diff --git "a/checkpoint-3000/trainer_state.json" "b/checkpoint-3000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-3000/trainer_state.json"
@@ -0,0 +1,18043 @@
+{
+  "best_metric": 19.604205318491033,
+  "best_model_checkpoint": "./whisper-small-swedish-Test-3000/checkpoint-3000",
+  "epoch": 3.8809831824062098,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 4.395,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 4.0449,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 4.7341,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 3.8688,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-08,
+      "loss": 5.1651,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-08,
+      "loss": 4.2447,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 4.4,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-08,
+      "loss": 4.2329,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 4.6475,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 4.5232,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4e-07,
+      "loss": 4.3821,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6e-07,
+      "loss": 4.3285,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8e-07,
+      "loss": 4.2704,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 4.2228,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2e-07,
+      "loss": 3.8452,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 4.5873,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6e-07,
+      "loss": 4.2215,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8e-07,
+      "loss": 4.6304,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.8465,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2e-07,
+      "loss": 3.9467,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 3.4544,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6e-07,
+      "loss": 4.0918,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8e-07,
+      "loss": 4.0405,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.6531,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 3.1822,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4e-07,
+      "loss": 4.4433,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 3.9565,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 3.7222,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.7677,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.2e-07,
+      "loss": 4.1824,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.4e-07,
+      "loss": 3.2356,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6e-07,
+      "loss": 4.0897,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 3.2607,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.1454,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 3.8168,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.4e-07,
+      "loss": 3.2422,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.6e-07,
+      "loss": 3.3574,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 3.4595,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.5301,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.2e-07,
+      "loss": 2.9723,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.4e-07,
+      "loss": 3.2563,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.6e-07,
+      "loss": 3.1801,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.8e-07,
+      "loss": 3.1175,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.3198,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 2.51,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 3.0512,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.6e-07,
+      "loss": 2.6373,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.8e-07,
+      "loss": 3.1265,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.1742,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 2.6901,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 2.881,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 2.7274,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 2.7614,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.5636,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.02e-06,
+      "loss": 2.8604,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.04e-06,
+      "loss": 2.5257,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.06e-06,
+      "loss": 2.6418,
+      "step": 57
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.08e-06,
+      "loss": 3.0077,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1e-06,
+      "loss": 2.2389,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.12e-06,
+      "loss": 2.5802,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.14e-06,
+      "loss": 2.7213,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 2.7884,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 2.293,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.4187,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 2.2956,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 2.3675,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.26e-06,
+      "loss": 2.6477,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.28e-06,
+      "loss": 2.2633,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3e-06,
+      "loss": 2.437,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.32e-06,
+      "loss": 2.5359,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.34e-06,
+      "loss": 2.0851,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 2.5844,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 2.0759,
+      "step": 73
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 1.874,
+      "step": 74
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.42e-06,
+      "loss": 1.9707,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.44e-06,
+      "loss": 1.7433,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.46e-06,
+      "loss": 1.9944,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.48e-06,
+      "loss": 1.8037,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.5e-06,
+      "loss": 1.7155,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.52e-06,
+      "loss": 1.9057,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.54e-06,
+      "loss": 1.8214,
+      "step": 81
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.56e-06,
+      "loss": 1.4923,
+      "step": 82
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 1.6633,
+      "step": 83
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.5372,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 1.6179,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 1.0725,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 1.5516,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 1.3554,
+      "step": 88
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 1.3521,
+      "step": 89
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.72e-06,
+      "loss": 1.4382,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.74e-06,
+      "loss": 1.2501,
+      "step": 91
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.76e-06,
+      "loss": 1.4096,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 1.6289,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 1.4058,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 1.129,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 1.4644,
+      "step": 96
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 1.4118,
+      "step": 97
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 1.3409,
+      "step": 98
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 1.1649,
+      "step": 99
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 1.1552,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94e-06,
+      "loss": 1.0825,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 1.1697,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.98e-06,
+      "loss": 1.2379,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.0347,
+      "step": 104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.02e-06,
+      "loss": 1.1671,
+      "step": 105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.04e-06,
+      "loss": 1.1411,
+      "step": 106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.06e-06,
+      "loss": 1.103,
+      "step": 107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.08e-06,
+      "loss": 1.2268,
+      "step": 108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 1.3028,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.12e-06,
+      "loss": 1.1445,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 1.2331,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.16e-06,
+      "loss": 1.1188,
+      "step": 112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 1.3088,
+      "step": 113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.2e-06,
+      "loss": 1.2728,
+      "step": 114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 1.119,
+      "step": 115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.24e-06,
+      "loss": 1.1833,
+      "step": 116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 1.1106,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.28e-06,
+      "loss": 0.9778,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 1.0656,
+      "step": 119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 1.1527,
+      "step": 120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 1.0438,
+      "step": 121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 0.9516,
+      "step": 122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.38e-06,
+      "loss": 1.1298,
+      "step": 123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.1249,
+      "step": 124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.42e-06,
+      "loss": 1.1475,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 1.231,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.46e-06,
+      "loss": 0.9551,
+      "step": 127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 1.2245,
+      "step": 128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5e-06,
+      "loss": 0.8682,
+      "step": 129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.52e-06,
+      "loss": 0.9014,
+      "step": 130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 1.0704,
+      "step": 131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.56e-06,
+      "loss": 1.0338,
+      "step": 132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 1.0788,
+      "step": 133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.6e-06,
+      "loss": 0.9233,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 1.0947,
+      "step": 135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.64e-06,
+      "loss": 0.9273,
+      "step": 136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 1.1656,
+      "step": 137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.68e-06,
+      "loss": 0.9767,
+      "step": 138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 1.1423,
+      "step": 139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 0.8623,
+      "step": 140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 1.0794,
+      "step": 141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 0.9867,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.9671,
+      "step": 143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.0643,
+      "step": 144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.82e-06,
+      "loss": 1.0043,
+      "step": 145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.84e-06,
+      "loss": 0.9117,
+      "step": 146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.86e-06,
+      "loss": 1.0169,
+      "step": 147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.88e-06,
+      "loss": 0.9239,
+      "step": 148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9e-06,
+      "loss": 1.0685,
+      "step": 149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.92e-06,
+      "loss": 0.8899,
+      "step": 150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 0.9628,
+      "step": 151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.96e-06,
+      "loss": 1.1378,
+      "step": 152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 1.0353,
+      "step": 153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3e-06,
+      "loss": 0.8729,
+      "step": 154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 0.9006,
+      "step": 155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.04e-06,
+      "loss": 0.9142,
+      "step": 156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 0.9569,
+      "step": 157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.08e-06,
+      "loss": 0.9067,
+      "step": 158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 0.7984,
+      "step": 159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.12e-06,
+      "loss": 0.8346,
+      "step": 160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 1.0714,
+      "step": 161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 0.9894,
+      "step": 162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 1.1956,
+      "step": 163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.928,
+      "step": 164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 0.9189,
+      "step": 165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 0.9178,
+      "step": 166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 0.9468,
+      "step": 167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 0.8104,
+      "step": 168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 0.8992,
+      "step": 169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 0.8997,
+      "step": 170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 0.7763,
+      "step": 171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 1.0102,
+      "step": 172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.7885,
+      "step": 173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.8626,
+      "step": 174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 0.8934,
+      "step": 175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.44e-06,
+      "loss": 0.8363,
+      "step": 176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.46e-06,
+      "loss": 0.917,
+      "step": 177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.48e-06,
+      "loss": 0.8555,
+      "step": 178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5e-06,
+      "loss": 0.9317,
+      "step": 179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.52e-06,
+      "loss": 0.8936,
+      "step": 180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.54e-06,
+      "loss": 0.9832,
+      "step": 181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 1.0079,
+      "step": 182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8884,
+      "step": 183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.8029,
+      "step": 184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.62e-06,
+      "loss": 0.7647,
+      "step": 185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 0.8811,
+      "step": 186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.66e-06,
+      "loss": 0.8564,
+      "step": 187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 0.9085,
+      "step": 188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7e-06,
+      "loss": 0.8664,
+      "step": 189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 0.8463,
+      "step": 190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.74e-06,
+      "loss": 0.7253,
+      "step": 191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 0.8904,
+      "step": 192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8905,
+      "step": 193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.8285,
+      "step": 194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 0.8176,
+      "step": 195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 0.8995,
+      "step": 196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.86e-06,
+      "loss": 0.8233,
+      "step": 197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.88e-06,
+      "loss": 0.8045,
+      "step": 198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 0.8363,
+      "step": 199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 0.7637,
+      "step": 200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.94e-06,
+      "loss": 0.7741,
+      "step": 201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.96e-06,
+      "loss": 0.705,
+      "step": 202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.8856,
+      "step": 203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7075,
+      "step": 204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 0.7965,
+      "step": 205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.04e-06,
+      "loss": 0.6272,
+      "step": 206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 0.7303,
+      "step": 207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.08e-06,
+      "loss": 0.7233,
+      "step": 208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1e-06,
+      "loss": 0.814,
+      "step": 209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.12e-06,
+      "loss": 0.7339,
+      "step": 210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.14e-06,
+      "loss": 0.702,
+      "step": 211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.16e-06,
+      "loss": 0.7194,
+      "step": 212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.18e-06,
+      "loss": 0.6023,
+      "step": 213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.6186,
+      "step": 214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.22e-06,
+      "loss": 0.6669,
+      "step": 215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.24e-06,
+      "loss": 0.7588,
+      "step": 216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.26e-06,
+      "loss": 0.7871,
+      "step": 217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 0.7603,
+      "step": 218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3e-06,
+      "loss": 0.6377,
+      "step": 219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.32e-06,
+      "loss": 0.6332,
+      "step": 220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.34e-06,
+      "loss": 0.7565,
+      "step": 221
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 0.697,
+      "step": 222
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7049,
+      "step": 223
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4e-06,
+      "loss": 0.6418,
+      "step": 224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.42e-06,
+      "loss": 0.5892,
+      "step": 225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 0.7336,
+      "step": 226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 0.6141,
+      "step": 227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.48e-06,
+      "loss": 0.5998,
+      "step": 228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5e-06,
+      "loss": 0.6392,
+      "step": 229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 0.7224,
+      "step": 230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 0.6227,
+      "step": 231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.56e-06,
+      "loss": 0.619,
+      "step": 232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7276,
+      "step": 233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.6601,
+      "step": 234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 0.7069,
+      "step": 235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 0.5525,
+      "step": 236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.66e-06,
+      "loss": 0.5791,
+      "step": 237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 0.6904,
+      "step": 238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7e-06,
+      "loss": 0.5832,
+      "step": 239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 0.6208,
+      "step": 240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.74e-06,
+      "loss": 0.6315,
+      "step": 241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.76e-06,
+      "loss": 0.6102,
+      "step": 242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.78e-06,
+      "loss": 0.6959,
+      "step": 243
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.5723,
+      "step": 244
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 0.5515,
+      "step": 245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.84e-06,
+      "loss": 0.5445,
+      "step": 246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.86e-06,
+      "loss": 0.5422,
+      "step": 247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 0.5842,
+      "step": 248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 0.5495,
+      "step": 249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.92e-06,
+      "loss": 0.5371,
+      "step": 250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.94e-06,
+      "loss": 0.55,
+      "step": 251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 0.497,
+      "step": 252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.5995,
+      "step": 253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5e-06,
+      "loss": 0.6282,
+      "step": 254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.02e-06,
+      "loss": 0.6977,
+      "step": 255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.04e-06,
+      "loss": 0.4138,
+      "step": 256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 0.5982,
+      "step": 257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 0.5051,
+      "step": 258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.1e-06,
+      "loss": 0.5628,
+      "step": 259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.12e-06,
+      "loss": 0.4942,
+      "step": 260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 0.4225,
+      "step": 261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 0.4663,
+      "step": 262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.18e-06,
+      "loss": 0.5791,
+      "step": 263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.2e-06,
+      "loss": 0.4888,
+      "step": 264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 0.4557,
+      "step": 265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 0.4799,
+      "step": 266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 0.4994,
+      "step": 267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.28e-06,
+      "loss": 0.2801,
+      "step": 268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 0.4254,
+      "step": 269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 0.442,
+      "step": 270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 0.3498,
+      "step": 271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.36e-06,
+      "loss": 0.4445,
+      "step": 272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.2333,
+      "step": 273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 0.321,
+      "step": 274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 0.2796,
+      "step": 275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 0.2785,
+      "step": 276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 0.4092,
+      "step": 277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 0.1976,
+      "step": 278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.3942,
+      "step": 279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 0.344,
+      "step": 280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 0.2907,
+      "step": 281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 0.2686,
+      "step": 282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.302,
+      "step": 283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.3522,
+      "step": 284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 0.2832,
+      "step": 285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.64e-06,
+      "loss": 0.2382,
+      "step": 286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.66e-06,
+      "loss": 0.3785,
+      "step": 287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.68e-06,
+      "loss": 0.2599,
+      "step": 288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.7e-06,
+      "loss": 0.3906,
+      "step": 289
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.72e-06,
+      "loss": 0.4287,
+      "step": 290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.74e-06,
+      "loss": 0.4113,
+      "step": 291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.76e-06,
+      "loss": 0.2168,
+      "step": 292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.78e-06,
+      "loss": 0.2779,
+      "step": 293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8e-06,
+      "loss": 0.2524,
+      "step": 294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.82e-06,
+      "loss": 0.4853,
+      "step": 295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.84e-06,
+      "loss": 0.2412,
+      "step": 296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.86e-06,
+      "loss": 0.2211,
+      "step": 297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 0.3164,
+      "step": 298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.9e-06,
+      "loss": 0.251,
+      "step": 299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.92e-06,
+      "loss": 0.2844,
+      "step": 300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.94e-06,
+      "loss": 0.3333,
+      "step": 301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 0.3683,
+      "step": 302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.98e-06,
+      "loss": 0.5333,
+      "step": 303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6e-06,
+      "loss": 0.2359,
+      "step": 304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.02e-06,
+      "loss": 0.391,
+      "step": 305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 0.3465,
+      "step": 306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 0.4912,
+      "step": 307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.08e-06,
+      "loss": 0.3711,
+      "step": 308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.1e-06,
+      "loss": 0.3932,
+      "step": 309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 0.2139,
+      "step": 310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 0.3667,
+      "step": 311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.16e-06,
+      "loss": 0.2327,
+      "step": 312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.18e-06,
+      "loss": 0.2835,
+      "step": 313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 0.3368,
+      "step": 314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 0.3004,
+      "step": 315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.24e-06,
+      "loss": 0.3797,
+      "step": 316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.26e-06,
+      "loss": 0.3416,
+      "step": 317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 0.2959,
+      "step": 318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 0.3807,
+      "step": 319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 0.2279,
+      "step": 320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.34e-06,
+      "loss": 0.3407,
+      "step": 321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 0.3498,
+      "step": 322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.3344,
+      "step": 323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.2296,
+      "step": 324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.42e-06,
+      "loss": 0.3635,
+      "step": 325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 0.1841,
+      "step": 326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 0.3908,
+      "step": 327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 0.2328,
+      "step": 328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.3268,
+      "step": 329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 0.2612,
+      "step": 330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 0.3278,
+      "step": 331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 0.3723,
+      "step": 332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.1958,
+      "step": 333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 0.2878,
+      "step": 334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 0.1966,
+      "step": 335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 0.3217,
+      "step": 336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 0.3566,
+      "step": 337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 0.311,
+      "step": 338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 0.2637,
+      "step": 339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 0.241,
+      "step": 340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 0.2138,
+      "step": 341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 0.29,
+      "step": 342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.2436,
+      "step": 343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.2874,
+      "step": 344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 0.41,
+      "step": 345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 0.3928,
+      "step": 346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 0.2678,
+      "step": 347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.88e-06,
+      "loss": 0.4438,
+      "step": 348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.9e-06,
+      "loss": 0.2744,
+      "step": 349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.92e-06,
+      "loss": 0.3031,
+      "step": 350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 0.2483,
+      "step": 351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.96e-06,
+      "loss": 0.3699,
+      "step": 352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.98e-06,
+      "loss": 0.3141,
+      "step": 353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7e-06,
+      "loss": 0.1779,
+      "step": 354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 0.2598,
+      "step": 355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.04e-06,
+      "loss": 0.4126,
+      "step": 356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.06e-06,
+      "loss": 0.2722,
+      "step": 357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.08e-06,
+      "loss": 0.3056,
+      "step": 358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 0.3284,
+      "step": 359
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 0.3319,
+      "step": 360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.14e-06,
+      "loss": 0.3226,
+      "step": 361
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.16e-06,
+      "loss": 0.3834,
+      "step": 362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.3085,
+      "step": 363
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.363,
+      "step": 364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.22e-06,
+      "loss": 0.433,
+      "step": 365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.24e-06,
+      "loss": 0.3258,
+      "step": 366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 0.2963,
+      "step": 367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 0.3056,
+      "step": 368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.3e-06,
+      "loss": 0.312,
+      "step": 369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.32e-06,
+      "loss": 0.4487,
+      "step": 370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 0.3073,
+      "step": 371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 0.2704,
