0998893682679479e-07,
+      "loss": 0.0063,
+      "step": 4699
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0926199633097156e-07,
+      "loss": 0.0075,
+      "step": 4700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0853743952143836e-07,
+      "loss": 0.0076,
+      "step": 4701
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0781526675133492e-07,
+      "loss": 0.0099,
+      "step": 4702
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0709547837263967e-07,
+      "loss": 0.0068,
+      "step": 4703
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0637807473616812e-07,
+      "loss": 0.0072,
+      "step": 4704
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0566305619157502e-07,
+      "loss": 0.0138,
+      "step": 4705
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0495042308735104e-07,
+      "loss": 0.0078,
+      "step": 4706
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0424017577082556e-07,
+      "loss": 0.0071,
+      "step": 4707
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0353231458816338e-07,
+      "loss": 0.0162,
+      "step": 4708
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0282683988436792e-07,
+      "loss": 0.0069,
+      "step": 4709
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0212375200327973e-07,
+      "loss": 0.0064,
+      "step": 4710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0142305128757468e-07,
+      "loss": 0.0048,
+      "step": 4711
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.007247380787657e-07,
+      "loss": 0.0065,
+      "step": 4712
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0002881271720222e-07,
+      "loss": 0.0075,
+      "step": 4713
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.933527554207012e-08,
+      "loss": 0.0071,
+      "step": 4714
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.864412689139124e-08,
+      "loss": 0.0081,
+      "step": 4715
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.795536710202169e-08,
+      "loss": 0.0083,
+      "step": 4716
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.726899650965626e-08,
+      "loss": 0.0081,
+      "step": 4717
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.658501544882182e-08,
+      "loss": 0.0066,
+      "step": 4718
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.590342425288446e-08,
+      "loss": 0.007,
+      "step": 4719
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.522422325404234e-08,
+      "loss": 0.0074,
+      "step": 4720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.454741278333013e-08,
+      "loss": 0.0069,
+      "step": 4721
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.387299317061615e-08,
+      "loss": 0.0069,
+      "step": 4722
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.320096474460527e-08,
+      "loss": 0.0062,
+      "step": 4723
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.253132783283548e-08,
+      "loss": 0.0062,
+      "step": 4724
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.186408276168012e-08,
+      "loss": 0.0089,
+      "step": 4725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.119922985634633e-08,
+      "loss": 0.0082,
+      "step": 4726
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.053676944087542e-08,
+      "loss": 0.0086,
+      "step": 4727
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.987670183814134e-08,
+      "loss": 0.0071,
+      "step": 4728
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.921902736985399e-08,
+      "loss": 0.0075,
+      "step": 4729
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.856374635655696e-08,
+      "loss": 0.0087,
+      "step": 4730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.791085911762476e-08,
+      "loss": 0.0067,
+      "step": 4731
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.726036597126619e-08,
+      "loss": 0.0062,
+      "step": 4732
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.661226723452542e-08,
+      "loss": 0.0119,
+      "step": 4733
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.596656322327645e-08,
+      "loss": 0.0074,
+      "step": 4734
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.53232542522292e-08,
+      "loss": 0.0068,
+      "step": 4735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.468234063492287e-08,
+      "loss": 0.0084,
+      "step": 4736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.404382268373145e-08,
+      "loss": 0.0114,
+      "step": 4737
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.340770070986215e-08,
+      "loss": 0.0083,
+      "step": 4738
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.277397502335194e-08,
+      "loss": 0.0047,
+      "step": 4739
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.214264593307097e-08,
+      "loss": 0.0103,
+      "step": 4740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.151371374672146e-08,
+      "loss": 0.0073,
+      "step": 4741
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.088717877083706e-08,
+      "loss": 0.0078,
+      "step": 4742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.02630413107841e-08,
+      "loss": 0.0075,
+      "step": 4743
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.964130167075923e-08,
+      "loss": 0.0057,
+      "step": 4744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.90219601537906e-08,
+      "loss": 0.0074,
+      "step": 4745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.840501706173786e-08,
+      "loss": 0.0345,
+      "step": 4746
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.779047269529105e-08,
+      "loss": 0.0071,
+      "step": 4747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.717832735397335e-08,
+      "loss": 0.0058,
+      "step": 4748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.656858133613498e-08,
+      "loss": 0.0098,