+      "step": 372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.4159,
+      "step": 373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.4e-06,
+      "loss": 0.1866,
+      "step": 374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 0.237,
+      "step": 375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 0.3067,
+      "step": 376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 0.2606,
+      "step": 377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.48e-06,
+      "loss": 0.3672,
+      "step": 378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.1962,
+      "step": 379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 0.3782,
+      "step": 380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 0.2287,
+      "step": 381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 0.3213,
+      "step": 382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.58e-06,
+      "loss": 0.4088,
+      "step": 383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.2305,
+      "step": 384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 0.3139,
+      "step": 385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 0.2927,
+      "step": 386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 0.2558,
+      "step": 387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 0.3695,
+      "step": 388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.7e-06,
+      "loss": 0.3994,
+      "step": 389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.72e-06,
+      "loss": 0.2252,
+      "step": 390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.74e-06,
+      "loss": 0.3546,
+      "step": 391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.76e-06,
+      "loss": 0.3023,
+      "step": 392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.78e-06,
+      "loss": 0.2558,
+      "step": 393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 0.4484,
+      "step": 394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 0.321,
+      "step": 395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 0.3118,
+      "step": 396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 0.4435,
+      "step": 397
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.88e-06,
+      "loss": 0.3486,
+      "step": 398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.9e-06,
+      "loss": 0.315,
+      "step": 399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.92e-06,
+      "loss": 0.3483,
+      "step": 400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.94e-06,
+      "loss": 0.4146,
+      "step": 401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 0.3463,
+      "step": 402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.285,
+      "step": 403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.3474,
+      "step": 404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 0.2894,
+      "step": 405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 0.3511,
+      "step": 406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.06e-06,
+      "loss": 0.2705,
+      "step": 407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.08e-06,
+      "loss": 0.4045,
+      "step": 408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.1e-06,
+      "loss": 0.3134,
+      "step": 409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 0.3489,
+      "step": 410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.14e-06,
+      "loss": 0.4218,
+      "step": 411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.16e-06,
+      "loss": 0.1292,
+      "step": 412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.18e-06,
+      "loss": 0.3295,
+      "step": 413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.2e-06,
+      "loss": 0.2268,
+      "step": 414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 0.2561,
+      "step": 415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.24e-06,
+      "loss": 0.2405,
+      "step": 416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.26e-06,
+      "loss": 0.3239,
+      "step": 417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.28e-06,
+      "loss": 0.3493,
+      "step": 418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.3e-06,
+      "loss": 0.305,
+      "step": 419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.32e-06,
+      "loss": 0.3314,
+      "step": 420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.34e-06,
+      "loss": 0.2124,
+      "step": 421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.36e-06,
+      "loss": 0.2721,
+      "step": 422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.4691,
+      "step": 423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.2838,
+      "step": 424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.42e-06,
+      "loss": 0.3031,
+      "step": 425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.44e-06,
+      "loss": 0.2494,
+      "step": 426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.46e-06,
+      "loss": 0.3551,
+      "step": 427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.48e-06,
+      "loss": 0.2791,
+      "step": 428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.5e-06,
+      "loss": 0.3033,
+      "step": 429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52e-06,
+      "loss": 0.308,
+      "step": 430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 0.2979,
+      "step": 431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 0.359,
+      "step": 432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.2798,
+      "step": 433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6e-06,
+      "loss": 0.3068,
+      "step": 434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62e-06,
+      "loss": 0.2803,
+      "step": 435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64e-06,
+      "loss": 0.459,
+      "step": 436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.66e-06,
+      "loss": 0.4329,
+      "step": 437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.68e-06,
+      "loss": 0.2973,
+      "step": 438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 0.2501,
+      "step": 439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 0.3572,
+      "step": 440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 0.1968,
+      "step": 441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.76e-06,
+      "loss": 0.2159,
+      "step": 442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.78e-06,
+      "loss": 0.2454,
+      "step": 443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.8e-06,
+      "loss": 0.3614,
+      "step": 444
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.82e-06,
+      "loss": 0.2704,
+      "step": 445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.84e-06,
+      "loss": 0.428,
+      "step": 446
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 0.2519,
+      "step": 447
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 0.2947,
+      "step": 448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 0.3261,
+      "step": 449
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 0.1945,
+      "step": 450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.94e-06,
+      "loss": 0.2397,
+      "step": 451
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.96e-06,
+      "loss": 0.3664,
+      "step": 452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.98e-06,
+      "loss": 0.4378,
+      "step": 453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9e-06,
+      "loss": 0.2026,
+      "step": 454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 0.2417,
+      "step": 455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 0.3146,
+      "step": 456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 0.2723,
+      "step": 457
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 0.2811,
+      "step": 458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 0.324,
+      "step": 459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.12e-06,
+      "loss": 0.1598,
+      "step": 460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.14e-06,
+      "loss": 0.1718,
+      "step": 461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.16e-06,
+      "loss": 0.2258,
+      "step": 462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.1702,
+      "step": 463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.2132,
+      "step": 464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 0.3255,
+      "step": 465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 0.2599,
+      "step": 466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 0.2064,
+      "step": 467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 0.4327,
+      "step": 468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.3e-06,
+      "loss": 0.3729,
+      "step": 469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.32e-06,
+      "loss": 0.2827,
+      "step": 470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 0.3283,
+      "step": 471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 0.1889,
+      "step": 472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.38e-06,
+      "loss": 0.3773,
+      "step": 473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.4e-06,
+      "loss": 0.261,
+      "step": 474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.42e-06,
+      "loss": 0.246,
+      "step": 475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 0.2314,
+      "step": 476
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 0.4631,
+      "step": 477
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.48e-06,
+      "loss": 0.3212,
+      "step": 478
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.5e-06,
+      "loss": 0.3073,
+      "step": 479
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.52e-06,
+      "loss": 0.3138,
+      "step": 480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.54e-06,
+      "loss": 0.2911,
+      "step": 481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.56e-06,
+      "loss": 0.5408,
+      "step": 482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.58e-06,
+      "loss": 0.3885,
+      "step": 483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.3509,
+      "step": 484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 0.2593,
+      "step": 485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 0.2224,
+      "step": 486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.66e-06,
+      "loss": 0.2753,
+      "step": 487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.68e-06,
+      "loss": 0.3228,
+      "step": 488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.7e-06,
+      "loss": 0.335,
+      "step": 489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.72e-06,
+      "loss": 0.2073,
+      "step": 490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.74e-06,
+      "loss": 0.2735,
+      "step": 491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 0.2525,
+      "step": 492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.2572,
+      "step": 493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.3139,
+      "step": 494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 0.2932,
+      "step": 495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.84e-06,
+      "loss": 0.1818,
+      "step": 496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.86e-06,
+      "loss": 0.4578,
+      "step": 497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.88e-06,
+      "loss": 0.1838,
+      "step": 498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.9e-06,
+      "loss": 0.3674,
+      "step": 499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 0.3474,
+      "step": 500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 0.1497,
+      "step": 501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 0.265,
+      "step": 502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.2817,
+      "step": 503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1e-05,
+      "loss": 0.4129,
+      "step": 504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.996e-06,
+      "loss": 0.3632,
+      "step": 505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.992e-06,
+      "loss": 0.2263,
+      "step": 506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.988000000000002e-06,
+      "loss": 0.3614,
+      "step": 507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.984e-06,
+      "loss": 0.1892,
+      "step": 508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.3237,
+      "step": 509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.976e-06,
+      "loss": 0.2606,
+      "step": 510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.972e-06,
+      "loss": 0.3468,
+      "step": 511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.968000000000002e-06,
+      "loss": 0.21,
+      "step": 512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.964e-06,
+      "loss": 0.3122,
+      "step": 513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 0.2837,
+      "step": 514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.956000000000001e-06,
+      "loss": 0.2633,
+      "step": 515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.952e-06,
+      "loss": 0.221,
+      "step": 516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.948e-06,
+      "loss": 0.2946,
+      "step": 517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.944e-06,
+      "loss": 0.2395,
+      "step": 518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 0.3417,
+      "step": 519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.936000000000001e-06,
+      "loss": 0.369,
+      "step": 520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.932e-06,
+      "loss": 0.2595,
+      "step": 521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.928e-06,
+      "loss": 0.3971,
+      "step": 522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.924e-06,
+      "loss": 0.331,
+      "step": 523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 0.3097,
+      "step": 524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.916000000000001e-06,
+      "loss": 0.4067,
+      "step": 525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.912000000000001e-06,
+      "loss": 0.3422,
+      "step": 526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.908e-06,
+      "loss": 0.2375,
+      "step": 527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.904e-06,
+      "loss": 0.2094,
+      "step": 528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.9e-06,
+      "loss": 0.4648,
+      "step": 529
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.896000000000001e-06,
+      "loss": 0.1561,
+      "step": 530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.892000000000001e-06,
+      "loss": 0.2537,
+      "step": 531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.888000000000001e-06,
+      "loss": 0.3205,
+      "step": 532
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.884e-06,
+      "loss": 0.32,
+      "step": 533
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.88e-06,
+      "loss": 0.1713,
+      "step": 534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.876000000000002e-06,
+      "loss": 0.2861,
+      "step": 535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.872e-06,
+      "loss": 0.3739,
+      "step": 536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.868000000000001e-06,
+      "loss": 0.3212,
+      "step": 537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.864e-06,
+      "loss": 0.2403,
+      "step": 538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.86e-06,
+      "loss": 0.2844,
+      "step": 539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.856000000000002e-06,
+      "loss": 0.2649,
+      "step": 540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.852e-06,
+      "loss": 0.2897,
+      "step": 541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.848000000000001e-06,
+      "loss": 0.2594,
+      "step": 542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.844000000000001e-06,
+      "loss": 0.2634,
+      "step": 543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.84e-06,
+      "loss": 0.3143,
+      "step": 544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.836e-06,
+      "loss": 0.3749,
+      "step": 545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.832e-06,
+      "loss": 0.2496,
+      "step": 546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.828000000000001e-06,
+      "loss": 0.3391,
+      "step": 547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.824000000000001e-06,
+      "loss": 0.2473,
+      "step": 548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 0.4312,
+      "step": 549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.816e-06,
+      "loss": 0.2964,
+      "step": 550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.812e-06,
+      "loss": 0.2127,
+      "step": 551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.808000000000002e-06,
+      "loss": 0.3125,
+      "step": 552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.804000000000001e-06,
+      "loss": 0.3331,
+      "step": 553
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.3688,
+      "step": 554
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.796e-06,
+      "loss": 0.4953,
+      "step": 555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.792e-06,
+      "loss": 0.1985,
+      "step": 556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.788e-06,
+      "loss": 0.3722,
+      "step": 557
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.784000000000002e-06,
+      "loss": 0.2816,
+      "step": 558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.2636,
+      "step": 559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.776000000000001e-06,
+      "loss": 0.4089,
+      "step": 560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.772e-06,
+      "loss": 0.2724,
+      "step": 561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.768e-06,
+      "loss": 0.3668,
+      "step": 562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.764000000000002e-06,
+      "loss": 0.2699,
+      "step": 563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 0.2179,
+      "step": 564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.756000000000001e-06,
+      "loss": 0.1884,
+      "step": 565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.752e-06,
+      "loss": 0.2811,
+      "step": 566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.748e-06,
+      "loss": 0.2378,
+      "step": 567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.744000000000002e-06,
+      "loss": 0.4268,
+      "step": 568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.74e-06,
+      "loss": 0.218,
+      "step": 569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.736000000000001e-06,
+      "loss": 0.3149,
+      "step": 570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.732000000000001e-06,
+      "loss": 0.3564,
+      "step": 571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.728e-06,
+      "loss": 0.343,
+      "step": 572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.724e-06,
+      "loss": 0.2316,
+      "step": 573
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.72e-06,
+      "loss": 0.2632,
+      "step": 574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.716000000000002e-06,
+      "loss": 0.3552,
+      "step": 575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.712e-06,
+      "loss": 0.3059,
+      "step": 576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.708000000000001e-06,
+      "loss": 0.2269,
+      "step": 577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.704e-06,
+      "loss": 0.2736,
+      "step": 578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.7e-06,
+      "loss": 0.3073,
+      "step": 579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.696000000000002e-06,
+      "loss": 0.3384,
+      "step": 580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.692e-06,
+      "loss": 0.2345,
+      "step": 581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.688000000000001e-06,
+      "loss": 0.2857,
+      "step": 582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.684e-06,
+      "loss": 0.372,
+      "step": 583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.68e-06,
+      "loss": 0.314,
+      "step": 584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.676e-06,
+      "loss": 0.2379,
+      "step": 585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.672e-06,
+      "loss": 0.3584,
+      "step": 586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.668000000000001e-06,
+      "loss": 0.3373,
+      "step": 587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.664000000000001e-06,
+      "loss": 0.3914,
+      "step": 588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.66e-06,
+      "loss": 0.3533,
+      "step": 589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.656e-06,
+      "loss": 0.4148,
+      "step": 590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.652e-06,
+      "loss": 0.335,
+      "step": 591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.648000000000001e-06,
+      "loss": 0.169,
+      "step": 592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.644000000000001e-06,
+      "loss": 0.2846,
+      "step": 593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 0.3905,
+      "step": 594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.636e-06,
+      "loss": 0.1586,
+      "step": 595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.632e-06,
+      "loss": 0.3985,
+      "step": 596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.628e-06,
+      "loss": 0.3578,
+      "step": 597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.624000000000001e-06,
+      "loss": 0.2484,
+      "step": 598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 0.2742,
+      "step": 599
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.616e-06,
+      "loss": 0.3772,
+      "step": 600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.612000000000002e-06,
+      "loss": 0.3139,
+      "step": 601
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.608e-06,
+      "loss": 0.3422,
+      "step": 602
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.604000000000002e-06,
+      "loss": 0.348,
+      "step": 603
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.2182,
+      "step": 604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.596000000000001e-06,
+      "loss": 0.3246,
+      "step": 605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.592e-06,
+      "loss": 0.3222,
+      "step": 606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.588e-06,
+      "loss": 0.1845,
+      "step": 607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.584000000000002e-06,
+      "loss": 0.3602,
+      "step": 608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.58e-06,
+      "loss": 0.3014,
+      "step": 609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.576000000000001e-06,
+      "loss": 0.3469,
+      "step": 610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.572000000000001e-06,
+      "loss": 0.2426,
+      "step": 611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.568e-06,
+      "loss": 0.2529,
+      "step": 612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.564e-06,
+      "loss": 0.3864,
+      "step": 613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.56e-06,
+      "loss": 0.2398,
+      "step": 614
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.556000000000001e-06,
+      "loss": 0.2013,
+      "step": 615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.552000000000001e-06,
+      "loss": 0.2163,
+      "step": 616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.548e-06,
+      "loss": 0.3782,
+      "step": 617
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.544e-06,
+      "loss": 0.2869,
+      "step": 618
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.54e-06,
+      "loss": 0.3902,
+      "step": 619
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.536000000000002e-06,
+      "loss": 0.2746,
+      "step": 620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.532000000000001e-06,
+      "loss": 0.2899,
+      "step": 621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.528000000000001e-06,
+      "loss": 0.3081,
+      "step": 622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.524e-06,
+      "loss": 0.31,
+      "step": 623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.52e-06,
+      "loss": 0.1896,
+      "step": 624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.516e-06,
+      "loss": 0.2874,
+      "step": 625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.512000000000001e-06,
+      "loss": 0.2751,
+      "step": 626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.508000000000001e-06,
+      "loss": 0.3423,
+      "step": 627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.504e-06,
+      "loss": 0.2271,
+      "step": 628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.5e-06,
+      "loss": 0.2415,
+      "step": 629
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.496e-06,
+      "loss": 0.3402,
+      "step": 630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.492000000000002e-06,
+      "loss": 0.2455,
+      "step": 631
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.488000000000001e-06,
+      "loss": 0.4361,
+      "step": 632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.484000000000001e-06,
+      "loss": 0.3939,
+      "step": 633
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.48e-06,
+      "loss": 0.4223,
+      "step": 634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.476e-06,
+      "loss": 0.4001,
+      "step": 635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.472000000000002e-06,
+      "loss": 0.2221,
+      "step": 636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.468e-06,
+      "loss": 0.3439,
+      "step": 637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.464000000000001e-06,
+      "loss": 0.2468,
+      "step": 638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 0.287,
+      "step": 639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.456e-06,
+      "loss": 0.3115,
+      "step": 640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.452000000000002e-06,
+      "loss": 0.2726,
+      "step": 641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.448e-06,
+      "loss": 0.2245,
+      "step": 642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.444000000000001e-06,
+      "loss": 0.4504,
+      "step": 643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 0.2485,
+      "step": 644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.436e-06,
+      "loss": 0.1988,
+      "step": 645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.432e-06,
+      "loss": 0.2461,
+      "step": 646
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.428e-06,
+      "loss": 0.369,
+      "step": 647
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.424000000000002e-06,
+      "loss": 0.2894,
+      "step": 648
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.42e-06,
+      "loss": 0.3942,
+      "step": 649
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.416000000000001e-06,
+      "loss": 0.2274,
+      "step": 650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.412e-06,
+      "loss": 0.304,
+      "step": 651
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.408e-06,
+      "loss": 0.2698,
+      "step": 652
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.404e-06,
+      "loss": 0.2397,
+      "step": 653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.4e-06,
+      "loss": 0.3322,
+      "step": 654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.396000000000001e-06,
+      "loss": 0.3536,
+      "step": 655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.392000000000001e-06,
+      "loss": 0.3073,
+      "step": 656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.388e-06,
+      "loss": 0.2737,
+      "step": 657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.384e-06,
+      "loss": 0.3598,
+      "step": 658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.38e-06,
+      "loss": 0.2622,
+      "step": 659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.376000000000001e-06,
+      "loss": 0.2508,
+      "step": 660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.372000000000001e-06,
+      "loss": 0.2842,
+      "step": 661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.368e-06,
+      "loss": 0.24,
+      "step": 662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.364e-06,
+      "loss": 0.1773,
+      "step": 663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 0.2817,
+      "step": 664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.356e-06,
+      "loss": 0.2767,
+      "step": 665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.352000000000001e-06,
+      "loss": 0.2552,
+      "step": 666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.348000000000001e-06,
+      "loss": 0.1924,
+      "step": 667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.344e-06,
+      "loss": 0.182,
+      "step": 668
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 0.4504,
+      "step": 669
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.336e-06,
+      "loss": 0.2314,
+      "step": 670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.332000000000001e-06,
+      "loss": 0.2739,
+      "step": 671
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.328000000000001e-06,
+      "loss": 0.246,
+      "step": 672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.324000000000001e-06,
+      "loss": 0.3521,
+      "step": 673
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.32e-06,
+      "loss": 0.2002,
+      "step": 674
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.316e-06,
+      "loss": 0.2559,
+      "step": 675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.312000000000002e-06,
+      "loss": 0.2834,
+      "step": 676
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.308e-06,
+      "loss": 0.249,
+      "step": 677
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.304000000000001e-06,
+      "loss": 0.2359,
+      "step": 678
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.3e-06,
+      "loss": 0.1828,
+      "step": 679
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.296e-06,
+      "loss": 0.3129,
+      "step": 680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.292000000000002e-06,
+      "loss": 0.3182,
+      "step": 681
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.288e-06,
+      "loss": 0.3882,
+      "step": 682
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.284000000000001e-06,
+      "loss": 0.4197,
+      "step": 683
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 0.1819,
+      "step": 684
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.276e-06,
+      "loss": 0.2142,
+      "step": 685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.272e-06,
+      "loss": 0.2563,
+      "step": 686
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.268e-06,
+      "loss": 0.232,
+      "step": 687
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.264000000000001e-06,
+      "loss": 0.3775,
+      "step": 688
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 0.2043,
+      "step": 689
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.256e-06,
+      "loss": 0.2981,