+      "step": 4749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.59612349389599e-08,
+      "loss": 0.007,
+      "step": 4750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.535628845846077e-08,
+      "loss": 0.0058,
+      "step": 4751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.475374218948118e-08,
+      "loss": 0.0116,
+      "step": 4752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.415359642569564e-08,
+      "loss": 0.0098,
+      "step": 4753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.355585145960743e-08,
+      "loss": 0.0064,
+      "step": 4754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.296050758254958e-08,
+      "loss": 0.0061,
+      "step": 4755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.236756508468612e-08,
+      "loss": 0.0093,
+      "step": 4756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.177702425500977e-08,
+      "loss": 0.0071,
+      "step": 4757
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.118888538134361e-08,
+      "loss": 0.007,
+      "step": 4758
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.060314875033836e-08,
+      "loss": 0.006,
+      "step": 4759
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.001981464747565e-08,
+      "loss": 0.0081,
+      "step": 4760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.943888335706472e-08,
+      "loss": 0.0109,
+      "step": 4761
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.88603551622452e-08,
+      "loss": 0.0075,
+      "step": 4762
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.828423034498488e-08,
+      "loss": 0.0061,
+      "step": 4763
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.771050918607913e-08,
+      "loss": 0.0098,
+      "step": 4764
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.713919196515317e-08,
+      "loss": 0.0116,
+      "step": 4765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.657027896065982e-08,
+      "loss": 0.0062,
+      "step": 4766
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.60037704498806e-08,
+      "loss": 0.0091,
+      "step": 4767
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.543966670892465e-08,
+      "loss": 0.008,
+      "step": 4768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.487796801272983e-08,
+      "loss": 0.0069,
+      "step": 4769
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.431867463506047e-08,
+      "loss": 0.0089,
+      "step": 4770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.376178684850965e-08,
+      "loss": 0.0072,
+      "step": 4771
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.3207304924498e-08,
+      "loss": 0.0069,
+      "step": 4772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.265522913327326e-08,
+      "loss": 0.0048,
+      "step": 4773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.210555974391075e-08,
+      "loss": 0.0066,
+      "step": 4774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.15582970243117e-08,
+      "loss": 0.0103,
+      "step": 4775
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.101344124120557e-08,
+      "loss": 0.0054,
+      "step": 4776
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.047099266014877e-08,
+      "loss": 0.0064,
+      "step": 4777
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.993095154552431e-08,
+      "loss": 0.0077,
+      "step": 4778
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.939331816054161e-08,
+      "loss": 0.0103,
+      "step": 4779
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8858092767236084e-08,
+      "loss": 0.0083,
+      "step": 4780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8325275626470166e-08,
+      "loss": 0.0107,
+      "step": 4781
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.7794866997933355e-08,
+      "loss": 0.0076,
+      "step": 4782
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.726686714013996e-08,
+      "loss": 0.0076,
+      "step": 4783
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.674127631043025e-08,
+      "loss": 0.0127,
+      "step": 4784
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.621809476497098e-08,
+      "loss": 0.0075,
+      "step": 4785
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.569732275875428e-08,
+      "loss": 0.006,
+      "step": 4786
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.517896054559879e-08,
+      "loss": 0.0064,
+      "step": 4787
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.466300837814797e-08,
+      "loss": 0.0057,
+      "step": 4788
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.414946650786957e-08,
+      "loss": 0.0098,
+      "step": 4789
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.363833518505834e-08,
+      "loss": 0.0094,
+      "step": 4790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.312961465883393e-08,
+      "loss": 0.0062,
+      "step": 4791
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.262330517713965e-08,
+      "loss": 0.0212,
+      "step": 4792
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2119406986745336e-08,
+      "loss": 0.0091,
+      "step": 4793
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.161792033324398e-08,
+      "loss": 0.0077,
+      "step": 4794
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.111884546105506e-08,
+      "loss": 0.0058,
+      "step": 4795
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.062218261342122e-08,
+      "loss": 0.0062,
+      "step": 4796
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.012793203240995e-08,
+      "loss": 0.0055,
+      "step": 4797
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9636093958913e-08,
+      "loss": 0.0066,
+      "step": 4798
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.914666863264528e-08,
+      "loss": 0.0073,
+      "step": 4799
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.865965629214819e-08,
+      "loss": 0.0071,
+      "step": 4800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8175057174785766e-08,
+      "loss": 0.0077,
+      "step": 4801
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.769287151674407e-08,
+      "loss": 0.0063,
+      "step": 4802
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7213099553035655e-08,
+      "loss": 0.0075,
+      "step": 4803
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6735741517495715e-08,
+      "loss": 0.0098,
+      "step": 4804
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.626079764278202e-08,
+      "loss": 0.0066,
+      "step": 4805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.578826816037718e-08,
+      "loss": 0.0072,
+      "step": 4806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.531815330058586e-08,
+      "loss": 0.0067,
+      "step": 4807
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.485045329253646e-08,
+      "loss": 0.0059,
+      "step": 4808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.438516836417994e-08,
+      "loss": 0.0071,
+      "step": 4809
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.392229874229159e-08,
+      "loss": 0.0071,
+      "step": 4810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.346184465246761e-08,
+      "loss": 0.0078,
+      "step": 4811
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3003806319127376e-08,
+      "loss": 0.0139,
+      "step": 4812
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2548183965513415e-08,
+      "loss": 0.0111,
+      "step": 4813
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.209497781369143e-08,
+      "loss": 0.0068,
+      "step": 4814
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.164418808454806e-08,
+      "loss": 0.0066,
+      "step": 4815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1195814997792014e-08,
+      "loss": 0.0059,
+      "step": 4816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0749858771956253e-08,
+      "loss": 0.0091,
+      "step": 4817
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.030631962439302e-08,
+      "loss": 0.0484,
+      "step": 4818
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.986519777127884e-08,
+      "loss": 0.0084,
+      "step": 4819
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9426493427611177e-08,
+      "loss": 0.0067,
+      "step": 4820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.899020680720844e-08,
+      "loss": 0.0059,
+      "step": 4821
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.855633812271165e-08,
+      "loss": 0.0062,
+      "step": 4822
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.812488758558386e-08,
+      "loss": 0.0055,
+      "step": 4823
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.769585540610799e-08,
+      "loss": 0.0095,
+      "step": 4824
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.726924179339009e-08,
+      "loss": 0.0094,
+      "step": 4825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.684504695535496e-08,
+      "loss": 0.0069,
+      "step": 4826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.642327109875166e-08,
+      "loss": 0.0142,
+      "step": 4827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.600391442914741e-08,
+      "loss": 0.006,
+      "step": 4828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.558697715093207e-08,
+      "loss": 0.0149,
+      "step": 4829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.517245946731529e-08,
+      "loss": 0.0066,
+      "step": 4830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.47603615803288e-08,
+      "loss": 0.0103,
+      "step": 4831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.435068369082306e-08,
+      "loss": 0.011,
+      "step": 4832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.394342599847111e-08,
+      "loss": 0.0074,
+      "step": 4833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3538588701765296e-08,
+      "loss": 0.0076,
+      "step": 4834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.313617199801777e-08,
+      "loss": 0.0112,
+      "step": 4835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2736176083362216e-08,
+      "loss": 0.008,
+      "step": 4836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2338601152751e-08,
+      "loss": 0.0092,
+      "step": 4837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.194344739995803e-08,
+      "loss": 0.0104,
+      "step": 4838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1550715017575895e-08,
+      "loss": 0.0057,
+      "step": 4839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1160404197018155e-08,
+      "loss": 0.0075,
+      "step": 4840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.077251512851709e-08,
+      "loss": 0.0088,
+      "step": 4841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.038704800112535e-08,
+      "loss": 0.018,
+      "step": 4842
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0004003002714886e-08,
+      "loss": 0.0062,
+      "step": 4843
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9623380319976912e-08,
+      "loss": 0.0072,
+      "step": 4844
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9245180138423033e-08,
+      "loss": 0.0385,
+      "step": 4845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8869402642382473e-08,
+      "loss": 0.0085,
+      "step": 4846
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8496048015005385e-08,
+      "loss": 0.0101,
+      "step": 4847
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8125116438260104e-08,
+      "loss": 0.0089,
+      "step": 4848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7756608092933678e-08,
+      "loss": 0.0078,
+      "step": 4849
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7390523158633552e-08,
+      "loss": 0.0078,
+      "step": 4850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7026861813783668e-08,
+      "loss": 0.0045,
+      "step": 4851