+      "step": 690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.252e-06,
+      "loss": 0.3429,
+      "step": 691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.248e-06,
+      "loss": 0.4232,
+      "step": 692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.244e-06,
+      "loss": 0.2896,
+      "step": 693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 0.2443,
+      "step": 694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.236000000000001e-06,
+      "loss": 0.2761,
+      "step": 695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.232e-06,
+      "loss": 0.313,
+      "step": 696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.228e-06,
+      "loss": 0.2477,
+      "step": 697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.224e-06,
+      "loss": 0.3407,
+      "step": 698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 0.3627,
+      "step": 699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.216000000000001e-06,
+      "loss": 0.1817,
+      "step": 700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.212000000000001e-06,
+      "loss": 0.2482,
+      "step": 701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.208e-06,
+      "loss": 0.2465,
+      "step": 702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.204e-06,
+      "loss": 0.3675,
+      "step": 703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.2657,
+      "step": 704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.196e-06,
+      "loss": 0.2882,
+      "step": 705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.192000000000001e-06,
+      "loss": 0.3487,
+      "step": 706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.188e-06,
+      "loss": 0.2705,
+      "step": 707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.184e-06,
+      "loss": 0.3923,
+      "step": 708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.1698,
+      "step": 709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.176e-06,
+      "loss": 0.1779,
+      "step": 710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.172000000000001e-06,
+      "loss": 0.321,
+      "step": 711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.168000000000001e-06,
+      "loss": 0.3136,
+      "step": 712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.164e-06,
+      "loss": 0.3573,
+      "step": 713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.16e-06,
+      "loss": 0.2281,
+      "step": 714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.156e-06,
+      "loss": 0.3204,
+      "step": 715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.152000000000001e-06,
+      "loss": 0.3089,
+      "step": 716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.148e-06,
+      "loss": 0.2835,
+      "step": 717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.144000000000001e-06,
+      "loss": 0.2605,
+      "step": 718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.14e-06,
+      "loss": 0.2242,
+      "step": 719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.136e-06,
+      "loss": 0.1901,
+      "step": 720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.132000000000002e-06,
+      "loss": 0.2148,
+      "step": 721
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.128e-06,
+      "loss": 0.3584,
+      "step": 722
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.124000000000001e-06,
+      "loss": 0.4136,
+      "step": 723
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.12e-06,
+      "loss": 0.2754,
+      "step": 724
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.116e-06,
+      "loss": 0.4235,
+      "step": 725
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.112e-06,
+      "loss": 0.227,
+      "step": 726
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.108000000000002e-06,
+      "loss": 0.1987,
+      "step": 727
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.104000000000001e-06,
+      "loss": 0.2156,
+      "step": 728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 0.1791,
+      "step": 729
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.096e-06,
+      "loss": 0.2886,
+      "step": 730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.092e-06,
+      "loss": 0.2282,
+      "step": 731
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.088000000000002e-06,
+      "loss": 0.1876,
+      "step": 732
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.084e-06,
+      "loss": 0.2989,
+      "step": 733
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 0.3286,
+      "step": 734
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.076000000000001e-06,
+      "loss": 0.2473,
+      "step": 735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.072e-06,
+      "loss": 0.3985,
+      "step": 736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.068000000000002e-06,
+      "loss": 0.2999,
+      "step": 737
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.064e-06,
+      "loss": 0.2273,
+      "step": 738
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 0.244,
+      "step": 739
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.056000000000001e-06,
+      "loss": 0.2679,
+      "step": 740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.052e-06,
+      "loss": 0.2926,
+      "step": 741
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.048e-06,
+      "loss": 0.3593,
+      "step": 742
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.044e-06,
+      "loss": 0.6355,
+      "step": 743
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 0.2782,
+      "step": 744
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.036e-06,
+      "loss": 0.3379,
+      "step": 745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.032000000000001e-06,
+      "loss": 0.2195,
+      "step": 746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.028e-06,
+      "loss": 0.3361,
+      "step": 747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.024e-06,
+      "loss": 0.2818,
+      "step": 748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 0.3004,
+      "step": 749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.016e-06,
+      "loss": 0.2489,
+      "step": 750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.012000000000001e-06,
+      "loss": 0.1813,
+      "step": 751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.008e-06,
+      "loss": 0.2775,
+      "step": 752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.004e-06,
+      "loss": 0.2917,
+      "step": 753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9e-06,
+      "loss": 0.3361,
+      "step": 754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.996e-06,
+      "loss": 0.3199,
+      "step": 755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.992000000000001e-06,
+      "loss": 0.3171,
+      "step": 756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.988000000000001e-06,
+      "loss": 0.3482,
+      "step": 757
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.984e-06,
+      "loss": 0.1777,
+      "step": 758
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.98e-06,
+      "loss": 0.295,
+      "step": 759
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.976e-06,
+      "loss": 0.3122,
+      "step": 760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.972000000000002e-06,
+      "loss": 0.3317,
+      "step": 761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.968000000000001e-06,
+      "loss": 0.2925,
+      "step": 762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.964000000000001e-06,
+      "loss": 0.2616,
+      "step": 763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.96e-06,
+      "loss": 0.3828,
+      "step": 764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.956e-06,
+      "loss": 0.2886,
+      "step": 765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.952e-06,
+      "loss": 0.3652,
+      "step": 766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.948000000000001e-06,
+      "loss": 0.3134,
+      "step": 767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.944000000000001e-06,
+      "loss": 0.252,
+      "step": 768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.94e-06,
+      "loss": 0.2338,
+      "step": 769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.936e-06,
+      "loss": 0.1338,
+      "step": 770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.932e-06,
+      "loss": 0.2757,
+      "step": 771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.928000000000002e-06,
+      "loss": 0.3608,
+      "step": 772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.924e-06,
+      "loss": 0.3668,
+      "step": 773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 0.1438,
+      "step": 774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.916e-06,
+      "loss": 0.1254,
+      "step": 775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.912e-06,
+      "loss": 0.1565,
+      "step": 776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.908000000000002e-06,
+      "loss": 0.2048,
+      "step": 777
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.904e-06,
+      "loss": 0.1604,
+      "step": 778
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 0.0888,
+      "step": 779
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.896000000000001e-06,
+      "loss": 0.0749,
+      "step": 780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.892e-06,
+      "loss": 0.1513,
+      "step": 781
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.888e-06,
+      "loss": 0.0951,
+      "step": 782
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.884e-06,
+      "loss": 0.1925,
+      "step": 783
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 0.1755,
+      "step": 784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.876e-06,
+      "loss": 0.1329,
+      "step": 785
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.872e-06,
+      "loss": 0.1521,
+      "step": 786
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.868e-06,
+      "loss": 0.237,
+      "step": 787
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.864e-06,
+      "loss": 0.1038,
+      "step": 788
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 0.1183,
+      "step": 789
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.856000000000001e-06,
+      "loss": 0.1859,
+      "step": 790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.852000000000001e-06,
+      "loss": 0.1116,
+      "step": 791
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.848e-06,
+      "loss": 0.1447,
+      "step": 792
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.844e-06,
+      "loss": 0.1164,
+      "step": 793
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.84e-06,
+      "loss": 0.1807,
+      "step": 794
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.836000000000001e-06,
+      "loss": 0.0992,
+      "step": 795
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.832000000000001e-06,
+      "loss": 0.1047,
+      "step": 796
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.828000000000001e-06,
+      "loss": 0.079,
+      "step": 797
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.824e-06,
+      "loss": 0.1567,
+      "step": 798
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.82e-06,
+      "loss": 0.1279,
+      "step": 799
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.816000000000002e-06,
+      "loss": 0.1195,
+      "step": 800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.812000000000001e-06,
+      "loss": 0.0888,
+      "step": 801
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.808000000000001e-06,
+      "loss": 0.2005,
+      "step": 802
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.804e-06,
+      "loss": 0.1651,
+      "step": 803
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.8e-06,
+      "loss": 0.1014,
+      "step": 804
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.796000000000002e-06,
+      "loss": 0.1609,
+      "step": 805
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.792e-06,
+      "loss": 0.1158,
+      "step": 806
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.788000000000001e-06,
+      "loss": 0.1546,
+      "step": 807
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.784000000000001e-06,
+      "loss": 0.1654,
+      "step": 808
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.78e-06,
+      "loss": 0.1998,
+      "step": 809
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.776e-06,
+      "loss": 0.1242,
+      "step": 810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.772e-06,
+      "loss": 0.2829,
+      "step": 811
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.768000000000001e-06,
+      "loss": 0.1192,
+      "step": 812
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.764e-06,
+      "loss": 0.1178,
+      "step": 813
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.76e-06,
+      "loss": 0.1474,
+      "step": 814
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.756e-06,
+      "loss": 0.1169,
+      "step": 815
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.752e-06,
+      "loss": 0.1215,
+      "step": 816
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.748000000000002e-06,
+      "loss": 0.1408,
+      "step": 817
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.744e-06,
+      "loss": 0.0953,
+      "step": 818
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 0.131,
+      "step": 819
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.736e-06,
+      "loss": 0.0889,
+      "step": 820
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.732e-06,
+      "loss": 0.1747,
+      "step": 821
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.728e-06,
+      "loss": 0.1724,
+      "step": 822
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.724e-06,
+      "loss": 0.1865,
+      "step": 823
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 0.1988,
+      "step": 824
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.716000000000001e-06,
+      "loss": 0.1187,
+      "step": 825
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.712e-06,
+      "loss": 0.0973,
+      "step": 826
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.708e-06,
+      "loss": 0.1722,
+      "step": 827
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.704e-06,
+      "loss": 0.1176,
+      "step": 828
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 0.1163,
+      "step": 829
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.696000000000001e-06,
+      "loss": 0.1267,
+      "step": 830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.692e-06,
+      "loss": 0.1044,
+      "step": 831
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.688e-06,
+      "loss": 0.1249,
+      "step": 832
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.684e-06,
+      "loss": 0.1715,
+      "step": 833
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.68e-06,
+      "loss": 0.1314,
+      "step": 834
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.676000000000001e-06,
+      "loss": 0.2246,
+      "step": 835
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.672000000000001e-06,
+      "loss": 0.1919,
+      "step": 836
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.668e-06,
+      "loss": 0.0998,
+      "step": 837
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.664e-06,
+      "loss": 0.16,
+      "step": 838
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.66e-06,
+      "loss": 0.0727,
+      "step": 839
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.656000000000001e-06,
+      "loss": 0.2466,
+      "step": 840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.652000000000001e-06,
+      "loss": 0.1847,
+      "step": 841
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.648000000000001e-06,
+      "loss": 0.1088,
+      "step": 842
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.644e-06,
+      "loss": 0.1432,
+      "step": 843
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.64e-06,
+      "loss": 0.143,
+      "step": 844
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.636000000000002e-06,
+      "loss": 0.1621,
+      "step": 845
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.632e-06,
+      "loss": 0.2637,
+      "step": 846
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.628000000000001e-06,
+      "loss": 0.1955,
+      "step": 847
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.624e-06,
+      "loss": 0.1412,
+      "step": 848
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.62e-06,
+      "loss": 0.0522,
+      "step": 849
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.616000000000002e-06,
+      "loss": 0.156,
+      "step": 850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.612e-06,
+      "loss": 0.2004,
+      "step": 851
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.608000000000001e-06,
+      "loss": 0.0463,
+      "step": 852
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.604000000000001e-06,
+      "loss": 0.1624,
+      "step": 853
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.6e-06,
+      "loss": 0.0692,
+      "step": 854
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.596e-06,
+      "loss": 0.0943,
+      "step": 855
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.592e-06,
+      "loss": 0.1161,
+      "step": 856
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.588000000000001e-06,
+      "loss": 0.2075,
+      "step": 857
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.584000000000001e-06,
+      "loss": 0.1856,
+      "step": 858
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.1388,
+      "step": 859
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.576e-06,
+      "loss": 0.1465,
+      "step": 860
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.572e-06,
+      "loss": 0.1504,
+      "step": 861
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.568e-06,
+      "loss": 0.1524,
+      "step": 862
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.564000000000001e-06,
+      "loss": 0.1402,
+      "step": 863
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 0.1331,
+      "step": 864
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.556e-06,
+      "loss": 0.1277,
+      "step": 865
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.552e-06,
+      "loss": 0.1184,
+      "step": 866
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.548e-06,
+      "loss": 0.1165,
+      "step": 867
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.544000000000002e-06,
+      "loss": 0.1155,
+      "step": 868
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 0.1728,
+      "step": 869
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.536000000000001e-06,
+      "loss": 0.143,
+      "step": 870
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.532e-06,
+      "loss": 0.1313,
+      "step": 871
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.528e-06,
+      "loss": 0.0926,
+      "step": 872
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.524000000000002e-06,
+      "loss": 0.1128,
+      "step": 873
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.52e-06,
+      "loss": 0.0934,
+      "step": 874
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.516000000000001e-06,
+      "loss": 0.0547,
+      "step": 875
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.512e-06,
+      "loss": 0.3281,
+      "step": 876
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.508e-06,
+      "loss": 0.0906,
+      "step": 877
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.504000000000002e-06,
+      "loss": 0.1503,
+      "step": 878
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.5e-06,
+      "loss": 0.1083,
+      "step": 879
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.496000000000001e-06,
+      "loss": 0.2172,
+      "step": 880
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.492000000000001e-06,
+      "loss": 0.2603,
+      "step": 881
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.488e-06,
+      "loss": 0.2176,
+      "step": 882
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.484e-06,
+      "loss": 0.1908,
+      "step": 883
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.48e-06,
+      "loss": 0.1277,
+      "step": 884
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.476000000000002e-06,
+      "loss": 0.1062,
+      "step": 885
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.472e-06,
+      "loss": 0.1235,
+      "step": 886
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.468000000000001e-06,
+      "loss": 0.1725,
+      "step": 887
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.464e-06,
+      "loss": 0.1147,
+      "step": 888
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.46e-06,
+      "loss": 0.1095,
+      "step": 889
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.456000000000002e-06,
+      "loss": 0.0996,
+      "step": 890
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.452e-06,
+      "loss": 0.1072,
+      "step": 891
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.448000000000001e-06,
+      "loss": 0.0952,
+      "step": 892
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.444e-06,
+      "loss": 0.1173,
+      "step": 893
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.44e-06,
+      "loss": 0.1329,
+      "step": 894
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.436e-06,
+      "loss": 0.142,
+      "step": 895
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.432e-06,
+      "loss": 0.0892,
+      "step": 896
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.428000000000001e-06,
+      "loss": 0.1121,
+      "step": 897
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.424000000000001e-06,
+      "loss": 0.1789,
+      "step": 898
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.42e-06,
+      "loss": 0.2394,
+      "step": 899
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.416e-06,
+      "loss": 0.1111,
+      "step": 900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.412e-06,
+      "loss": 0.1132,
+      "step": 901
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.408e-06,
+      "loss": 0.104,
+      "step": 902
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.404000000000001e-06,
+      "loss": 0.1279,
+      "step": 903
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.1059,
+      "step": 904
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.396e-06,
+      "loss": 0.1628,
+      "step": 905
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.392e-06,
+      "loss": 0.2063,
+      "step": 906
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.388e-06,
+      "loss": 0.1683,
+      "step": 907
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.384000000000001e-06,
+      "loss": 0.189,
+      "step": 908
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.2506,
+      "step": 909
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.376e-06,
+      "loss": 0.2187,
+      "step": 910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.372e-06,
+      "loss": 0.1443,
+      "step": 911
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.368e-06,
+      "loss": 0.2061,
+      "step": 912
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.364000000000002e-06,
+      "loss": 0.1017,
+      "step": 913
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.36e-06,
+      "loss": 0.1954,
+      "step": 914
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.356000000000001e-06,
+      "loss": 0.1428,
+      "step": 915
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.352e-06,
+      "loss": 0.1576,
+      "step": 916
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.348e-06,
+      "loss": 0.1385,
+      "step": 917
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.344000000000002e-06,
+      "loss": 0.1192,
+      "step": 918
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.34e-06,
+      "loss": 0.118,
+      "step": 919
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.336000000000001e-06,
+      "loss": 0.1793,
+      "step": 920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.332000000000001e-06,
+      "loss": 0.0945,
+      "step": 921
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.328e-06,
+      "loss": 0.157,
+      "step": 922
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.324e-06,
+      "loss": 0.0886,
+      "step": 923
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.32e-06,
+      "loss": 0.2463,
+      "step": 924
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.316000000000001e-06,
+      "loss": 0.1773,
+      "step": 925
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.312000000000001e-06,
+      "loss": 0.131,
+      "step": 926
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.308e-06,
+      "loss": 0.1555,
+      "step": 927
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.304e-06,
+      "loss": 0.1896,
+      "step": 928
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.3e-06,
+      "loss": 0.1232,
+      "step": 929
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.296000000000002e-06,
+      "loss": 0.1613,
+      "step": 930
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.292000000000001e-06,
+      "loss": 0.1605,
+      "step": 931
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.288000000000001e-06,
+      "loss": 0.1627,
+      "step": 932
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.284e-06,
+      "loss": 0.0917,
+      "step": 933
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.28e-06,
+      "loss": 0.101,
+      "step": 934
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.276e-06,
+      "loss": 0.1426,
+      "step": 935
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.272000000000001e-06,
+      "loss": 0.2281,
+      "step": 936
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.268000000000001e-06,
+      "loss": 0.1234,
+      "step": 937
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.264e-06,
+      "loss": 0.1608,
+      "step": 938
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.26e-06,
+      "loss": 0.1038,
+      "step": 939
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.256e-06,
+      "loss": 0.212,
+      "step": 940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.252000000000002e-06,
+      "loss": 0.2145,
+      "step": 941
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.248e-06,
+      "loss": 0.137,
+      "step": 942
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.244000000000001e-06,
+      "loss": 0.1402,
+      "step": 943
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.24e-06,
+      "loss": 0.1558,
+      "step": 944
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.236e-06,
+      "loss": 0.1523,
+      "step": 945
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.232000000000002e-06,
+      "loss": 0.1776,
+      "step": 946
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.228e-06,
+      "loss": 0.2269,
+      "step": 947
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.224000000000001e-06,
+      "loss": 0.1123,
+      "step": 948
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 0.1269,
+      "step": 949
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.216e-06,
+      "loss": 0.0989,
+      "step": 950
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.212e-06,
+      "loss": 0.1036,
+      "step": 951
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.208e-06,
+      "loss": 0.1431,
+      "step": 952
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.204000000000001e-06,
+      "loss": 0.1782,
+      "step": 953
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.2e-06,
+      "loss": 0.0816,
+      "step": 954
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.196e-06,
+      "loss": 0.1293,
+      "step": 955
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.192e-06,
+      "loss": 0.189,
+      "step": 956
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.188e-06,
+      "loss": 0.1025,
+      "step": 957
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.184000000000002e-06,
+      "loss": 0.1473,
+      "step": 958
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.18e-06,
+      "loss": 0.2016,
+      "step": 959
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.176000000000001e-06,
+      "loss": 0.2099,
+      "step": 960
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.172e-06,
+      "loss": 0.0965,
+      "step": 961
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.168e-06,
+      "loss": 0.144,
+      "step": 962
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.164e-06,
+      "loss": 0.1238,
+      "step": 963
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.16e-06,
+      "loss": 0.223,
+      "step": 964
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.156000000000001e-06,
+      "loss": 0.0954,
+      "step": 965
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.152000000000001e-06,
+      "loss": 0.0897,
+      "step": 966
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.148e-06,
+      "loss": 0.1558,
+      "step": 967
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.144e-06,
+      "loss": 0.1342,
+      "step": 968
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.14e-06,
+      "loss": 0.1415,
+      "step": 969
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.136000000000001e-06,
+      "loss": 0.1176,
+      "step": 970
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.132000000000001e-06,
+      "loss": 0.3732,
+      "step": 971
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.128e-06,
+      "loss": 0.1233,
+      "step": 972
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.124e-06,
+      "loss": 0.0987,
+      "step": 973
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 0.1321,
+      "step": 974
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.116e-06,
+      "loss": 0.1718,
+      "step": 975
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.112000000000001e-06,
+      "loss": 0.1323,
+      "step": 976
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.108000000000001e-06,
+      "loss": 0.1366,
+      "step": 977
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.104e-06,
+      "loss": 0.11,
+      "step": 978
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.1e-06,
+      "loss": 0.264,
+      "step": 979
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.096e-06,
+      "loss": 0.1057,
+      "step": 980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.092000000000001e-06,
+      "loss": 0.0768,
+      "step": 981
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.088e-06,
+      "loss": 0.2009,
+      "step": 982
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.084000000000001e-06,
+      "loss": 0.1345,
+      "step": 983
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.08e-06,
+      "loss": 0.0929,
+      "step": 984
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.076e-06,
+      "loss": 0.1056,
+      "step": 985
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.072000000000002e-06,
+      "loss": 0.2265,
+      "step": 986
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.068e-06,
+      "loss": 0.1408,
+      "step": 987
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.064000000000001e-06,
+      "loss": 0.1028,
+      "step": 988
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.06e-06,
+      "loss": 0.262,
+      "step": 989
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.056e-06,
+      "loss": 0.1044,
+      "step": 990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.052e-06,
+      "loss": 0.0766,
+      "step": 991
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.048e-06,
+      "loss": 0.1797,
+      "step": 992
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.044000000000001e-06,
+      "loss": 0.1194,
+      "step": 993
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 0.1138,
+      "step": 994
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.036e-06,
+      "loss": 0.0959,
+      "step": 995
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.032e-06,
+      "loss": 0.1396,
+      "step": 996
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.028e-06,
+      "loss": 0.0875,
+      "step": 997
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.024000000000001e-06,
+      "loss": 0.154,
+      "step": 998
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 0.1099,
+      "step": 999
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.016e-06,
+      "loss": 0.1448,
+      "step": 1000
+    },
+    {
+      "epoch": 1.29,
+      "eval_loss": 0.2952553331851959,
+      "eval_runtime": 1898.7791,
+      "eval_samples_per_second": 2.67,
+      "eval_steps_per_second": 0.167,
+      "eval_wer": 21.42453819472453,
+      "step": 1000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.012e-06,
+      "loss": 0.1386,
+      "step": 1001
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.008e-06,
+      "loss": 0.1304,
+      "step": 1002
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.004e-06,
+      "loss": 0.2707,
+      "step": 1003
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.2071,
+      "step": 1004
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.996000000000001e-06,
+      "loss": 0.124,
+      "step": 1005
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.992e-06,
+      "loss": 0.1486,
+      "step": 1006
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.988e-06,
+      "loss": 0.1004,
+      "step": 1007
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.984e-06,
+      "loss": 0.13,
+      "step": 1008
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.1051,
+      "step": 1009
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.976000000000001e-06,
+      "loss": 0.1737,
+      "step": 1010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.972000000000001e-06,
+      "loss": 0.2589,
+      "step": 1011
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.968e-06,
+      "loss": 0.1319,
+      "step": 1012
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.964e-06,
+      "loss": 0.1352,
+      "step": 1013
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 0.2023,
+      "step": 1014
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.956e-06,
+      "loss": 0.1062,
+      "step": 1015
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.952000000000001e-06,
+      "loss": 0.1876,
+      "step": 1016
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.948e-06,
+      "loss": 0.138,
+      "step": 1017
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.944e-06,
+      "loss": 0.2378,
+      "step": 1018
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.94e-06,
+      "loss": 0.1471,
+      "step": 1019
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.936e-06,
+      "loss": 0.1428,
+      "step": 1020
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.932000000000001e-06,
+      "loss": 0.1441,
+      "step": 1021
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.928e-06,
+      "loss": 0.1409,
+      "step": 1022
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.924e-06,
+      "loss": 0.1014,
+      "step": 1023
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.92e-06,
+      "loss": 0.1224,
+      "step": 1024
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.916e-06,
+      "loss": 0.1367,
+      "step": 1025
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.912000000000001e-06,
+      "loss": 0.176,
+      "step": 1026
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.908e-06,
+      "loss": 0.1827,
+      "step": 1027
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.904000000000001e-06,
+      "loss": 0.1543,
+      "step": 1028
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.9e-06,
+      "loss": 0.0909,
+      "step": 1029
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.896e-06,
+      "loss": 0.1185,
+      "step": 1030
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.892e-06,
+      "loss": 0.097,
+      "step": 1031
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.888e-06,
+      "loss": 0.1033,
+      "step": 1032
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.884000000000001e-06,
+      "loss": 0.1572,
+      "step": 1033
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.88e-06,
+      "loss": 0.1049,
+      "step": 1034
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.876e-06,
+      "loss": 0.1372,
+      "step": 1035
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.872e-06,
+      "loss": 0.0853,
+      "step": 1036
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.868000000000002e-06,
+      "loss": 0.1275,
+      "step": 1037
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.864000000000001e-06,
+      "loss": 0.1811,
+      "step": 1038
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 0.1621,
+      "step": 1039
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.856e-06,
+      "loss": 0.1761,
+      "step": 1040
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.852e-06,
+      "loss": 0.1713,
+      "step": 1041
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.848000000000002e-06,
+      "loss": 0.1073,
+      "step": 1042
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.844e-06,
+      "loss": 0.0964,
+      "step": 1043
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 0.1265,
+      "step": 1044
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.836000000000001e-06,
+      "loss": 0.2001,
+      "step": 1045
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.832e-06,
+      "loss": 0.1423,
+      "step": 1046
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.828000000000002e-06,
+      "loss": 0.1714,
+      "step": 1047
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.824e-06,
+      "loss": 0.1265,
+      "step": 1048
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 0.1495,
+      "step": 1049
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.816000000000001e-06,
+      "loss": 0.1195,
+      "step": 1050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.812e-06,
+      "loss": 0.1095,
+      "step": 1051
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.808e-06,
+      "loss": 0.1727,
+      "step": 1052
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.804e-06,
+      "loss": 0.1289,
+      "step": 1053
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 0.1179,
+      "step": 1054
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.796e-06,
+      "loss": 0.162,
+      "step": 1055
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.792000000000001e-06,
+      "loss": 0.1081,
+      "step": 1056
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.788e-06,
+      "loss": 0.0859,
+      "step": 1057
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.784e-06,
+      "loss": 0.1174,
+      "step": 1058
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.78e-06,
+      "loss": 0.173,
+      "step": 1059
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.776e-06,
+      "loss": 0.1816,
+      "step": 1060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.772000000000001e-06,
+      "loss": 0.2123,
+      "step": 1061
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.768e-06,
+      "loss": 0.0951,
+      "step": 1062
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.764e-06,
+      "loss": 0.1144,
+      "step": 1063
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.76e-06,
+      "loss": 0.1382,
+      "step": 1064
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.756e-06,
+      "loss": 0.1122,
+      "step": 1065
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.752000000000001e-06,
+      "loss": 0.0903,
+      "step": 1066
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.748000000000001e-06,
+      "loss": 0.109,
+      "step": 1067
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.744e-06,
+      "loss": 0.1127,
+      "step": 1068
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.74e-06,
+      "loss": 0.2231,
+      "step": 1069
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.736e-06,
+      "loss": 0.1834,
+      "step": 1070
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.732e-06,
+      "loss": 0.2401,
+      "step": 1071
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.728000000000001e-06,
+      "loss": 0.1341,
+      "step": 1072
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.724000000000001e-06,
+      "loss": 0.1427,
+      "step": 1073
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.72e-06,
+      "loss": 0.0939,
+      "step": 1074
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.716e-06,
+      "loss": 0.1018,
+      "step": 1075
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.712e-06,
+      "loss": 0.2179,
+      "step": 1076
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.708000000000001e-06,
+      "loss": 0.0968,
+      "step": 1077
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.704000000000001e-06,
+      "loss": 0.1793,
+      "step": 1078
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.7e-06,
+      "loss": 0.1588,
+      "step": 1079
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.696e-06,
+      "loss": 0.205,
+      "step": 1080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.692e-06,
+      "loss": 0.0825,
+      "step": 1081
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.688000000000002e-06,
+      "loss": 0.1155,
+      "step": 1082
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.684e-06,
+      "loss": 0.1212,
+      "step": 1083
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 0.2666,
+      "step": 1084
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.676e-06,
+      "loss": 0.1301,
+      "step": 1085
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.672e-06,
+      "loss": 0.1972,
+      "step": 1086
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.668000000000002e-06,
+      "loss": 0.0937,
+      "step": 1087
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.664e-06,
+      "loss": 0.0835,
+      "step": 1088
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 0.1889,
+      "step": 1089
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.656000000000001e-06,
+      "loss": 0.1721,
+      "step": 1090
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.652e-06,
+      "loss": 0.0834,
+      "step": 1091
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.648e-06,
+      "loss": 0.157,
+      "step": 1092
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.644e-06,
+      "loss": 0.1933,
+      "step": 1093
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 0.1189,
+      "step": 1094
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.636e-06,
+      "loss": 0.1184,
+      "step": 1095
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.632e-06,
+      "loss": 0.0911,
+      "step": 1096
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.628000000000001e-06,
+      "loss": 0.2054,
+      "step": 1097
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.624e-06,
+      "loss": 0.1153,
+      "step": 1098
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 0.1493,
+      "step": 1099
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.616000000000001e-06,
+      "loss": 0.1763,
+      "step": 1100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.612e-06,
+      "loss": 0.104,
+      "step": 1101
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.608000000000001e-06,
+      "loss": 0.1433,
+      "step": 1102
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.604e-06,
+      "loss": 0.1833,
+      "step": 1103
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.1695,
+      "step": 1104
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.5960000000000015e-06,
+      "loss": 0.1627,
+      "step": 1105
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.592e-06,
+      "loss": 0.0772,
+      "step": 1106
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.588000000000001e-06,
+      "loss": 0.1665,
+      "step": 1107
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.5840000000000006e-06,
+      "loss": 0.1093,
+      "step": 1108
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.58e-06,
+      "loss": 0.1027,
+      "step": 1109
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.576000000000001e-06,
+      "loss": 0.0974,
+      "step": 1110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.5720000000000005e-06,
+      "loss": 0.1246,
+      "step": 1111
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.568000000000001e-06,
+      "loss": 0.2064,
+      "step": 1112
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.564e-06,
+      "loss": 0.1099,
+      "step": 1113
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 0.1755,
+      "step": 1114
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.556000000000001e-06,
+      "loss": 0.0937,
+      "step": 1115
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.552000000000001e-06,
+      "loss": 0.1357,
+      "step": 1116
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.548000000000001e-06,
+      "loss": 0.182,
+      "step": 1117
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.544e-06,
+      "loss": 0.122,
+      "step": 1118
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 0.126,
+      "step": 1119
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.536000000000001e-06,
+      "loss": 0.1463,
+      "step": 1120
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.532e-06,
+      "loss": 0.1445,
+      "step": 1121
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.528000000000001e-06,
+      "loss": 0.1551,
+      "step": 1122
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.524e-06,
+      "loss": 0.1456,
+      "step": 1123
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 0.1625,
+      "step": 1124
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.516000000000001e-06,
+      "loss": 0.1213,
+      "step": 1125
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.512e-06,
+      "loss": 0.1383,
+      "step": 1126
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.508000000000001e-06,
+      "loss": 0.1403,
+      "step": 1127
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.5040000000000005e-06,
+      "loss": 0.1464,
+      "step": 1128
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.0962,
+      "step": 1129
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.496000000000001e-06,
+      "loss": 0.1951,
+      "step": 1130
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.4920000000000004e-06,
+      "loss": 0.1047,
+      "step": 1131
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.488000000000001e-06,
+      "loss": 0.1532,
+      "step": 1132
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.484e-06,
+      "loss": 0.1901,
+      "step": 1133
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.48e-06,
+      "loss": 0.1235,
+      "step": 1134
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.476000000000001e-06,
+      "loss": 0.1245,
+      "step": 1135
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.472000000000001e-06,
+      "loss": 0.2217,
+      "step": 1136
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.468000000000001e-06,
+      "loss": 0.1732,
+      "step": 1137
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.464e-06,
+      "loss": 0.109,
+      "step": 1138
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 0.1847,
+      "step": 1139
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.456000000000001e-06,
+      "loss": 0.0982,
+      "step": 1140
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.452e-06,
+      "loss": 0.2368,
+      "step": 1141
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.4480000000000005e-06,
+      "loss": 0.1808,
+      "step": 1142
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.444e-06,
+      "loss": 0.1195,
+      "step": 1143
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 0.0853,
+      "step": 1144
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.436000000000001e-06,
+      "loss": 0.1248,
+      "step": 1145
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.432e-06,
+      "loss": 0.1166,
+      "step": 1146
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.428000000000001e-06,
+      "loss": 0.1649,
+      "step": 1147
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.424e-06,
+      "loss": 0.0712,
+      "step": 1148
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 0.0825,
+      "step": 1149
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.416000000000001e-06,
+      "loss": 0.161,
+      "step": 1150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.412e-06,
+      "loss": 0.1366,
+      "step": 1151
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.408000000000001e-06,
+      "loss": 0.0737,
+      "step": 1152
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.404e-06,
+      "loss": 0.0914,
+      "step": 1153
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.4e-06,
+      "loss": 0.0783,
+      "step": 1154
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.396000000000001e-06,
+      "loss": 0.2263,
+      "step": 1155
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.3920000000000005e-06,
+      "loss": 0.1318,
+      "step": 1156
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.388000000000001e-06,
+      "loss": 0.1161,
+      "step": 1157
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.384e-06,
+      "loss": 0.1057,
+      "step": 1158
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.0868,
+      "step": 1159
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.376000000000001e-06,
+      "loss": 0.1766,
+      "step": 1160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.372e-06,
+      "loss": 0.1295,
+      "step": 1161
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.3680000000000004e-06,
+      "loss": 0.1443,
+      "step": 1162
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.364000000000001e-06,
+      "loss": 0.1934,
+      "step": 1163
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 0.1559,
+      "step": 1164
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.356000000000001e-06,
+      "loss": 0.141,
+      "step": 1165
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.352e-06,
+      "loss": 0.1531,
+      "step": 1166
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.348000000000001e-06,
+      "loss": 0.0843,
+      "step": 1167
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.344000000000001e-06,
+      "loss": 0.2386,
+      "step": 1168
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 0.1085,
+      "step": 1169
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.3360000000000006e-06,
+      "loss": 0.0712,
+      "step": 1170
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.332e-06,
+      "loss": 0.2099,
+      "step": 1171
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.328000000000001e-06,
+      "loss": 0.1541,
+      "step": 1172
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.324000000000001e-06,
+      "loss": 0.1581,
+      "step": 1173
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.32e-06,
+      "loss": 0.1495,
+      "step": 1174
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.316000000000001e-06,
+      "loss": 0.1153,
+      "step": 1175
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.3120000000000005e-06,
+      "loss": 0.1694,
+      "step": 1176
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.308000000000001e-06,
+      "loss": 0.1652,
+      "step": 1177
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.304000000000001e-06,
+      "loss": 0.1104,
+      "step": 1178
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.3e-06,
+      "loss": 0.1501,
+      "step": 1179
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.296000000000001e-06,
+      "loss": 0.1607,
+      "step": 1180
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.292e-06,
+      "loss": 0.154,
+      "step": 1181
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.288e-06,
+      "loss": 0.1457,
+      "step": 1182
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.284000000000001e-06,
+      "loss": 0.1327,
+      "step": 1183
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 0.1402,
+      "step": 1184
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.276000000000001e-06,
+      "loss": 0.1168,
+      "step": 1185
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.272e-06,
+      "loss": 0.1313,
+      "step": 1186
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.2680000000000005e-06,
+      "loss": 0.2107,
+      "step": 1187
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.264000000000001e-06,
+      "loss": 0.2792,
+      "step": 1188
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 0.1629,
+      "step": 1189
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.2560000000000005e-06,
+      "loss": 0.1017,
+      "step": 1190
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.252e-06,
+      "loss": 0.1681,
+      "step": 1191
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.248000000000001e-06,
+      "loss": 0.1619,
+      "step": 1192
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.244000000000001e-06,
+      "loss": 0.1133,
+      "step": 1193
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.24e-06,
+      "loss": 0.1275,
+      "step": 1194
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.236000000000001e-06,
+      "loss": 0.0885,
+      "step": 1195
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.232e-06,
+      "loss": 0.1063,
+      "step": 1196
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.228000000000001e-06,
+      "loss": 0.116,
+      "step": 1197
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.224000000000001e-06,
+      "loss": 0.2418,
+      "step": 1198
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.22e-06,
+      "loss": 0.1846,
+      "step": 1199
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.216000000000001e-06,
+      "loss": 0.1594,
+      "step": 1200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.212e-06,
+      "loss": 0.1982,
+      "step": 1201
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.208e-06,
+      "loss": 0.14,
+      "step": 1202
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.204000000000001e-06,
+      "loss": 0.1341,
+      "step": 1203
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.1514,
+      "step": 1204
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.196000000000001e-06,
+      "loss": 0.0789,
+      "step": 1205
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.192e-06,
+      "loss": 0.1198,
+      "step": 1206
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.1880000000000005e-06,
+      "loss": 0.2377,
+      "step": 1207
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.184000000000001e-06,
+      "loss": 0.1118,
+      "step": 1208
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.139,
+      "step": 1209
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.176e-06,
+      "loss": 0.0938,
+      "step": 1210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.172e-06,
+      "loss": 0.0807,
+      "step": 1211
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.168000000000001e-06,
+      "loss": 0.3037,
+      "step": 1212
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.164000000000001e-06,
+      "loss": 0.149,
+      "step": 1213
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.16e-06,
+      "loss": 0.1308,
+      "step": 1214
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.156000000000001e-06,
+      "loss": 0.1185,
+      "step": 1215
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.152e-06,
+      "loss": 0.1322,
+      "step": 1216
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.148000000000001e-06,
+      "loss": 0.1494,
+      "step": 1217
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.1440000000000005e-06,
+      "loss": 0.0688,
+      "step": 1218
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.14e-06,
+      "loss": 0.1562,
+      "step": 1219
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.136000000000001e-06,
+      "loss": 0.0991,
+      "step": 1220
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.132e-06,
+      "loss": 0.1705,
+      "step": 1221
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.128e-06,
+      "loss": 0.1129,
+      "step": 1222
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.124000000000001e-06,
+      "loss": 0.1288,
+      "step": 1223
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 0.1139,
+      "step": 1224
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.116000000000001e-06,
+      "loss": 0.1148,
+      "step": 1225
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.1120000000000015e-06,
+      "loss": 0.1425,
+      "step": 1226
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.108e-06,
+      "loss": 0.1364,
+      "step": 1227
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.104000000000001e-06,
+      "loss": 0.1083,
+      "step": 1228
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 0.1633,
+      "step": 1229
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.096e-06,
+      "loss": 0.1459,
+      "step": 1230
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.092000000000001e-06,
+      "loss": 0.1476,
+      "step": 1231
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.088000000000001e-06,
+      "loss": 0.0686,
+      "step": 1232
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.084000000000001e-06,
+      "loss": 0.1772,
+      "step": 1233
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.08e-06,
+      "loss": 0.1739,
+      "step": 1234
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.0760000000000005e-06,
+      "loss": 0.1275,
+      "step": 1235
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.072000000000001e-06,
+      "loss": 0.1096,
+      "step": 1236
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.068000000000001e-06,
+      "loss": 0.0881,
+      "step": 1237
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.0640000000000005e-06,
+      "loss": 0.202,
+      "step": 1238
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.06e-06,
+      "loss": 0.1831,
+      "step": 1239
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.056000000000001e-06,
+      "loss": 0.1006,
+      "step": 1240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.052000000000001e-06,
+      "loss": 0.081,
+      "step": 1241
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.048e-06,
+      "loss": 0.1935,
+      "step": 1242
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.044000000000001e-06,
+      "loss": 0.1319,
+      "step": 1243
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.04e-06,
+      "loss": 0.1329,
+      "step": 1244
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.036000000000001e-06,
+      "loss": 0.1568,
+      "step": 1245
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.0320000000000015e-06,
+      "loss": 0.0902,
+      "step": 1246
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.028e-06,
+      "loss": 0.0938,
+      "step": 1247
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.024000000000001e-06,
+      "loss": 0.0881,
+      "step": 1248
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 0.1615,
+      "step": 1249
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.016e-06,
+      "loss": 0.0701,
+      "step": 1250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.012000000000001e-06,
+      "loss": 0.0949,
+      "step": 1251
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.0080000000000005e-06,
+      "loss": 0.1247,
+      "step": 1252
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.004000000000001e-06,
+      "loss": 0.1614,
+      "step": 1253
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7e-06,
+      "loss": 0.0757,
+      "step": 1254
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.9960000000000004e-06,
+      "loss": 0.1546,
+      "step": 1255
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.992000000000001e-06,
+      "loss": 0.1502,
+      "step": 1256
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.988000000000001e-06,
+      "loss": 0.1096,
+      "step": 1257
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.984e-06,
+      "loss": 0.1056,
+      "step": 1258
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.98e-06,
+      "loss": 0.1642,
+      "step": 1259
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.976000000000001e-06,
+      "loss": 0.0905,
+      "step": 1260
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.972000000000001e-06,
+      "loss": 0.1385,
+      "step": 1261
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.968e-06,
+      "loss": 0.0926,
+      "step": 1262
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.964000000000001e-06,
+      "loss": 0.16,
+      "step": 1263
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.96e-06,
+      "loss": 0.1184,
+      "step": 1264
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.956000000000001e-06,
+      "loss": 0.1517,
+      "step": 1265
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.952000000000001e-06,
+      "loss": 0.1627,
+      "step": 1266
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.948e-06,
+      "loss": 0.1023,
+      "step": 1267
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.944000000000001e-06,
+      "loss": 0.1741,
+      "step": 1268
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 0.1954,
+      "step": 1269
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.936e-06,
+      "loss": 0.1287,
+      "step": 1270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.932000000000001e-06,
+      "loss": 0.225,
+      "step": 1271
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.928e-06,
+      "loss": 0.1244,
+      "step": 1272
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.924000000000001e-06,
+      "loss": 0.104,
+      "step": 1273
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.92e-06,
+      "loss": 0.2303,
+      "step": 1274
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.916e-06,
+      "loss": 0.1457,
+      "step": 1275
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.912000000000001e-06,
+      "loss": 0.0982,
+      "step": 1276
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.908000000000001e-06,
+      "loss": 0.137,
+      "step": 1277
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.904e-06,
+      "loss": 0.1318,
+      "step": 1278
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.9e-06,
+      "loss": 0.1331,
+      "step": 1279
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.8960000000000006e-06,
+      "loss": 0.1299,
+      "step": 1280
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.892000000000001e-06,
+      "loss": 0.0673,
+      "step": 1281