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6665624235629463e-08,
+      "loss": 0.0089,
+      "step": 4852
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6306810600233435e-08,
+      "loss": 0.0091,
+      "step": 4853
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5950421082476805e-08,
+      "loss": 0.0052,
+      "step": 4854
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5596455856058966e-08,
+      "loss": 0.0095,
+      "step": 4855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5244915093499134e-08,
+      "loss": 0.0054,
+      "step": 4856
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.489579896613359e-08,
+      "loss": 0.006,
+      "step": 4857
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4549107644117888e-08,
+      "loss": 0.0067,
+      "step": 4858
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4204841296424086e-08,
+      "loss": 0.0062,
+      "step": 4859
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.386300009084408e-08,
+      "loss": 0.0066,
+      "step": 4860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3523584193986816e-08,
+      "loss": 0.0058,
+      "step": 4861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3186593771280518e-08,
+      "loss": 0.0066,
+      "step": 4862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.285202898696881e-08,
+      "loss": 0.0078,
+      "step": 4863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.251989000411514e-08,
+      "loss": 0.0093,
+      "step": 4864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.219017698460002e-08,
+      "loss": 0.0071,
+      "step": 4865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1862890089121567e-08,
+      "loss": 0.0067,
+      "step": 4866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1538029477195522e-08,
+      "loss": 0.0056,
+      "step": 4867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1215595307154667e-08,
+      "loss": 0.0102,
+      "step": 4868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0895587736149414e-08,
+      "loss": 0.0074,
+      "step": 4869
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.057800692014833e-08,
+      "loss": 0.0074,
+      "step": 4870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.026285301393538e-08,
+      "loss": 0.0092,
+      "step": 4871
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.995012617111436e-08,
+      "loss": 0.0063,
+      "step": 4872
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.963982654410279e-08,
+      "loss": 0.0241,
+      "step": 4873
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9331954284137476e-08,
+      "loss": 0.0063,
+      "step": 4874
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9026509541272276e-08,
+      "loss": 0.0086,
+      "step": 4875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8723492464376992e-08,
+      "loss": 0.0091,
+      "step": 4876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.842290320113793e-08,
+      "loss": 0.0159,
+      "step": 4877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8124741898058462e-08,
+      "loss": 0.0083,
+      "step": 4878
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7829008700460116e-08,
+      "loss": 0.0063,
+      "step": 4879
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.753570375247815e-08,
+      "loss": 0.0103,
+      "step": 4880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7244827197067103e-08,
+      "loss": 0.0066,
+      "step": 4881
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6956379175995796e-08,
+      "loss": 0.0075,
+      "step": 4882
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6670359829850657e-08,
+      "loss": 0.0057,
+      "step": 4883
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6386769298034067e-08,
+      "loss": 0.0135,
+      "step": 4884
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.610560771876435e-08,
+      "loss": 0.0064,
+      "step": 4885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.582687522907633e-08,
+      "loss": 0.0054,
+      "step": 4886
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5550571964820793e-08,
+      "loss": 0.01,
+      "step": 4887
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5276698060665007e-08,
+      "loss": 0.0072,
+      "step": 4888
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.500525365009109e-08,
+      "loss": 0.0058,
+      "step": 4889
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4736238865398766e-08,
+      "loss": 0.0092,
+      "step": 4890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4469653837701491e-08,
+      "loss": 0.0069,
+      "step": 4891
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4205498696930332e-08,
+      "loss": 0.0068,
+      "step": 4892
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3943773571831188e-08,
+      "loss": 0.0066,
+      "step": 4893
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3684478589964801e-08,
+      "loss": 0.0061,
+      "step": 4894
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3427613877709523e-08,
+      "loss": 0.0062,
+      "step": 4895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3173179560257432e-08,
+      "loss": 0.004,
+      "step": 4896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.292117576161711e-08,
+      "loss": 0.0072,
+      "step": 4897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2671602604612531e-08,
+      "loss": 0.0095,
+      "step": 4898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2424460210881394e-08,
+      "loss": 0.0086,
+      "step": 4899
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2179748700879013e-08,
+      "loss": 0.0073,
+      "step": 4900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1937468193873869e-08,
+      "loss": 0.0052,
+      "step": 4901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1697618807951504e-08,
+      "loss": 0.0081,