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.888e-06,
+      "loss": 0.1524,
+      "step": 1282
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.8840000000000005e-06,
+      "loss": 0.0927,
+      "step": 1283
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.88e-06,
+      "loss": 0.1806,
+      "step": 1284
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.876000000000001e-06,
+      "loss": 0.0937,
+      "step": 1285
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.872000000000001e-06,
+      "loss": 0.0678,
+      "step": 1286
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.868e-06,
+      "loss": 0.1426,
+      "step": 1287
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.864000000000001e-06,
+      "loss": 0.0926,
+      "step": 1288
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 0.0955,
+      "step": 1289
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.856e-06,
+      "loss": 0.1981,
+      "step": 1290
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.852000000000001e-06,
+      "loss": 0.1996,
+      "step": 1291
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.848e-06,
+      "loss": 0.1723,
+      "step": 1292
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.844000000000001e-06,
+      "loss": 0.1623,
+      "step": 1293
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 0.1296,
+      "step": 1294
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.836e-06,
+      "loss": 0.1143,
+      "step": 1295
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.832000000000001e-06,
+      "loss": 0.1408,
+      "step": 1296
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.8280000000000005e-06,
+      "loss": 0.1067,
+      "step": 1297
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.824e-06,
+      "loss": 0.1876,
+      "step": 1298
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 0.0506,
+      "step": 1299
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.8160000000000005e-06,
+      "loss": 0.1069,
+      "step": 1300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.812000000000001e-06,
+      "loss": 0.1829,
+      "step": 1301
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.808e-06,
+      "loss": 0.0934,
+      "step": 1302
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.804e-06,
+      "loss": 0.1622,
+      "step": 1303
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.1105,
+      "step": 1304
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.796000000000001e-06,
+      "loss": 0.1891,
+      "step": 1305
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.792000000000001e-06,
+      "loss": 0.1049,
+      "step": 1306
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.788e-06,
+      "loss": 0.2565,
+      "step": 1307
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.784000000000001e-06,
+      "loss": 0.1723,
+      "step": 1308
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.1134,
+      "step": 1309
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.776e-06,
+      "loss": 0.152,
+      "step": 1310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.7720000000000006e-06,
+      "loss": 0.1277,
+      "step": 1311
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.768e-06,
+      "loss": 0.1215,
+      "step": 1312
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.764000000000001e-06,
+      "loss": 0.1573,
+      "step": 1313
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 0.1929,
+      "step": 1314
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.756e-06,
+      "loss": 0.0997,
+      "step": 1315
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.752000000000001e-06,
+      "loss": 0.1531,
+      "step": 1316
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.7480000000000004e-06,
+      "loss": 0.0891,
+      "step": 1317
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.744e-06,
+      "loss": 0.2842,
+      "step": 1318
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 0.096,
+      "step": 1319
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.736e-06,
+      "loss": 0.1407,
+      "step": 1320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.732000000000001e-06,
+      "loss": 0.0961,
+      "step": 1321
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.728e-06,
+      "loss": 0.1034,
+      "step": 1322
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.724e-06,
+      "loss": 0.1381,
+      "step": 1323
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 0.107,
+      "step": 1324
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.716000000000001e-06,
+      "loss": 0.1299,
+      "step": 1325
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.712000000000001e-06,
+      "loss": 0.1302,
+      "step": 1326
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.708e-06,
+      "loss": 0.1361,
+      "step": 1327
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.7040000000000005e-06,
+      "loss": 0.12,
+      "step": 1328
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 0.0981,
+      "step": 1329
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.696e-06,
+      "loss": 0.0893,
+      "step": 1330
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.6920000000000005e-06,
+      "loss": 0.1157,
+      "step": 1331
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.688e-06,
+      "loss": 0.1063,
+      "step": 1332
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.684000000000001e-06,
+      "loss": 0.0849,
+      "step": 1333
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 0.1762,
+      "step": 1334
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.676e-06,
+      "loss": 0.0857,
+      "step": 1335
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.672000000000001e-06,
+      "loss": 0.0969,
+      "step": 1336
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.668e-06,
+      "loss": 0.2307,
+      "step": 1337
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.664e-06,
+      "loss": 0.1894,
+      "step": 1338
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 0.1214,
+      "step": 1339
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.656e-06,
+      "loss": 0.1522,
+      "step": 1340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.652000000000001e-06,
+      "loss": 0.0543,
+      "step": 1341
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.648e-06,
+      "loss": 0.1113,
+      "step": 1342
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.644e-06,
+      "loss": 0.1674,
+      "step": 1343
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 0.0937,
+      "step": 1344
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.6360000000000005e-06,
+      "loss": 0.13,
+      "step": 1345
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.632000000000001e-06,
+      "loss": 0.0407,
+      "step": 1346
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.628e-06,
+      "loss": 0.1811,
+      "step": 1347
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.6240000000000004e-06,
+      "loss": 0.1009,
+      "step": 1348
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 0.1035,
+      "step": 1349
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.616e-06,
+      "loss": 0.1594,
+      "step": 1350
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.612e-06,
+      "loss": 0.1612,
+      "step": 1351
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.608000000000001e-06,
+      "loss": 0.1416,
+      "step": 1352
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.604000000000001e-06,
+      "loss": 0.1259,
+      "step": 1353
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 0.1868,
+      "step": 1354
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.596e-06,
+      "loss": 0.2227,
+      "step": 1355
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.592000000000001e-06,
+      "loss": 0.1269,
+      "step": 1356
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.588000000000001e-06,
+      "loss": 0.138,
+      "step": 1357
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.584e-06,
+      "loss": 0.0905,
+      "step": 1358
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.1869,
+      "step": 1359
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.576e-06,
+      "loss": 0.112,
+      "step": 1360
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.572000000000001e-06,
+      "loss": 0.1309,
+      "step": 1361
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.568000000000001e-06,
+      "loss": 0.2712,
+      "step": 1362
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.564e-06,
+      "loss": 0.0768,
+      "step": 1363
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 0.1235,
+      "step": 1364
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.556e-06,
+      "loss": 0.1258,
+      "step": 1365
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.552000000000001e-06,
+      "loss": 0.2008,
+      "step": 1366
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.548000000000001e-06,
+      "loss": 0.1499,
+      "step": 1367
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.544e-06,
+      "loss": 0.0955,
+      "step": 1368
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 0.145,
+      "step": 1369
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.536e-06,
+      "loss": 0.1677,
+      "step": 1370
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.532e-06,
+      "loss": 0.1137,
+      "step": 1371
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.528000000000001e-06,
+      "loss": 0.0875,
+      "step": 1372
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.5240000000000006e-06,
+      "loss": 0.2921,
+      "step": 1373
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 0.1205,
+      "step": 1374
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.516e-06,
+      "loss": 0.1692,
+      "step": 1375
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.5120000000000005e-06,
+      "loss": 0.0939,
+      "step": 1376
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.508000000000001e-06,
+      "loss": 0.1223,
+      "step": 1377
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.504e-06,
+      "loss": 0.2177,
+      "step": 1378
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.118,
+      "step": 1379
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.496e-06,
+      "loss": 0.1236,
+      "step": 1380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.492000000000001e-06,
+      "loss": 0.1218,
+      "step": 1381
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.488000000000001e-06,
+      "loss": 0.12,
+      "step": 1382
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.484e-06,
+      "loss": 0.1333,
+      "step": 1383
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 0.1601,
+      "step": 1384
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.476e-06,
+      "loss": 0.0556,
+      "step": 1385
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.472000000000001e-06,
+      "loss": 0.2154,
+      "step": 1386
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.468000000000001e-06,
+      "loss": 0.1269,
+      "step": 1387
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.464e-06,
+      "loss": 0.0904,
+      "step": 1388
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 0.1167,
+      "step": 1389
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.456e-06,
+      "loss": 0.1916,
+      "step": 1390
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.452e-06,
+      "loss": 0.1542,
+      "step": 1391
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.448000000000001e-06,
+      "loss": 0.1178,
+      "step": 1392
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.4440000000000005e-06,
+      "loss": 0.1396,
+      "step": 1393
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 0.1292,
+      "step": 1394
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.436e-06,
+      "loss": 0.163,
+      "step": 1395
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.432e-06,
+      "loss": 0.1384,
+      "step": 1396
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.428000000000001e-06,
+      "loss": 0.1768,
+      "step": 1397
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.424e-06,
+      "loss": 0.1507,
+      "step": 1398
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.42e-06,
+      "loss": 0.1046,
+      "step": 1399
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.416e-06,
+      "loss": 0.1021,
+      "step": 1400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.412000000000001e-06,
+      "loss": 0.0807,
+      "step": 1401
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.408000000000001e-06,
+      "loss": 0.1041,
+      "step": 1402
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.404e-06,
+      "loss": 0.1161,
+      "step": 1403
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.097,
+      "step": 1404
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.396e-06,
+      "loss": 0.1464,
+      "step": 1405
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.392000000000001e-06,
+      "loss": 0.0986,
+      "step": 1406
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.3880000000000005e-06,
+      "loss": 0.1444,
+      "step": 1407
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.384e-06,
+      "loss": 0.152,
+      "step": 1408
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.1511,
+      "step": 1409
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.376e-06,
+      "loss": 0.1512,
+      "step": 1410
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.372e-06,
+      "loss": 0.0671,
+      "step": 1411
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.368000000000001e-06,
+      "loss": 0.2108,
+      "step": 1412
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.364e-06,
+      "loss": 0.1396,
+      "step": 1413
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 0.1461,
+      "step": 1414
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.356000000000001e-06,
+      "loss": 0.1477,
+      "step": 1415
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.352e-06,
+      "loss": 0.1896,
+      "step": 1416
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.348000000000001e-06,
+      "loss": 0.1572,
+      "step": 1417
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.344e-06,
+      "loss": 0.0866,
+      "step": 1418
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.34e-06,
+      "loss": 0.1038,
+      "step": 1419
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.336000000000001e-06,
+      "loss": 0.0669,
+      "step": 1420
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.3320000000000005e-06,
+      "loss": 0.1053,
+      "step": 1421
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.328000000000001e-06,
+      "loss": 0.1651,
+      "step": 1422
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.324e-06,
+      "loss": 0.1083,
+      "step": 1423
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 0.1339,
+      "step": 1424
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.316000000000001e-06,
+      "loss": 0.0779,
+      "step": 1425
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.312000000000001e-06,
+      "loss": 0.1462,
+      "step": 1426
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.308e-06,
+      "loss": 0.196,
+      "step": 1427
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.304e-06,
+      "loss": 0.1388,
+      "step": 1428
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 0.1753,
+      "step": 1429
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.296000000000001e-06,
+      "loss": 0.1241,
+      "step": 1430
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.292e-06,
+      "loss": 0.0755,
+      "step": 1431
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.288000000000001e-06,
+      "loss": 0.0598,
+      "step": 1432
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.284e-06,
+      "loss": 0.2034,
+      "step": 1433
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 0.1201,
+      "step": 1434
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.2760000000000006e-06,
+      "loss": 0.1391,
+      "step": 1435
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.272e-06,
+      "loss": 0.0754,
+      "step": 1436
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.268000000000001e-06,
+      "loss": 0.1178,
+      "step": 1437
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.264e-06,
+      "loss": 0.1153,
+      "step": 1438
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.26e-06,
+      "loss": 0.1733,
+      "step": 1439
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.256000000000001e-06,
+      "loss": 0.0884,
+      "step": 1440
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.2520000000000004e-06,
+      "loss": 0.1147,
+      "step": 1441
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.248000000000001e-06,
+      "loss": 0.111,
+      "step": 1442
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.244e-06,
+      "loss": 0.1433,
+      "step": 1443
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.24e-06,
+      "loss": 0.0551,
+      "step": 1444
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.236000000000001e-06,
+      "loss": 0.1728,
+      "step": 1445
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.232000000000001e-06,
+      "loss": 0.1556,
+      "step": 1446
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.228e-06,
+      "loss": 0.0895,
+      "step": 1447
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.224e-06,
+      "loss": 0.0885,
+      "step": 1448
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 0.0989,
+      "step": 1449
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.216000000000001e-06,
+      "loss": 0.0855,
+      "step": 1450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.212e-06,
+      "loss": 0.1847,
+      "step": 1451
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.2080000000000005e-06,
+      "loss": 0.083,
+      "step": 1452
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.204e-06,
+      "loss": 0.1055,
+      "step": 1453
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 0.1229,
+      "step": 1454
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.196000000000001e-06,
+      "loss": 0.1226,
+      "step": 1455
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.192e-06,
+      "loss": 0.1191,
+      "step": 1456
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.188000000000001e-06,
+      "loss": 0.1877,
+      "step": 1457
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.184e-06,
+      "loss": 0.2268,
+      "step": 1458
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.18e-06,
+      "loss": 0.1467,
+      "step": 1459
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.176000000000001e-06,
+      "loss": 0.1681,
+      "step": 1460
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.172e-06,
+      "loss": 0.1573,
+      "step": 1461
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.168000000000001e-06,
+      "loss": 0.0662,
+      "step": 1462
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.164e-06,
+      "loss": 0.0684,
+      "step": 1463
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.16e-06,
+      "loss": 0.182,
+      "step": 1464
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.156000000000001e-06,
+      "loss": 0.2206,
+      "step": 1465
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.1520000000000006e-06,
+      "loss": 0.1174,
+      "step": 1466
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.148e-06,
+      "loss": 0.1211,
+      "step": 1467
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.144e-06,
+      "loss": 0.1081,
+      "step": 1468
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 0.1357,
+      "step": 1469
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.136000000000001e-06,
+      "loss": 0.1091,
+      "step": 1470
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.132e-06,
+      "loss": 0.1508,
+      "step": 1471
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.1280000000000005e-06,
+      "loss": 0.1975,
+      "step": 1472
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.124000000000001e-06,
+      "loss": 0.158,
+      "step": 1473
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 0.2262,
+      "step": 1474
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.116000000000001e-06,
+      "loss": 0.0771,
+      "step": 1475
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.112e-06,
+      "loss": 0.0931,
+      "step": 1476
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.108000000000001e-06,
+      "loss": 0.0951,
+      "step": 1477
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.104000000000001e-06,
+      "loss": 0.0854,
+      "step": 1478
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1e-06,
+      "loss": 0.1283,
+      "step": 1479
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.096000000000001e-06,
+      "loss": 0.1924,
+      "step": 1480
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.092e-06,
+      "loss": 0.0917,
+      "step": 1481
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.088000000000001e-06,
+      "loss": 0.2734,
+      "step": 1482
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.084000000000001e-06,
+      "loss": 0.1369,
+      "step": 1483
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.08e-06,
+      "loss": 0.096,
+      "step": 1484
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.076000000000001e-06,
+      "loss": 0.0868,
+      "step": 1485
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0720000000000005e-06,
+      "loss": 0.0907,
+      "step": 1486
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.068e-06,
+      "loss": 0.0698,
+      "step": 1487
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.064000000000001e-06,
+      "loss": 0.1394,
+      "step": 1488
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 0.1954,
+      "step": 1489
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.056000000000001e-06,
+      "loss": 0.1277,
+      "step": 1490
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.052e-06,
+      "loss": 0.1432,
+      "step": 1491
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.048e-06,
+      "loss": 0.1162,
+      "step": 1492
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.044000000000001e-06,
+      "loss": 0.1255,
+      "step": 1493
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 0.157,
+      "step": 1494
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.036000000000001e-06,
+      "loss": 0.1053,
+      "step": 1495
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.032e-06,
+      "loss": 0.2011,
+      "step": 1496
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.0280000000000006e-06,
+      "loss": 0.1132,
+      "step": 1497
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.024000000000001e-06,
+      "loss": 0.1575,
+      "step": 1498
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.02e-06,
+      "loss": 0.1031,
+      "step": 1499
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.0160000000000005e-06,
+      "loss": 0.1014,
+      "step": 1500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.012e-06,
+      "loss": 0.1209,
+      "step": 1501
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.008000000000001e-06,
+      "loss": 0.2586,
+      "step": 1502
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.004000000000001e-06,
+      "loss": 0.0485,
+      "step": 1503
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6e-06,
+      "loss": 0.1978,
+      "step": 1504
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.996000000000001e-06,
+      "loss": 0.0848,
+      "step": 1505
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.992e-06,
+      "loss": 0.1936,
+      "step": 1506
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.988e-06,
+      "loss": 0.1474,
+      "step": 1507
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.984000000000001e-06,
+      "loss": 0.0979,
+      "step": 1508
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.98e-06,
+      "loss": 0.1236,
+      "step": 1509
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.976000000000001e-06,
+      "loss": 0.1175,
+      "step": 1510
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.972e-06,
+      "loss": 0.1455,
+      "step": 1511
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.968e-06,
+      "loss": 0.1253,
+      "step": 1512
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.964000000000001e-06,
+      "loss": 0.1799,
+      "step": 1513
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 0.1005,
+      "step": 1514
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.956000000000001e-06,
+      "loss": 0.1355,
+      "step": 1515
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.952e-06,
+      "loss": 0.0855,
+      "step": 1516
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.9480000000000005e-06,
+      "loss": 0.1067,
+      "step": 1517
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.944000000000001e-06,
+      "loss": 0.2548,
+      "step": 1518
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.94e-06,
+      "loss": 0.1913,
+      "step": 1519
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.9360000000000004e-06,
+      "loss": 0.1248,
+      "step": 1520
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.932e-06,
+      "loss": 0.1474,
+      "step": 1521
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.928000000000001e-06,
+      "loss": 0.1659,
+      "step": 1522
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.924000000000001e-06,
+      "loss": 0.1134,
+      "step": 1523
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.92e-06,
+      "loss": 0.1372,
+      "step": 1524
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.916000000000001e-06,
+      "loss": 0.0877,
+      "step": 1525
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.912e-06,
+      "loss": 0.1675,
+      "step": 1526
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.908e-06,
+      "loss": 0.095,
+      "step": 1527
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.9040000000000006e-06,
+      "loss": 0.112,
+      "step": 1528
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.9e-06,
+      "loss": 0.2132,
+      "step": 1529
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.896000000000001e-06,
+      "loss": 0.0985,
+      "step": 1530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.892e-06,
+      "loss": 0.104,
+      "step": 1531
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.888e-06,
+      "loss": 0.124,
+      "step": 1532
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.884000000000001e-06,
+      "loss": 0.1052,
+      "step": 1533
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 0.0808,
+      "step": 1534
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.876000000000001e-06,
+      "loss": 0.0612,
+      "step": 1535
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.872000000000001e-06,
+      "loss": 0.1368,
+      "step": 1536
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.868e-06,
+      "loss": 0.1951,
+      "step": 1537
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.864000000000001e-06,
+      "loss": 0.1834,
+      "step": 1538
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.86e-06,
+      "loss": 0.1137,
+      "step": 1539
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.856e-06,
+      "loss": 0.1359,
+      "step": 1540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.852000000000001e-06,
+      "loss": 0.0738,
+      "step": 1541
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.848000000000001e-06,
+      "loss": 0.0863,
+      "step": 1542
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.844000000000001e-06,
+      "loss": 0.1237,
+      "step": 1543
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.84e-06,
+      "loss": 0.1196,
+      "step": 1544
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.8360000000000005e-06,
+      "loss": 0.1212,
+      "step": 1545
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.832000000000001e-06,
+      "loss": 0.0854,
+      "step": 1546
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.828e-06,
+      "loss": 0.0451,
+      "step": 1547
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.8240000000000005e-06,
+      "loss": 0.0407,
+      "step": 1548
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.82e-06,
+      "loss": 0.0478,
+      "step": 1549
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.816000000000001e-06,
+      "loss": 0.0583,
+      "step": 1550
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.812000000000001e-06,
+      "loss": 0.0644,
+      "step": 1551
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.808e-06,
+      "loss": 0.043,
+      "step": 1552
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.804000000000001e-06,
+      "loss": 0.0381,
+      "step": 1553
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.8e-06,
+      "loss": 0.0637,
+      "step": 1554
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.796000000000001e-06,
+      "loss": 0.0351,
+      "step": 1555
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.792000000000001e-06,
+      "loss": 0.0443,
+      "step": 1556
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.788e-06,
+      "loss": 0.0519,
+      "step": 1557
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.784000000000001e-06,
+      "loss": 0.0409,
+      "step": 1558
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.78e-06,
+      "loss": 0.0446,
+      "step": 1559
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.776e-06,
+      "loss": 0.0635,
+      "step": 1560
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.772000000000001e-06,
+      "loss": 0.0851,
+      "step": 1561
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.7680000000000005e-06,
+      "loss": 0.0295,
+      "step": 1562
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.764000000000001e-06,
+      "loss": 0.045,
+      "step": 1563
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.76e-06,
+      "loss": 0.0478,
+      "step": 1564
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.7560000000000005e-06,
+      "loss": 0.0452,
+      "step": 1565
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.752000000000001e-06,
+      "loss": 0.0491,
+      "step": 1566
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.748e-06,
+      "loss": 0.0406,
+      "step": 1567
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.744e-06,
+      "loss": 0.0554,
+      "step": 1568
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.74e-06,
+      "loss": 0.0728,
+      "step": 1569
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.736000000000001e-06,
+      "loss": 0.0496,
+      "step": 1570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.732000000000001e-06,
+      "loss": 0.0553,
+      "step": 1571
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.728e-06,
+      "loss": 0.0351,
+      "step": 1572
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.724000000000001e-06,
+      "loss": 0.0599,
+      "step": 1573
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.72e-06,
+      "loss": 0.0285,
+      "step": 1574
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.716000000000001e-06,
+      "loss": 0.0545,
+      "step": 1575
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.7120000000000005e-06,
+      "loss": 0.0295,
+      "step": 1576
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.708e-06,
+      "loss": 0.0687,
+      "step": 1577
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.704000000000001e-06,
+      "loss": 0.0251,
+      "step": 1578
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.7e-06,
+      "loss": 0.0491,
+      "step": 1579
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.696e-06,
+      "loss": 0.0679,
+      "step": 1580
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.692000000000001e-06,
+      "loss": 0.0549,
+      "step": 1581
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.6880000000000004e-06,
+      "loss": 0.0467,
+      "step": 1582
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.684000000000001e-06,
+      "loss": 0.0475,
+      "step": 1583
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.68e-06,
+      "loss": 0.0925,
+      "step": 1584
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.676e-06,
+      "loss": 0.0377,
+      "step": 1585
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.672000000000001e-06,
+      "loss": 0.0837,
+      "step": 1586
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.668e-06,
+      "loss": 0.0406,
+      "step": 1587
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.664e-06,
+      "loss": 0.0418,
+      "step": 1588
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.66e-06,
+      "loss": 0.0525,
+      "step": 1589
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.6560000000000006e-06,
+      "loss": 0.0534,
+      "step": 1590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.652000000000001e-06,
+      "loss": 0.0458,
+      "step": 1591
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.648e-06,
+      "loss": 0.0454,
+      "step": 1592
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.6440000000000005e-06,
+      "loss": 0.0544,
+      "step": 1593
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.64e-06,
+      "loss": 0.0594,
+      "step": 1594
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.636000000000001e-06,
+      "loss": 0.0281,
+      "step": 1595
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.6320000000000005e-06,
+      "loss": 0.049,
+      "step": 1596
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.628e-06,
+      "loss": 0.0424,
+      "step": 1597
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.624000000000001e-06,
+      "loss": 0.0703,
+      "step": 1598
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 0.0654,
+      "step": 1599
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.616e-06,
+      "loss": 0.0201,
+      "step": 1600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.612000000000001e-06,
+      "loss": 0.075,
+      "step": 1601
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.608e-06,
+      "loss": 0.0399,
+      "step": 1602
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.604000000000001e-06,
+      "loss": 0.0645,
+      "step": 1603
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.0551,
+      "step": 1604
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.596e-06,
+      "loss": 0.0375,
+      "step": 1605
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.592000000000001e-06,
+      "loss": 0.0941,
+      "step": 1606
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.588e-06,
+      "loss": 0.0515,
+      "step": 1607
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.584e-06,
+      "loss": 0.0777,
+      "step": 1608
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.0711,
+      "step": 1609
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.5760000000000005e-06,
+      "loss": 0.0678,
+      "step": 1610
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.572000000000001e-06,
+      "loss": 0.0909,
+      "step": 1611
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.568e-06,
+      "loss": 0.0306,
+      "step": 1612
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.5640000000000004e-06,
+      "loss": 0.0792,
+      "step": 1613
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 0.0309,
+      "step": 1614
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.556000000000001e-06,
+      "loss": 0.0393,
+      "step": 1615
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.552e-06,
+      "loss": 0.0357,
+      "step": 1616
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.548e-06,
+      "loss": 0.065,
+      "step": 1617
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.544000000000001e-06,
+      "loss": 0.0665,
+      "step": 1618
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 0.0528,
+      "step": 1619
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.536e-06,
+      "loss": 0.0304,
+      "step": 1620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.5320000000000006e-06,
+      "loss": 0.0623,
+      "step": 1621
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.528e-06,
+      "loss": 0.062,
+      "step": 1622
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.524000000000001e-06,
+      "loss": 0.0269,
+      "step": 1623
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 0.0488,
+      "step": 1624
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.516e-06,
+      "loss": 0.0503,
+      "step": 1625
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.512000000000001e-06,
+      "loss": 0.0838,
+      "step": 1626
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.508e-06,
+      "loss": 0.0666,
+      "step": 1627
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.504e-06,
+      "loss": 0.0271,
+      "step": 1628
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.0339,
+      "step": 1629
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.496e-06,
+      "loss": 0.0494,
+      "step": 1630
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.492000000000001e-06,
+      "loss": 0.0457,
+      "step": 1631
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.488e-06,
+      "loss": 0.0684,
+      "step": 1632
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.484e-06,
+      "loss": 0.0807,
+      "step": 1633
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 0.0401,
+      "step": 1634
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.476000000000001e-06,
+      "loss": 0.1044,
+      "step": 1635
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.472e-06,
+      "loss": 0.0615,
+      "step": 1636
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.468e-06,
+      "loss": 0.0704,
+      "step": 1637
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.4640000000000005e-06,
+      "loss": 0.038,
+      "step": 1638
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 0.0445,
+      "step": 1639
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.456e-06,
+      "loss": 0.0659,
+      "step": 1640
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.4520000000000005e-06,
+      "loss": 0.0675,
+      "step": 1641
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.448e-06,
+      "loss": 0.0288,
+      "step": 1642
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.444000000000001e-06,
+      "loss": 0.0335,
+      "step": 1643
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 0.0446,
+      "step": 1644
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.436e-06,
+      "loss": 0.0272,
+      "step": 1645
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.432000000000001e-06,
+      "loss": 0.0527,
+      "step": 1646
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.4279999999999995e-06,
+      "loss": 0.116,
+      "step": 1647
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.424e-06,
+      "loss": 0.0358,
+      "step": 1648
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 0.0572,
+      "step": 1649
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.416e-06,
+      "loss": 0.0489,
+      "step": 1650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.412000000000001e-06,
+      "loss": 0.0662,
+      "step": 1651
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.408e-06,
+      "loss": 0.0438,
+      "step": 1652
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.404e-06,
+      "loss": 0.1023,
+      "step": 1653
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 0.0239,
+      "step": 1654
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.3960000000000005e-06,
+      "loss": 0.0579,
+      "step": 1655
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.392e-06,
+      "loss": 0.0905,
+      "step": 1656
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.388e-06,
+      "loss": 0.0741,
+      "step": 1657
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.3840000000000005e-06,
+      "loss": 0.0689,
+      "step": 1658
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.0865,
+      "step": 1659
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.376e-06,
+      "loss": 0.036,
+      "step": 1660
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.372e-06,
+      "loss": 0.0705,
+      "step": 1661
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.368000000000001e-06,
+      "loss": 0.0903,
+      "step": 1662
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.364000000000001e-06,
+      "loss": 0.036,
+      "step": 1663
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.36e-06,
+      "loss": 0.0469,
+      "step": 1664
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.356e-06,
+      "loss": 0.0444,
+      "step": 1665
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.352000000000001e-06,
+      "loss": 0.0411,
+      "step": 1666
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.348000000000001e-06,
+      "loss": 0.0411,
+      "step": 1667
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.344e-06,
+      "loss": 0.0786,
+      "step": 1668
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 0.0318,
+      "step": 1669
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.336e-06,
+      "loss": 0.0315,
+      "step": 1670
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.332000000000001e-06,
+      "loss": 0.0568,
+      "step": 1671
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.328000000000001e-06,
+      "loss": 0.0605,
+      "step": 1672
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.324e-06,
+      "loss": 0.0693,
+      "step": 1673
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 0.0464,
+      "step": 1674
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.3160000000000004e-06,
+      "loss": 0.0552,
+      "step": 1675
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.312e-06,
+      "loss": 0.0941,
+      "step": 1676
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.308000000000001e-06,
+      "loss": 0.0446,
+      "step": 1677
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.304e-06,
+      "loss": 0.0472,
+      "step": 1678
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 0.0633,
+      "step": 1679
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.296e-06,
+      "loss": 0.0506,
+      "step": 1680
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.292e-06,
+      "loss": 0.0299,
+      "step": 1681
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.288000000000001e-06,
+      "loss": 0.0762,
+      "step": 1682
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.2840000000000006e-06,
+      "loss": 0.0297,
+      "step": 1683
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.28e-06,
+      "loss": 0.0357,
+      "step": 1684
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.276e-06,
+      "loss": 0.0342,
+      "step": 1685
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.2720000000000005e-06,
+      "loss": 0.0603,
+      "step": 1686
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.268000000000001e-06,
+      "loss": 0.0719,
+      "step": 1687
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.264e-06,
+      "loss": 0.0392,
+      "step": 1688
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 0.0491,
+      "step": 1689
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.256e-06,
+      "loss": 0.0422,
+      "step": 1690
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.252000000000001e-06,
+      "loss": 0.0877,
+      "step": 1691
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.248000000000001e-06,
+      "loss": 0.0259,
+      "step": 1692
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.244e-06,
+      "loss": 0.09,
+      "step": 1693
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 0.0744,
+      "step": 1694
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.236e-06,
+      "loss": 0.0674,
+      "step": 1695
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.232e-06,
+      "loss": 0.0291,
+      "step": 1696
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.228000000000001e-06,
+      "loss": 0.0626,
+      "step": 1697
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.224e-06,
+      "loss": 0.0511,
+      "step": 1698
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 0.0554,
+      "step": 1699
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.216e-06,
+      "loss": 0.1238,
+      "step": 1700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.212e-06,
+      "loss": 0.0702,
+      "step": 1701
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.208000000000001e-06,
+      "loss": 0.068,
+      "step": 1702
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.2040000000000005e-06,
+      "loss": 0.052,
+      "step": 1703
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.2e-06,
+      "loss": 0.0359,
+      "step": 1704
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.196e-06,
+      "loss": 0.057,
+      "step": 1705
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.1920000000000004e-06,
+      "loss": 0.0515,
+      "step": 1706
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.188000000000001e-06,
+      "loss": 0.0471,
+      "step": 1707
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.184e-06,
+      "loss": 0.0502,
+      "step": 1708
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.18e-06,
+      "loss": 0.0275,
+      "step": 1709
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.176e-06,
+      "loss": 0.047,
+      "step": 1710
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.172000000000001e-06,
+      "loss": 0.0534,
+      "step": 1711
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.168000000000001e-06,
+      "loss": 0.0505,
+      "step": 1712
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.164e-06,
+      "loss": 0.0407,
+      "step": 1713
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 0.0368,
+      "step": 1714
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.156e-06,
+      "loss": 0.0811,
+      "step": 1715
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.152e-06,
+      "loss": 0.065,
+      "step": 1716
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1480000000000005e-06,
+      "loss": 0.0493,
+      "step": 1717
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.144e-06,
+      "loss": 0.0632,
+      "step": 1718
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 0.0352,
+      "step": 1719
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.136e-06,
+      "loss": 0.0463,
+      "step": 1720
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.132e-06,
+      "loss": 0.0594,
+      "step": 1721
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.128000000000001e-06,
+      "loss": 0.0393,
+      "step": 1722
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.124e-06,
+      "loss": 0.0585,
+      "step": 1723
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.12e-06,
+      "loss": 0.0385,
+      "step": 1724
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.116000000000001e-06,
+      "loss": 0.0385,
+      "step": 1725
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.112e-06,
+      "loss": 0.049,
+      "step": 1726
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.108000000000001e-06,
+      "loss": 0.0304,
+      "step": 1727
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.104e-06,
+      "loss": 0.0289,
+      "step": 1728
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.1e-06,
+      "loss": 0.0519,
+      "step": 1729
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.096000000000001e-06,
+      "loss": 0.0402,
+      "step": 1730
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0920000000000005e-06,
+      "loss": 0.0388,
+      "step": 1731
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.088000000000001e-06,
+      "loss": 0.0286,
+      "step": 1732
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.084e-06,
+      "loss": 0.0791,
+      "step": 1733
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 0.1179,
+      "step": 1734
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.076000000000001e-06,
+      "loss": 0.0357,
+      "step": 1735
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.072e-06,
+      "loss": 0.0295,
+      "step": 1736
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0680000000000004e-06,
+      "loss": 0.0508,
+      "step": 1737
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.064e-06,
+      "loss": 0.0444,
+      "step": 1738
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 0.0524,
+      "step": 1739
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.056000000000001e-06,
+      "loss": 0.0329,
+      "step": 1740
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.052e-06,
+      "loss": 0.0799,
+      "step": 1741
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.048000000000001e-06,
+      "loss": 0.0563,
+      "step": 1742
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.044e-06,
+      "loss": 0.0652,
+      "step": 1743
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.04e-06,
+      "loss": 0.0483,
+      "step": 1744
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.0360000000000006e-06,
+      "loss": 0.0503,
+      "step": 1745
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.032e-06,
+      "loss": 0.0433,
+      "step": 1746
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.028000000000001e-06,
+      "loss": 0.1046,
+      "step": 1747
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.024e-06,
+      "loss": 0.0476,
+      "step": 1748
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.02e-06,
+      "loss": 0.0443,
+      "step": 1749
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.016000000000001e-06,
+      "loss": 0.0526,
+      "step": 1750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.0120000000000005e-06,
+      "loss": 0.048,
+      "step": 1751
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.008000000000001e-06,
+      "loss": 0.0397,
+      "step": 1752
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.004e-06,
+      "loss": 0.0427,
+      "step": 1753
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5e-06,
+      "loss": 0.0715,
+      "step": 1754
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.996e-06,
+      "loss": 0.0317,
+      "step": 1755
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.992e-06,
+      "loss": 0.055,
+      "step": 1756
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.988e-06,
+      "loss": 0.0587,
+      "step": 1757
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.984000000000001e-06,
+      "loss": 0.0367,
+      "step": 1758
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.0494,
+      "step": 1759
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.976e-06,
+      "loss": 0.0637,
+      "step": 1760
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.972e-06,
+      "loss": 0.0619,
+      "step": 1761
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9680000000000005e-06,
+      "loss": 0.0404,
+      "step": 1762
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.964e-06,
+      "loss": 0.0822,
+      "step": 1763
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 0.051,
+      "step": 1764
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9560000000000005e-06,
+      "loss": 0.0756,
+      "step": 1765
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.952e-06,
+      "loss": 0.0397,
+      "step": 1766
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.948000000000001e-06,
+      "loss": 0.0767,
+      "step": 1767
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.9440000000000004e-06,
+      "loss": 0.0436,
+      "step": 1768
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.94e-06,
+      "loss": 0.0494,
+      "step": 1769
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.936e-06,
+      "loss": 0.0576,
+      "step": 1770
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.932e-06,
+      "loss": 0.06,
+      "step": 1771
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.928000000000001e-06,
+      "loss": 0.0461,
+      "step": 1772
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.924000000000001e-06,
+      "loss": 0.0397,
+      "step": 1773
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.92e-06,
+      "loss": 0.0724,
+      "step": 1774
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.916e-06,
+      "loss": 0.043,
+      "step": 1775
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.9120000000000006e-06,
+      "loss": 0.0247,
+      "step": 1776
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.908e-06,
+      "loss": 0.0363,
+      "step": 1777
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.904000000000001e-06,
+      "loss": 0.0427,
+      "step": 1778
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 0.0536,
+      "step": 1779
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.896e-06,
+      "loss": 0.0786,
+      "step": 1780
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.892000000000001e-06,
+      "loss": 0.0423,
+      "step": 1781
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.8880000000000005e-06,
+      "loss": 0.0296,
+      "step": 1782
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.884e-06,
+      "loss": 0.0699,
+      "step": 1783
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 0.0404,
+      "step": 1784
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.876e-06,
+      "loss": 0.037,
+      "step": 1785
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.872000000000001e-06,
+      "loss": 0.0285,
+      "step": 1786
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.868000000000001e-06,
+      "loss": 0.0568,
+      "step": 1787
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.864e-06,
+      "loss": 0.0902,
+      "step": 1788
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.86e-06,
+      "loss": 0.0368,
+      "step": 1789
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.856e-06,
+      "loss": 0.0694,
+      "step": 1790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.852e-06,
+      "loss": 0.0478,
+      "step": 1791
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.848000000000001e-06,
+      "loss": 0.0259,
+      "step": 1792
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.8440000000000005e-06,
+      "loss": 0.0412,
+      "step": 1793
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.84e-06,
+      "loss": 0.0365,
+      "step": 1794
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.836e-06,
+      "loss": 0.0694,
+      "step": 1795
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.8320000000000005e-06,
+      "loss": 0.0516,
+      "step": 1796
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.828e-06,
+      "loss": 0.0548,
+      "step": 1797
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.824000000000001e-06,
+      "loss": 0.0353,
+      "step": 1798
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 0.0528,
+      "step": 1799
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.816e-06,
+      "loss": 0.0533,
+      "step": 1800
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.812000000000001e-06,
+      "loss": 0.0452,
+      "step": 1801
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.808e-06,
+      "loss": 0.0414,
+      "step": 1802
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.804e-06,
+      "loss": 0.065,
+      "step": 1803
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.0533,
+      "step": 1804
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.796e-06,
+      "loss": 0.0403,
+      "step": 1805
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.792000000000001e-06,
+      "loss": 0.0591,
+      "step": 1806
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.7880000000000006e-06,
+      "loss": 0.0631,
+      "step": 1807
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.784e-06,
+      "loss": 0.0447,
+      "step": 1808
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.78e-06,
+      "loss": 0.0785,
+      "step": 1809
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.7760000000000005e-06,
+      "loss": 0.0411,
+      "step": 1810
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.772e-06,
+      "loss": 0.0522,
+      "step": 1811
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.768000000000001e-06,
+      "loss": 0.0632,
+      "step": 1812
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7640000000000005e-06,
+      "loss": 0.036,
+      "step": 1813
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.76e-06,
+      "loss": 0.0461,
+      "step": 1814
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.756000000000001e-06,
+      "loss": 0.0761,
+      "step": 1815
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.752e-06,
+      "loss": 0.0532,
+      "step": 1816
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.748e-06,
+      "loss": 0.0509,
+      "step": 1817
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.744000000000001e-06,
+      "loss": 0.0574,
+      "step": 1818
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.74e-06,
+      "loss": 0.039,
+      "step": 1819
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.736000000000001e-06,
+      "loss": 0.0754,
+      "step": 1820
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.732000000000001e-06,
+      "loss": 0.0597,
+      "step": 1821
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.728e-06,
+      "loss": 0.0465,
+      "step": 1822
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.724e-06,
+      "loss": 0.0403,
+      "step": 1823
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 0.0869,
+      "step": 1824
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.716e-06,
+      "loss": 0.0209,
+      "step": 1825
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.712000000000001e-06,
+      "loss": 0.0582,
+      "step": 1826
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.7080000000000005e-06,
+      "loss": 0.0805,
+      "step": 1827
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.704e-06,
+      "loss": 0.0466,
+      "step": 1828
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.7e-06,
+      "loss": 0.0401,
+      "step": 1829
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6960000000000004e-06,
+      "loss": 0.0446,
+      "step": 1830
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.692e-06,
+      "loss": 0.0363,
+      "step": 1831
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.688000000000001e-06,
+      "loss": 0.033,
+      "step": 1832
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.684e-06,
+      "loss": 0.0425,
+      "step": 1833
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 0.087,
+      "step": 1834
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.676000000000001e-06,
+      "loss": 0.0475,
+      "step": 1835
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.672e-06,
+      "loss": 0.0775,
+      "step": 1836
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.668e-06,
+      "loss": 0.0738,
+      "step": 1837
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.664000000000001e-06,
+      "loss": 0.0491,
+      "step": 1838
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.66e-06,
+      "loss": 0.0311,
+      "step": 1839
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.656000000000001e-06,
+      "loss": 0.0429,
+      "step": 1840
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6520000000000005e-06,
+      "loss": 0.0496,
+      "step": 1841
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.648e-06,
+      "loss": 0.0275,
+      "step": 1842
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.644e-06,
+      "loss": 0.0477,
+      "step": 1843
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 0.0632,
+      "step": 1844
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.636e-06,
+      "loss": 0.0749,
+      "step": 1845
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.632000000000001e-06,
+      "loss": 0.0317,
+      "step": 1846
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.628e-06,
+      "loss": 0.0503,
+      "step": 1847
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.624e-06,
+      "loss": 0.0403,
+      "step": 1848
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 0.0457,
+      "step": 1849
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.616e-06,
+      "loss": 0.0661,
+      "step": 1850
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.612e-06,
+      "loss": 0.0392,
+      "step": 1851
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.608000000000001e-06,
+      "loss": 0.0343,
+      "step": 1852
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.604e-06,
+      "loss": 0.0333,
+      "step": 1853
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.0734,
+      "step": 1854
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.5960000000000006e-06,
+      "loss": 0.0344,
+      "step": 1855
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.592e-06,
+      "loss": 0.0706,
+      "step": 1856
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.588e-06,
+      "loss": 0.0496,
+      "step": 1857
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.5840000000000005e-06,
+      "loss": 0.059,
+      "step": 1858
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.58e-06,
+      "loss": 0.0448,
+      "step": 1859
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.576000000000001e-06,
+      "loss": 0.048,
+      "step": 1860
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5720000000000004e-06,
+      "loss": 0.0331,
+      "step": 1861
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.568e-06,
+      "loss": 0.0306,
+      "step": 1862
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.564e-06,
+      "loss": 0.0488,
+      "step": 1863
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.56e-06,
+      "loss": 0.0645,
+      "step": 1864
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.556e-06,
+      "loss": 0.0349,
+      "step": 1865
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.552000000000001e-06,
+      "loss": 0.0829,
+      "step": 1866
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.548e-06,
+      "loss": 0.0428,
+      "step": 1867
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.544000000000001e-06,
+      "loss": 0.067,
+      "step": 1868
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 0.0781,
+      "step": 1869
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.536e-06,
+      "loss": 0.0355,
+      "step": 1870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.532e-06,
+      "loss": 0.0597,
+      "step": 1871
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5280000000000005e-06,
+      "loss": 0.0459,