+      "step": 4902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.146020066001119e-08,
+      "loss": 0.0082,
+      "step": 4903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1225213865767026e-08,
+      "loss": 0.0074,
+      "step": 4904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0992658539750179e-08,
+      "loss": 0.0058,
+      "step": 4905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.076253479530387e-08,
+      "loss": 0.0122,
+      "step": 4906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0534842744588381e-08,
+      "loss": 0.0142,
+      "step": 4907
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.030958249857772e-08,
+      "loss": 0.0067,
+      "step": 4908
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.008675416706073e-08,
+      "loss": 0.0064,
+      "step": 4909
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.866357858642206e-09,
+      "loss": 0.0071,
+      "step": 4910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.64839368074011e-09,
+      "loss": 0.0062,
+      "step": 4911
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.432861739586685e-09,
+      "loss": 0.0101,
+      "step": 4912
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.219762140231237e-09,
+      "loss": 0.0072,
+      "step": 4913
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.009094986534572e-09,
+      "loss": 0.0073,
+      "step": 4914
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.800860381173448e-09,
+      "loss": 0.0065,
+      "step": 4915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.595058425640012e-09,
+      "loss": 0.0083,
+      "step": 4916
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.391689220238474e-09,
+      "loss": 0.0103,
+      "step": 4917
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.190752864088436e-09,
+      "loss": 0.0085,
+      "step": 4918
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.992249455124889e-09,
+      "loss": 0.0186,
+      "step": 4919
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.796179090094891e-09,
+      "loss": 0.0057,
+      "step": 4920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.602541864561442e-09,
+      "loss": 0.0086,
+      "step": 4921
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.411337872900715e-09,
+      "loss": 0.0069,
+      "step": 4922
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.222567208303721e-09,
+      "loss": 0.0064,
+      "step": 4923
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.036229962774088e-09,
+      "loss": 0.0067,
+      "step": 4924
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.852326227130835e-09,
+      "loss": 0.0056,
+      "step": 4925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.670856091006151e-09,
+      "loss": 0.0072,
+      "step": 4926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.491819642846509e-09,
+      "loss": 0.0117,
+      "step": 4927
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.315216969912663e-09,
+      "loss": 0.0053,
+      "step": 4928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.141048158277429e-09,
+      "loss": 0.012,
+      "step": 4929
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.969313292830126e-09,
+      "loss": 0.0064,
+      "step": 4930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.800012457270466e-09,
+      "loss": 0.0064,
+      "step": 4931
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.633145734114665e-09,
+      "loss": 0.0054,
+      "step": 4932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.468713204692111e-09,
+      "loss": 0.0052,
+      "step": 4933
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.306714949143699e-09,
+      "loss": 0.0105,
+      "step": 4934
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.147151046426824e-09,
+      "loss": 0.0071,
+      "step": 4935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.990021574309834e-09,
+      "loss": 0.0071,
+      "step": 4936
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835326609376468e-09,
+      "loss": 0.0065,
+      "step": 4937
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.683066227023081e-09,
+      "loss": 0.0079,
+      "step": 4938
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.533240501459202e-09,
+      "loss": 0.0082,
+      "step": 4939
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.385849505708084e-09,
+      "loss": 0.0057,
+      "step": 4940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2408933116072635e-09,
+      "loss": 0.005,
+      "step": 4941
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.098371989805227e-09,
+      "loss": 0.0069,
+      "step": 4942
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9582856097658554e-09,
+      "loss": 0.0123,
+      "step": 4943
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.820634239765642e-09,
+      "loss": 0.0093,
+      "step": 4944
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.685417946894254e-09,
+      "loss": 0.0071,
+      "step": 4945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5526367970539765e-09,
+      "loss": 0.0079,
+      "step": 4946
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4222908549608193e-09,
+      "loss": 0.0082,
+      "step": 4947
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.294380184143964e-09,
+      "loss": 0.0057,
+      "step": 4948
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1689048469457638e-09,
+      "loss": 0.0082,
+      "step": 4949
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0458649045211897e-09,
+      "loss": 0.0063,
+      "step": 4950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9252604168383826e-09,
+      "loss": 0.0081,
+      "step": 4951
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8070914426786555e-09,
+      "loss": 0.0069,
+      "step": 4952
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6913580396359384e-09,