+      "step": 1872
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.524e-06,
+      "loss": 0.034,
+      "step": 1873
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 0.0715,
+      "step": 1874
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5160000000000005e-06,
+      "loss": 0.0971,
+      "step": 1875
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.512e-06,
+      "loss": 0.024,
+      "step": 1876
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.508e-06,
+      "loss": 0.0408,
+      "step": 1877
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.504e-06,
+      "loss": 0.0377,
+      "step": 1878
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5e-06,
+      "loss": 0.0461,
+      "step": 1879
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.496000000000001e-06,
+      "loss": 0.0664,
+      "step": 1880
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.492e-06,
+      "loss": 0.0472,
+      "step": 1881
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.488e-06,
+      "loss": 0.0462,
+      "step": 1882
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.484000000000001e-06,
+      "loss": 0.0576,
+      "step": 1883
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.48e-06,
+      "loss": 0.0653,
+      "step": 1884
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.476e-06,
+      "loss": 0.0302,
+      "step": 1885
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.4720000000000006e-06,
+      "loss": 0.0332,
+      "step": 1886
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.468e-06,
+      "loss": 0.0508,
+      "step": 1887
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.464000000000001e-06,
+      "loss": 0.0488,
+      "step": 1888
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 0.066,
+      "step": 1889
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.456e-06,
+      "loss": 0.0285,
+      "step": 1890
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.452e-06,
+      "loss": 0.043,
+      "step": 1891
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.4480000000000004e-06,
+      "loss": 0.0235,
+      "step": 1892
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.444e-06,
+      "loss": 0.0631,
+      "step": 1893
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 0.0485,
+      "step": 1894
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.436e-06,
+      "loss": 0.024,
+      "step": 1895
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.432e-06,
+      "loss": 0.0486,
+      "step": 1896
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.428000000000001e-06,
+      "loss": 0.065,
+      "step": 1897
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.424e-06,
+      "loss": 0.0608,
+      "step": 1898
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.42e-06,
+      "loss": 0.0857,
+      "step": 1899
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.416000000000001e-06,
+      "loss": 0.0566,
+      "step": 1900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.412e-06,
+      "loss": 0.0402,
+      "step": 1901
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.408000000000001e-06,
+      "loss": 0.0736,
+      "step": 1902
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.4040000000000005e-06,
+      "loss": 0.0428,
+      "step": 1903
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.4e-06,
+      "loss": 0.0693,
+      "step": 1904
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.396e-06,
+      "loss": 0.0343,
+      "step": 1905
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.3920000000000005e-06,
+      "loss": 0.0591,
+      "step": 1906
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.388e-06,
+      "loss": 0.0454,
+      "step": 1907
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.384000000000001e-06,
+      "loss": 0.0293,
+      "step": 1908
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.38e-06,
+      "loss": 0.0569,
+      "step": 1909
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.376e-06,
+      "loss": 0.043,
+      "step": 1910
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.372e-06,
+      "loss": 0.0179,
+      "step": 1911
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.368e-06,
+      "loss": 0.038,
+      "step": 1912
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.364e-06,
+      "loss": 0.0707,
+      "step": 1913
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 0.0392,
+      "step": 1914
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.356e-06,
+      "loss": 0.0227,
+      "step": 1915
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.352e-06,
+      "loss": 0.0808,
+      "step": 1916
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.3480000000000006e-06,
+      "loss": 0.0528,
+      "step": 1917
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.344e-06,
+      "loss": 0.0324,
+      "step": 1918
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.34e-06,
+      "loss": 0.023,
+      "step": 1919
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.3360000000000005e-06,
+      "loss": 0.0706,
+      "step": 1920
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.332e-06,
+      "loss": 0.0276,
+      "step": 1921
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.328000000000001e-06,
+      "loss": 0.0915,
+      "step": 1922
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.3240000000000004e-06,
+      "loss": 0.0776,
+      "step": 1923
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.32e-06,
+      "loss": 0.0538,
+      "step": 1924
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.316e-06,
+      "loss": 0.0522,
+      "step": 1925
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.312e-06,
+      "loss": 0.0353,
+      "step": 1926
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.308000000000001e-06,
+      "loss": 0.0382,
+      "step": 1927
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.304000000000001e-06,
+      "loss": 0.0256,
+      "step": 1928
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.3e-06,
+      "loss": 0.0202,
+      "step": 1929
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.296e-06,
+      "loss": 0.0369,
+      "step": 1930
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.292000000000001e-06,
+      "loss": 0.04,
+      "step": 1931
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.288e-06,
+      "loss": 0.0311,
+      "step": 1932
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.284e-06,
+      "loss": 0.0341,
+      "step": 1933
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 0.0697,
+      "step": 1934
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.276e-06,
+      "loss": 0.0462,
+      "step": 1935
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.272000000000001e-06,
+      "loss": 0.049,
+      "step": 1936
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.2680000000000005e-06,
+      "loss": 0.0756,
+      "step": 1937
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.264e-06,
+      "loss": 0.0217,
+      "step": 1938
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.26e-06,
+      "loss": 0.033,
+      "step": 1939
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.256e-06,
+      "loss": 0.0301,
+      "step": 1940
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.252000000000001e-06,
+      "loss": 0.0435,
+      "step": 1941
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.248000000000001e-06,
+      "loss": 0.0272,
+      "step": 1942
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.244e-06,
+      "loss": 0.0254,
+      "step": 1943
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.24e-06,
+      "loss": 0.0571,
+      "step": 1944
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.236e-06,
+      "loss": 0.0243,
+      "step": 1945
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.232e-06,
+      "loss": 0.089,
+      "step": 1946
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.228000000000001e-06,
+      "loss": 0.0661,
+      "step": 1947
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.2240000000000006e-06,
+      "loss": 0.048,
+      "step": 1948
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.22e-06,
+      "loss": 0.0512,
+      "step": 1949
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.216e-06,
+      "loss": 0.0632,
+      "step": 1950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.2120000000000005e-06,
+      "loss": 0.0525,
+      "step": 1951
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.208e-06,
+      "loss": 0.0426,
+      "step": 1952
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.204e-06,
+      "loss": 0.0856,
+      "step": 1953
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.0364,
+      "step": 1954
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.196e-06,
+      "loss": 0.0699,
+      "step": 1955
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.192000000000001e-06,
+      "loss": 0.045,
+      "step": 1956
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.188e-06,
+      "loss": 0.0398,
+      "step": 1957
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.184e-06,
+      "loss": 0.0363,
+      "step": 1958
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.18e-06,
+      "loss": 0.027,
+      "step": 1959
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.176e-06,
+      "loss": 0.0493,
+      "step": 1960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.172000000000001e-06,
+      "loss": 0.0328,
+      "step": 1961
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.168000000000001e-06,
+      "loss": 0.0371,
+      "step": 1962
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.164e-06,
+      "loss": 0.031,
+      "step": 1963
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.16e-06,
+      "loss": 0.0455,
+      "step": 1964
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.1560000000000005e-06,
+      "loss": 0.032,
+      "step": 1965
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.152e-06,
+      "loss": 0.0576,
+      "step": 1966
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.148000000000001e-06,
+      "loss": 0.0504,
+      "step": 1967
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.1440000000000005e-06,
+      "loss": 0.0762,
+      "step": 1968
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.14e-06,
+      "loss": 0.03,
+      "step": 1969
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.136000000000001e-06,
+      "loss": 0.046,
+      "step": 1970
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.132e-06,
+      "loss": 0.0768,
+      "step": 1971
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.128e-06,
+      "loss": 0.0548,
+      "step": 1972
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.124e-06,
+      "loss": 0.0333,
+      "step": 1973
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.12e-06,
+      "loss": 0.0437,
+      "step": 1974
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.116000000000001e-06,
+      "loss": 0.0519,
+      "step": 1975
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.112000000000001e-06,
+      "loss": 0.0534,
+      "step": 1976
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.108e-06,
+      "loss": 0.0857,
+      "step": 1977
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.104e-06,
+      "loss": 0.0367,
+      "step": 1978
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.1e-06,
+      "loss": 0.0563,
+      "step": 1979
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.096e-06,
+      "loss": 0.0311,
+      "step": 1980
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.092000000000001e-06,
+      "loss": 0.0604,
+      "step": 1981
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.0880000000000005e-06,
+      "loss": 0.0793,
+      "step": 1982
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.084e-06,
+      "loss": 0.0319,
+      "step": 1983
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.08e-06,
+      "loss": 0.0394,
+      "step": 1984
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.0760000000000004e-06,
+      "loss": 0.0863,
+      "step": 1985
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.072e-06,
+      "loss": 0.0499,
+      "step": 1986
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.068000000000001e-06,
+      "loss": 0.0405,
+      "step": 1987
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.064e-06,
+      "loss": 0.0249,
+      "step": 1988
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 0.0434,
+      "step": 1989
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.056000000000001e-06,
+      "loss": 0.035,
+      "step": 1990
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.052e-06,
+      "loss": 0.0534,
+      "step": 1991
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.048e-06,
+      "loss": 0.0591,
+      "step": 1992
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.044e-06,
+      "loss": 0.037,
+      "step": 1993
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.04e-06,
+      "loss": 0.0217,
+      "step": 1994
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.036000000000001e-06,
+      "loss": 0.0333,
+      "step": 1995
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.0320000000000005e-06,
+      "loss": 0.0428,
+      "step": 1996
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.028e-06,
+      "loss": 0.0249,
+      "step": 1997
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.024e-06,
+      "loss": 0.0617,
+      "step": 1998
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 0.0353,
+      "step": 1999
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.016e-06,
+      "loss": 0.0188,
+      "step": 2000
+    },
+    {
+      "epoch": 2.59,
+      "eval_loss": 0.28788238763809204,
+      "eval_runtime": 1845.7942,
+      "eval_samples_per_second": 2.746,
+      "eval_steps_per_second": 0.172,
+      "eval_wer": 20.088193380118845,
+      "step": 2000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.012000000000001e-06,
+      "loss": 0.0309,
+      "step": 2001
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.008e-06,
+      "loss": 0.0566,
+      "step": 2002
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.004e-06,
+      "loss": 0.0212,
+      "step": 2003
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.0421,
+      "step": 2004
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.996e-06,
+      "loss": 0.0481,
+      "step": 2005
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.992e-06,
+      "loss": 0.051,
+      "step": 2006
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.988000000000001e-06,
+      "loss": 0.0487,
+      "step": 2007
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.984e-06,
+      "loss": 0.0432,
+      "step": 2008
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.0405,
+      "step": 2009
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.9760000000000006e-06,
+      "loss": 0.0316,
+      "step": 2010
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.972e-06,
+      "loss": 0.0451,
+      "step": 2011
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.968e-06,
+      "loss": 0.0224,
+      "step": 2012
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.964e-06,
+      "loss": 0.06,
+      "step": 2013
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.96e-06,
+      "loss": 0.0477,
+      "step": 2014
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.956000000000001e-06,
+      "loss": 0.088,
+      "step": 2015
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.9520000000000004e-06,
+      "loss": 0.0624,
+      "step": 2016
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.948e-06,
+      "loss": 0.0574,
+      "step": 2017
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.944e-06,
+      "loss": 0.0377,
+      "step": 2018
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.94e-06,
+      "loss": 0.0459,
+      "step": 2019
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.936e-06,
+      "loss": 0.035,
+      "step": 2020
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.932000000000001e-06,
+      "loss": 0.0426,
+      "step": 2021
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.928e-06,
+      "loss": 0.0472,
+      "step": 2022
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.924000000000001e-06,
+      "loss": 0.028,
+      "step": 2023
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 0.0298,
+      "step": 2024
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.916e-06,
+      "loss": 0.0277,
+      "step": 2025
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.912e-06,
+      "loss": 0.0373,
+      "step": 2026
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.9080000000000005e-06,
+      "loss": 0.0547,
+      "step": 2027
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.904e-06,
+      "loss": 0.0486,
+      "step": 2028
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 0.0782,
+      "step": 2029
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.8960000000000005e-06,
+      "loss": 0.0396,
+      "step": 2030
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.892e-06,
+      "loss": 0.0452,
+      "step": 2031
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.888e-06,
+      "loss": 0.0545,
+      "step": 2032
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.884e-06,
+      "loss": 0.061,
+      "step": 2033
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.88e-06,
+      "loss": 0.0706,
+      "step": 2034
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.876000000000001e-06,
+      "loss": 0.0251,
+      "step": 2035
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.872e-06,
+      "loss": 0.0365,
+      "step": 2036
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.868e-06,
+      "loss": 0.0416,
+      "step": 2037
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.864000000000001e-06,
+      "loss": 0.0487,
+      "step": 2038
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.86e-06,
+      "loss": 0.0613,
+      "step": 2039
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.856e-06,
+      "loss": 0.0353,
+      "step": 2040
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.8520000000000006e-06,
+      "loss": 0.0458,
+      "step": 2041
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.848e-06,
+      "loss": 0.0533,
+      "step": 2042
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.844000000000001e-06,
+      "loss": 0.0407,
+      "step": 2043
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 0.0391,
+      "step": 2044
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.836e-06,
+      "loss": 0.0397,
+      "step": 2045
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.832e-06,
+      "loss": 0.0554,
+      "step": 2046
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.8280000000000004e-06,
+      "loss": 0.0453,
+      "step": 2047
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.824e-06,
+      "loss": 0.0655,
+      "step": 2048
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 0.039,
+      "step": 2049
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.816e-06,
+      "loss": 0.0763,
+      "step": 2050
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.812e-06,
+      "loss": 0.0808,
+      "step": 2051
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.8080000000000006e-06,
+      "loss": 0.0538,
+      "step": 2052
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8040000000000003e-06,
+      "loss": 0.0441,
+      "step": 2053
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.0303,
+      "step": 2054
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.796e-06,
+      "loss": 0.0296,
+      "step": 2055
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7920000000000003e-06,
+      "loss": 0.0454,
+      "step": 2056
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7880000000000004e-06,
+      "loss": 0.0413,
+      "step": 2057
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7840000000000005e-06,
+      "loss": 0.0517,
+      "step": 2058
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.0338,
+      "step": 2059
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7760000000000004e-06,
+      "loss": 0.0387,
+      "step": 2060
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.772e-06,
+      "loss": 0.0235,
+      "step": 2061
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7680000000000006e-06,
+      "loss": 0.0627,
+      "step": 2062
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7640000000000003e-06,
+      "loss": 0.0564,
+      "step": 2063
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 0.0241,
+      "step": 2064
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.756e-06,
+      "loss": 0.0234,
+      "step": 2065
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7520000000000002e-06,
+      "loss": 0.0668,
+      "step": 2066
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7480000000000004e-06,
+      "loss": 0.0559,
+      "step": 2067
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7440000000000005e-06,
+      "loss": 0.0451,
+      "step": 2068
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.74e-06,
+      "loss": 0.0887,
+      "step": 2069
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7360000000000003e-06,
+      "loss": 0.0494,
+      "step": 2070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.732e-06,
+      "loss": 0.0549,
+      "step": 2071
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7280000000000006e-06,
+      "loss": 0.081,
+      "step": 2072
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7240000000000003e-06,
+      "loss": 0.0498,
+      "step": 2073
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 0.0507,
+      "step": 2074
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.716e-06,
+      "loss": 0.061,
+      "step": 2075
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.712e-06,
+      "loss": 0.0352,
+      "step": 2076
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7080000000000003e-06,
+      "loss": 0.0523,
+      "step": 2077
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7040000000000005e-06,
+      "loss": 0.0724,
+      "step": 2078
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7e-06,
+      "loss": 0.0911,
+      "step": 2079
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6960000000000003e-06,
+      "loss": 0.045,
+      "step": 2080
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.692e-06,
+      "loss": 0.0399,
+      "step": 2081
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6880000000000005e-06,
+      "loss": 0.0496,
+      "step": 2082
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6840000000000002e-06,
+      "loss": 0.0625,
+      "step": 2083
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 0.0193,
+      "step": 2084
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.676e-06,
+      "loss": 0.0365,
+      "step": 2085
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6720000000000006e-06,
+      "loss": 0.0323,
+      "step": 2086
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6680000000000003e-06,
+      "loss": 0.0465,
+      "step": 2087
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6640000000000004e-06,
+      "loss": 0.0522,
+      "step": 2088
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.66e-06,
+      "loss": 0.0292,
+      "step": 2089
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6560000000000002e-06,
+      "loss": 0.0662,
+      "step": 2090
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6520000000000004e-06,
+      "loss": 0.0737,
+      "step": 2091
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6480000000000005e-06,
+      "loss": 0.0739,
+      "step": 2092
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.644e-06,
+      "loss": 0.0459,
+      "step": 2093
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 0.0654,
+      "step": 2094
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.636e-06,
+      "loss": 0.0848,
+      "step": 2095
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6320000000000005e-06,
+      "loss": 0.0477,
+      "step": 2096
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6280000000000002e-06,
+      "loss": 0.0421,
+      "step": 2097
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6240000000000004e-06,
+      "loss": 0.0556,
+      "step": 2098
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.62e-06,
+      "loss": 0.0941,
+      "step": 2099
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.616e-06,
+      "loss": 0.0222,
+      "step": 2100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6120000000000003e-06,
+      "loss": 0.0542,
+      "step": 2101
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6080000000000004e-06,
+      "loss": 0.0552,
+      "step": 2102
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.604e-06,
+      "loss": 0.0342,
+      "step": 2103
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.0422,
+      "step": 2104
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.596e-06,
+      "loss": 0.0499,
+      "step": 2105
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.5920000000000005e-06,
+      "loss": 0.049,
+      "step": 2106
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.588e-06,
+      "loss": 0.0944,
+      "step": 2107
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.5840000000000003e-06,
+      "loss": 0.0269,
+      "step": 2108
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.58e-06,
+      "loss": 0.0718,
+      "step": 2109
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.576e-06,
+      "loss": 0.0579,
+      "step": 2110
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.5720000000000003e-06,
+      "loss": 0.0435,
+      "step": 2111
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.5680000000000004e-06,
+      "loss": 0.0614,
+      "step": 2112
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.564e-06,
+      "loss": 0.0433,
+      "step": 2113
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 0.0477,
+      "step": 2114
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.5560000000000008e-06,
+      "loss": 0.0578,
+      "step": 2115
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.5520000000000005e-06,
+      "loss": 0.0695,
+      "step": 2116
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.548e-06,
+      "loss": 0.0578,
+      "step": 2117
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.5440000000000003e-06,
+      "loss": 0.0535,
+      "step": 2118
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.54e-06,
+      "loss": 0.0753,
+      "step": 2119
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.5360000000000005e-06,
+      "loss": 0.0384,
+      "step": 2120
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.5320000000000002e-06,
+      "loss": 0.0512,
+      "step": 2121
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5280000000000004e-06,
+      "loss": 0.0284,
+      "step": 2122
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.524e-06,
+      "loss": 0.0871,
+      "step": 2123
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.52e-06,
+      "loss": 0.0568,
+      "step": 2124
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5160000000000007e-06,
+      "loss": 0.0663,
+      "step": 2125
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5120000000000004e-06,
+      "loss": 0.0689,
+      "step": 2126
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.508e-06,
+      "loss": 0.059,
+      "step": 2127
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5040000000000002e-06,
+      "loss": 0.0776,
+      "step": 2128
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5e-06,
+      "loss": 0.0426,
+      "step": 2129
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4960000000000005e-06,
+      "loss": 0.0448,
+      "step": 2130
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.492e-06,
+      "loss": 0.0489,
+      "step": 2131
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4880000000000003e-06,
+      "loss": 0.0644,
+      "step": 2132
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.484e-06,
+      "loss": 0.0531,
+      "step": 2133
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.48e-06,
+      "loss": 0.067,
+      "step": 2134
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4760000000000007e-06,
+      "loss": 0.0776,
+      "step": 2135
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4720000000000004e-06,
+      "loss": 0.075,
+      "step": 2136
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.468e-06,
+      "loss": 0.0345,
+      "step": 2137
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.464e-06,
+      "loss": 0.0488,
+      "step": 2138
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.46e-06,
+      "loss": 0.0562,
+      "step": 2139
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.4560000000000005e-06,
+      "loss": 0.0421,
+      "step": 2140
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.452e-06,
+      "loss": 0.0361,
+      "step": 2141
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.4480000000000003e-06,
+      "loss": 0.0579,
+      "step": 2142
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.444e-06,
+      "loss": 0.0662,
+      "step": 2143
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.44e-06,
+      "loss": 0.0419,
+      "step": 2144
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.4360000000000006e-06,
+      "loss": 0.0493,
+      "step": 2145
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.4320000000000003e-06,
+      "loss": 0.0511,
+      "step": 2146
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.428e-06,
+      "loss": 0.0404,
+      "step": 2147
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.424e-06,
+      "loss": 0.0389,
+      "step": 2148
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 0.0581,
+      "step": 2149
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.4160000000000004e-06,
+      "loss": 0.0438,
+      "step": 2150
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.412e-06,
+      "loss": 0.0219,
+      "step": 2151
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.4080000000000002e-06,
+      "loss": 0.0202,
+      "step": 2152
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.404e-06,
+      "loss": 0.0588,
+      "step": 2153
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.0775,
+      "step": 2154
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.3960000000000006e-06,
+      "loss": 0.0359,
+      "step": 2155
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.3920000000000003e-06,
+      "loss": 0.0284,
+      "step": 2156
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.388e-06,
+      "loss": 0.0205,
+      "step": 2157
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.384e-06,
+      "loss": 0.0337,
+      "step": 2158
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.0296,
+      "step": 2159
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.3760000000000004e-06,
+      "loss": 0.0765,
+      "step": 2160
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.372e-06,
+      "loss": 0.0383,
+      "step": 2161
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.368e-06,
+      "loss": 0.0422,
+      "step": 2162
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.364e-06,
+      "loss": 0.0261,
+      "step": 2163
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 0.0275,
+      "step": 2164