+      "loss": 0.0063,
+      "step": 4953
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5780602641167774e-09,
+      "loss": 0.0083,
+      "step": 4954
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4671981713420003e-09,
+      "loss": 0.0061,
+      "step": 4955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.358771815344496e-09,
+      "loss": 0.0065,
+      "step": 4956
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2527812489692156e-09,
+      "loss": 0.0062,
+      "step": 4957
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.149226523874837e-09,
+      "loss": 0.0078,
+      "step": 4958
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0481076905332074e-09,
+      "loss": 0.0057,
+      "step": 4959
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9494247982282386e-09,
+      "loss": 0.0091,
+      "step": 4960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8531778950564572e-09,
+      "loss": 0.0092,
+      "step": 4961
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.759367027927561e-09,
+      "loss": 0.0059,
+      "step": 4962
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6679922425638651e-09,
+      "loss": 0.0131,
+      "step": 4963
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5790535835003006e-09,
+      "loss": 0.0067,
+      "step": 4964
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4925510940844157e-09,
+      "loss": 0.009,
+      "step": 4965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4084848164763742e-09,
+      "loss": 0.0068,
+      "step": 4966
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3268547916495124e-09,
+      "loss": 0.0081,
+      "step": 4967
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.247661059389227e-09,
+      "loss": 0.0055,
+      "step": 4968
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.170903658293532e-09,
+      "loss": 0.0061,
+      "step": 4969
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.096582625772502e-09,
+      "loss": 0.007,
+      "step": 4970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0246979980499395e-09,
+      "loss": 0.0114,
+      "step": 4971
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.55249810161152e-10,
+      "loss": 0.0333,
+      "step": 4972
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.88238095955174e-10,
+      "loss": 0.0063,
+      "step": 4973
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.236628880914365e-10,
+      "loss": 0.0183,
+      "step": 4974
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.615242180436521e-10,
+      "loss": 0.0131,
+      "step": 4975
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.018221160981498e-10,
+      "loss": 0.0113,
+      "step": 4976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.445566113516544e-10,
+      "loss": 0.0116,
+      "step": 4977
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.897277317157279e-10,
+      "loss": 0.0068,
+      "step": 4978
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.373355039128836e-10,
+      "loss": 0.007,
+      "step": 4979
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.87379953478806e-10,
+      "loss": 0.0065,
+      "step": 4980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.398611047612411e-10,
+      "loss": 0.0105,
+      "step": 4981
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9477898091944135e-10,
+      "loss": 0.0073,
+      "step": 4982
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.521336039263856e-10,
+      "loss": 0.0083,
+      "step": 4983
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1192499456766947e-10,
+      "loss": 0.0102,
+      "step": 4984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.741531724392843e-10,
+      "loss": 0.0142,
+      "step": 4985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.388181559515035e-10,
+      "loss": 0.0061,
+      "step": 4986
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0591996232610656e-10,
+      "loss": 0.0053,
+      "step": 4987
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7545860759693446e-10,
+      "loss": 0.0083,
+      "step": 4988
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4743410661044454e-10,
+      "loss": 0.0102,
+      "step": 4989
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2184647302626585e-10,
+      "loss": 0.0105,
+      "step": 4990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.869571931442334e-11,
+      "loss": 0.0149,
+      "step": 4991
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.798185675866876e-11,
+      "loss": 0.0079,
+      "step": 4992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.970489545537028e-11,
+      "loss": 0.007,
+      "step": 4993
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3864844311847235e-11,
+      "loss": 0.0108,
+      "step": 4994
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0461711048035415e-11,
+      "loss": 0.0062,
+      "step": 4995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9495502197042214e-11,
+      "loss": 0.0107,
+      "step": 4996
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0966223103481278e-11,
+      "loss": 0.009,
+      "step": 4997
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.873877924582715e-12,
+      "loss": 0.0099,
+      "step": 4998
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2184696296380083e-12,
+      "loss": 0.0093,
+      "step": 4999
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.0076,
+      "step": 5000
+    },
+    {
+      "epoch": 1.0,
+      "step": 5000,
+      "total_flos": 1.65603302178816e+18,
+      "train_loss": 0.016783093049097805,
+      "train_runtime": 19874.1523,
+      "train_samples_per_second": 16.101,
+      "train_steps_per_second": 0.252
+    }
+  ],
+  "max_steps": 5000,
+  "num_train_epochs": 9223372036854775807,
+  "total_flos": 1.65603302178816e+18,
+  "trial_name": null,
+  "trial_params": null
+}