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3560000000000006e-06,
+      "loss": 0.0564,
+      "step": 2165
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3520000000000003e-06,
+      "loss": 0.069,
+      "step": 2166
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.348e-06,
+      "loss": 0.0657,
+      "step": 2167
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.344e-06,
+      "loss": 0.0504,
+      "step": 2168
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 0.0532,
+      "step": 2169
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3360000000000003e-06,
+      "loss": 0.0344,
+      "step": 2170
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.332e-06,
+      "loss": 0.0522,
+      "step": 2171
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.328e-06,
+      "loss": 0.0334,
+      "step": 2172
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.324e-06,
+      "loss": 0.0599,
+      "step": 2173
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 0.0559,
+      "step": 2174
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3160000000000005e-06,
+      "loss": 0.0295,
+      "step": 2175
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3120000000000002e-06,
+      "loss": 0.0779,
+      "step": 2176
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.308e-06,
+      "loss": 0.039,
+      "step": 2177
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3040000000000005e-06,
+      "loss": 0.0512,
+      "step": 2178
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 0.0649,
+      "step": 2179
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.2960000000000003e-06,
+      "loss": 0.0721,
+      "step": 2180
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.292e-06,
+      "loss": 0.0839,
+      "step": 2181
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.288e-06,
+      "loss": 0.0535,
+      "step": 2182
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.2840000000000007e-06,
+      "loss": 0.0757,
+      "step": 2183
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 0.0452,
+      "step": 2184
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2760000000000005e-06,
+      "loss": 0.0388,
+      "step": 2185
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.272e-06,
+      "loss": 0.0398,
+      "step": 2186
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.268e-06,
+      "loss": 0.0489,
+      "step": 2187
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2640000000000004e-06,
+      "loss": 0.0752,
+      "step": 2188
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 0.058,
+      "step": 2189
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2560000000000003e-06,
+      "loss": 0.0366,
+      "step": 2190
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.252e-06,
+      "loss": 0.0506,
+      "step": 2191
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.248e-06,
+      "loss": 0.0335,
+      "step": 2192
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.2440000000000006e-06,
+      "loss": 0.0609,
+      "step": 2193
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 0.0548,
+      "step": 2194
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.2360000000000004e-06,
+      "loss": 0.0414,
+      "step": 2195
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.232e-06,
+      "loss": 0.0528,
+      "step": 2196
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.228e-06,
+      "loss": 0.0277,
+      "step": 2197
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.2240000000000004e-06,
+      "loss": 0.0268,
+      "step": 2198
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 0.0368,
+      "step": 2199
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.216e-06,
+      "loss": 0.0398,
+      "step": 2200
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.212e-06,
+      "loss": 0.0552,
+      "step": 2201
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.208e-06,
+      "loss": 0.027,
+      "step": 2202
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.2040000000000006e-06,
+      "loss": 0.0541,
+      "step": 2203
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.0263,
+      "step": 2204
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.1960000000000004e-06,
+      "loss": 0.0421,
+      "step": 2205
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.192e-06,
+      "loss": 0.0204,
+      "step": 2206
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.188e-06,
+      "loss": 0.0968,
+      "step": 2207
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.1840000000000003e-06,
+      "loss": 0.0542,
+      "step": 2208
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.0626,
+      "step": 2209
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.176e-06,
+      "loss": 0.0599,
+      "step": 2210
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.172e-06,
+      "loss": 0.0742,
+      "step": 2211
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.1680000000000004e-06,
+      "loss": 0.0715,
+      "step": 2212
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.1640000000000005e-06,
+      "loss": 0.0227,
+      "step": 2213
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 0.0718,
+      "step": 2214
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.1560000000000004e-06,
+      "loss": 0.0467,
+      "step": 2215
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.152e-06,
+      "loss": 0.0474,
+      "step": 2216
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.1480000000000006e-06,
+      "loss": 0.0464,
+      "step": 2217
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.1440000000000003e-06,
+      "loss": 0.0331,
+      "step": 2218
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 0.0318,
+      "step": 2219
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.136e-06,
+      "loss": 0.0288,
+      "step": 2220
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.132e-06,
+      "loss": 0.0325,
+      "step": 2221
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.1280000000000004e-06,
+      "loss": 0.032,
+      "step": 2222
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.1240000000000005e-06,
+      "loss": 0.0583,
+      "step": 2223
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.12e-06,
+      "loss": 0.0257,
+      "step": 2224
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.1160000000000003e-06,
+      "loss": 0.0705,
+      "step": 2225
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.112e-06,
+      "loss": 0.0339,
+      "step": 2226
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.1080000000000006e-06,
+      "loss": 0.0298,
+      "step": 2227
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.1040000000000003e-06,
+      "loss": 0.0653,
+      "step": 2228
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 0.0482,
+      "step": 2229
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.096e-06,
+      "loss": 0.037,
+      "step": 2230
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.092e-06,
+      "loss": 0.0449,
+      "step": 2231
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.0880000000000003e-06,
+      "loss": 0.0733,
+      "step": 2232
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.0840000000000005e-06,
+      "loss": 0.0249,
+      "step": 2233
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.08e-06,
+      "loss": 0.0211,
+      "step": 2234
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.0760000000000003e-06,
+      "loss": 0.0816,
+      "step": 2235
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.072e-06,
+      "loss": 0.0779,
+      "step": 2236
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.0680000000000005e-06,
+      "loss": 0.0506,
+      "step": 2237
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.0640000000000002e-06,
+      "loss": 0.0304,
+      "step": 2238
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 0.041,
+      "step": 2239
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.056e-06,
+      "loss": 0.0341,
+      "step": 2240
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.0520000000000006e-06,
+      "loss": 0.0369,
+      "step": 2241
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.0480000000000003e-06,
+      "loss": 0.0431,
+      "step": 2242
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.0440000000000004e-06,
+      "loss": 0.0288,
+      "step": 2243
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.04e-06,
+      "loss": 0.0705,
+      "step": 2244
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.0360000000000002e-06,
+      "loss": 0.0475,
+      "step": 2245
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.0320000000000004e-06,
+      "loss": 0.0336,
+      "step": 2246
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.0280000000000005e-06,
+      "loss": 0.0367,
+      "step": 2247
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.024e-06,
+      "loss": 0.0475,
+      "step": 2248
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 0.0297,
+      "step": 2249
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.016e-06,
+      "loss": 0.0295,
+      "step": 2250
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.0120000000000006e-06,
+      "loss": 0.072,
+      "step": 2251
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.0080000000000003e-06,
+      "loss": 0.023,
+      "step": 2252
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.0040000000000004e-06,
+      "loss": 0.038,
+      "step": 2253
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3e-06,
+      "loss": 0.049,
+      "step": 2254
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.996e-06,
+      "loss": 0.0348,
+      "step": 2255
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9920000000000003e-06,
+      "loss": 0.0235,
+      "step": 2256
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9880000000000004e-06,
+      "loss": 0.0529,
+      "step": 2257
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.984e-06,
+      "loss": 0.0272,
+      "step": 2258
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.0207,
+      "step": 2259
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.976e-06,
+      "loss": 0.1006,
+      "step": 2260
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9720000000000005e-06,
+      "loss": 0.0554,
+      "step": 2261
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9680000000000002e-06,
+      "loss": 0.0409,
+      "step": 2262
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9640000000000003e-06,
+      "loss": 0.0694,
+      "step": 2263
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.96e-06,
+      "loss": 0.0446,
+      "step": 2264
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.956e-06,
+      "loss": 0.0897,
+      "step": 2265
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9520000000000003e-06,
+      "loss": 0.0445,
+      "step": 2266
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9480000000000004e-06,
+      "loss": 0.1219,
+      "step": 2267
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.944e-06,
+      "loss": 0.0399,
+      "step": 2268
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 0.0544,
+      "step": 2269
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9360000000000003e-06,
+      "loss": 0.0342,
+      "step": 2270
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9320000000000005e-06,
+      "loss": 0.0808,
+      "step": 2271
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.928e-06,
+      "loss": 0.0373,
+      "step": 2272
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9240000000000003e-06,
+      "loss": 0.0385,
+      "step": 2273
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.92e-06,
+      "loss": 0.0605,
+      "step": 2274
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9160000000000005e-06,
+      "loss": 0.0321,
+      "step": 2275
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9120000000000002e-06,
+      "loss": 0.0297,
+      "step": 2276
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9080000000000004e-06,
+      "loss": 0.0458,
+      "step": 2277
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.904e-06,
+      "loss": 0.0328,
+      "step": 2278
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9e-06,
+      "loss": 0.0766,
+      "step": 2279
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.8960000000000003e-06,
+      "loss": 0.0401,
+      "step": 2280
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.8920000000000004e-06,
+      "loss": 0.0428,
+      "step": 2281
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.888e-06,
+      "loss": 0.0595,
+      "step": 2282
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.8840000000000003e-06,
+      "loss": 0.0632,
+      "step": 2283
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.88e-06,
+      "loss": 0.0462,
+      "step": 2284
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.8760000000000005e-06,
+      "loss": 0.0394,
+      "step": 2285
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.872e-06,
+      "loss": 0.047,
+      "step": 2286
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.8680000000000003e-06,
+      "loss": 0.0342,
+      "step": 2287
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.864e-06,
+      "loss": 0.0538,
+      "step": 2288
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.86e-06,
+      "loss": 0.069,
+      "step": 2289
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.8560000000000003e-06,
+      "loss": 0.0291,
+      "step": 2290
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.8520000000000004e-06,
+      "loss": 0.0573,
+      "step": 2291
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.848e-06,
+      "loss": 0.0228,
+      "step": 2292
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.8440000000000002e-06,
+      "loss": 0.0679,
+      "step": 2293
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.84e-06,
+      "loss": 0.0415,
+      "step": 2294
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.8360000000000005e-06,
+      "loss": 0.0502,
+      "step": 2295
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.832e-06,
+      "loss": 0.0367,
+      "step": 2296
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.8280000000000003e-06,
+      "loss": 0.0321,
+      "step": 2297
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.824e-06,
+      "loss": 0.0454,
+      "step": 2298
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.82e-06,
+      "loss": 0.0333,
+      "step": 2299
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8160000000000002e-06,
+      "loss": 0.07,
+      "step": 2300
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8120000000000004e-06,
+      "loss": 0.0479,
+      "step": 2301
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.808e-06,
+      "loss": 0.0771,
+      "step": 2302
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.804e-06,
+      "loss": 0.0422,
+      "step": 2303
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.0444,
+      "step": 2304
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7960000000000004e-06,
+      "loss": 0.0733,
+      "step": 2305
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.792e-06,
+      "loss": 0.063,
+      "step": 2306
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7880000000000002e-06,
+      "loss": 0.0305,
+      "step": 2307
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.784e-06,
+      "loss": 0.0829,
+      "step": 2308
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.0486,
+      "step": 2309
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.776e-06,
+      "loss": 0.0393,
+      "step": 2310
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.7720000000000003e-06,
+      "loss": 0.1005,
+      "step": 2311
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.768e-06,
+      "loss": 0.0648,
+      "step": 2312
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.764e-06,
+      "loss": 0.0369,
+      "step": 2313
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 0.0348,
+      "step": 2314
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.7560000000000004e-06,
+      "loss": 0.0469,
+      "step": 2315
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.752e-06,
+      "loss": 0.0362,
+      "step": 2316
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.748e-06,
+      "loss": 0.0546,
+      "step": 2317
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.744e-06,
+      "loss": 0.0706,
+      "step": 2318
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 0.0611,
+      "step": 2319
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.736e-06,
+      "loss": 0.0166,
+      "step": 2320
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7320000000000003e-06,
+      "loss": 0.0378,
+      "step": 2321
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.728e-06,
+      "loss": 0.0287,
+      "step": 2322
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.724e-06,
+      "loss": 0.0274,
+      "step": 2323
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 0.0113,
+      "step": 2324
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7160000000000003e-06,
+      "loss": 0.0141,
+      "step": 2325
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.712e-06,
+      "loss": 0.0227,
+      "step": 2326
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.708e-06,
+      "loss": 0.0331,
+      "step": 2327
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.704e-06,
+      "loss": 0.0171,
+      "step": 2328
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 0.0359,
+      "step": 2329
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.696e-06,
+      "loss": 0.014,
+      "step": 2330
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6920000000000002e-06,
+      "loss": 0.018,
+      "step": 2331
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.688e-06,
+      "loss": 0.0208,
+      "step": 2332
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6840000000000005e-06,
+      "loss": 0.0105,
+      "step": 2333
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.68e-06,
+      "loss": 0.0292,
+      "step": 2334
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6760000000000003e-06,
+      "loss": 0.022,
+      "step": 2335
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.672e-06,
+      "loss": 0.0272,
+      "step": 2336
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.668e-06,
+      "loss": 0.0166,
+      "step": 2337
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6640000000000007e-06,
+      "loss": 0.0299,
+      "step": 2338
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 0.0147,
+      "step": 2339
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.656e-06,
+      "loss": 0.0135,
+      "step": 2340
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.652e-06,
+      "loss": 0.0194,
+      "step": 2341
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.648e-06,
+      "loss": 0.0104,
+      "step": 2342
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.6440000000000004e-06,
+      "loss": 0.0498,
+      "step": 2343
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.64e-06,
+      "loss": 0.0131,
+      "step": 2344
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.6360000000000003e-06,
+      "loss": 0.0292,
+      "step": 2345
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.632e-06,
+      "loss": 0.0171,
+      "step": 2346
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.628e-06,
+      "loss": 0.0147,
+      "step": 2347
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.6240000000000006e-06,
+      "loss": 0.0192,
+      "step": 2348
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 0.0145,
+      "step": 2349
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.616e-06,
+      "loss": 0.0182,
+      "step": 2350
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.612e-06,
+      "loss": 0.0137,
+      "step": 2351
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.608e-06,
+      "loss": 0.0392,
+      "step": 2352
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.6040000000000004e-06,
+      "loss": 0.014,
+      "step": 2353
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.6e-06,
+      "loss": 0.0178,
+      "step": 2354
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.5960000000000002e-06,
+      "loss": 0.02,
+      "step": 2355
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.592e-06,
+      "loss": 0.0142,
+      "step": 2356
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.588e-06,
+      "loss": 0.0242,
+      "step": 2357
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.5840000000000006e-06,
+      "loss": 0.0285,
+      "step": 2358
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.068,
+      "step": 2359
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.576e-06,
+      "loss": 0.0318,
+      "step": 2360
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.572e-06,
+      "loss": 0.0291,
+      "step": 2361
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.568e-06,
+      "loss": 0.0319,
+      "step": 2362
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.5640000000000004e-06,
+      "loss": 0.0163,
+      "step": 2363
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.56e-06,
+      "loss": 0.0218,
+      "step": 2364
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.556e-06,
+      "loss": 0.037,
+      "step": 2365
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.552e-06,
+      "loss": 0.0186,
+      "step": 2366
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.5480000000000004e-06,
+      "loss": 0.0426,
+      "step": 2367
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.5440000000000005e-06,
+      "loss": 0.0147,
+      "step": 2368
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 0.0163,
+      "step": 2369
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.536e-06,
+      "loss": 0.0132,
+      "step": 2370
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.532e-06,
+      "loss": 0.02,
+      "step": 2371
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.5280000000000006e-06,
+      "loss": 0.0301,
+      "step": 2372
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.5240000000000003e-06,
+      "loss": 0.0238,
+      "step": 2373
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.52e-06,
+      "loss": 0.0181,
+      "step": 2374
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.516e-06,
+      "loss": 0.0223,
+      "step": 2375
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.512e-06,
+      "loss": 0.0157,
+      "step": 2376
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.5080000000000004e-06,
+      "loss": 0.0313,
+      "step": 2377
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.5040000000000005e-06,
+      "loss": 0.0185,
+      "step": 2378
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.5e-06,
+      "loss": 0.0241,
+      "step": 2379
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.496e-06,
+      "loss": 0.0147,
+      "step": 2380
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.4920000000000005e-06,
+      "loss": 0.0111,
+      "step": 2381
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.488e-06,
+      "loss": 0.0335,
+      "step": 2382
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.4840000000000003e-06,
+      "loss": 0.0369,
+      "step": 2383
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 0.0148,
+      "step": 2384
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.476e-06,
+      "loss": 0.0218,
+      "step": 2385
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.4720000000000002e-06,
+      "loss": 0.0145,
+      "step": 2386
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.468e-06,
+      "loss": 0.0116,
+      "step": 2387
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.4640000000000005e-06,
+      "loss": 0.0182,
+      "step": 2388
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.46e-06,
+      "loss": 0.0256,
+      "step": 2389
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.4560000000000003e-06,
+      "loss": 0.0277,
+      "step": 2390
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.4520000000000004e-06,
+      "loss": 0.0204,
+      "step": 2391
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.448e-06,
+      "loss": 0.0186,
+      "step": 2392
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.4440000000000002e-06,
+      "loss": 0.0277,
+      "step": 2393
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 0.0231,
+      "step": 2394
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.4360000000000005e-06,
+      "loss": 0.0174,
+      "step": 2395
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.432e-06,
+      "loss": 0.0229,
+      "step": 2396
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.428e-06,
+      "loss": 0.0187,
+      "step": 2397
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.4240000000000004e-06,
+      "loss": 0.0175,
+      "step": 2398
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.42e-06,
+      "loss": 0.024,
+      "step": 2399
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.4160000000000002e-06,
+      "loss": 0.0361,
+      "step": 2400
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4120000000000004e-06,
+      "loss": 0.048,
+      "step": 2401
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.408e-06,
+      "loss": 0.0264,
+      "step": 2402
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.404e-06,
+      "loss": 0.0254,
+      "step": 2403
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.0279,
+      "step": 2404
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.3960000000000004e-06,
+      "loss": 0.0146,
+      "step": 2405
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.392e-06,
+      "loss": 0.0246,
+      "step": 2406
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.3880000000000003e-06,
+      "loss": 0.0283,
+      "step": 2407
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.3840000000000004e-06,
+      "loss": 0.0106,
+      "step": 2408
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.38e-06,
+      "loss": 0.0119,
+      "step": 2409
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.376e-06,
+      "loss": 0.0111,
+      "step": 2410
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.3720000000000003e-06,
+      "loss": 0.0144,
+      "step": 2411
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.3680000000000005e-06,
+      "loss": 0.028,
+      "step": 2412
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.364e-06,
+      "loss": 0.0269,
+      "step": 2413
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 0.0289,
+      "step": 2414
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.3560000000000004e-06,
+      "loss": 0.0179,
+      "step": 2415
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.352e-06,
+      "loss": 0.0145,
+      "step": 2416
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3480000000000002e-06,
+      "loss": 0.0151,
+      "step": 2417
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3440000000000003e-06,
+      "loss": 0.0188,
+      "step": 2418
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 0.0131,
+      "step": 2419
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.336e-06,
+      "loss": 0.0257,
+      "step": 2420
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3320000000000003e-06,
+      "loss": 0.0293,
+      "step": 2421
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3280000000000004e-06,
+      "loss": 0.032,
+      "step": 2422
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.324e-06,
+      "loss": 0.0115,
+      "step": 2423
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 0.0176,
+      "step": 2424
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3160000000000004e-06,
+      "loss": 0.024,
+      "step": 2425
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.312e-06,
+      "loss": 0.0217,
+      "step": 2426
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.308e-06,
+      "loss": 0.0141,
+      "step": 2427
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3040000000000003e-06,
+      "loss": 0.0498,
+      "step": 2428
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 0.0271,
+      "step": 2429
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.296e-06,
+      "loss": 0.0263,
+      "step": 2430
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.2920000000000002e-06,
+      "loss": 0.0144,
+      "step": 2431
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2880000000000004e-06,
+      "loss": 0.0192,
+      "step": 2432
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.284e-06,
+      "loss": 0.0259,
+      "step": 2433
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.28e-06,
+      "loss": 0.0253,
+      "step": 2434
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2760000000000003e-06,
+      "loss": 0.0372,
+      "step": 2435
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2720000000000004e-06,
+      "loss": 0.0168,
+      "step": 2436
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.268e-06,
+      "loss": 0.0366,
+      "step": 2437
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2640000000000003e-06,
+      "loss": 0.0364,
+      "step": 2438
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 0.019,
+      "step": 2439
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.256e-06,
+      "loss": 0.0219,
+      "step": 2440
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.252e-06,
+      "loss": 0.0143,
+      "step": 2441
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2480000000000003e-06,
+      "loss": 0.0159,
+      "step": 2442
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.244e-06,
+      "loss": 0.0402,
+      "step": 2443
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.24e-06,
+      "loss": 0.0173,
+      "step": 2444
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2360000000000003e-06,
+      "loss": 0.0106,
+      "step": 2445
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2320000000000004e-06,
+      "loss": 0.0435,
+      "step": 2446
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.228e-06,
+      "loss": 0.0278,
+      "step": 2447
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.2240000000000002e-06,
+      "loss": 0.0359,
+      "step": 2448
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 0.0143,
+      "step": 2449
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.216e-06,
+      "loss": 0.0259,
+      "step": 2450
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.212e-06,
+      "loss": 0.0112,
+      "step": 2451
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.2080000000000003e-06,
+      "loss": 0.0243,
+      "step": 2452
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.2040000000000004e-06,
+      "loss": 0.0158,
+      "step": 2453
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.2e-06,
+      "loss": 0.0212,
+      "step": 2454
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1960000000000002e-06,
+      "loss": 0.0152,
+      "step": 2455
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1920000000000004e-06,
+      "loss": 0.0459,
+      "step": 2456
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.188e-06,
+      "loss": 0.0115,
+      "step": 2457
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.184e-06,
+      "loss": 0.0245,
+      "step": 2458
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.0143,
+      "step": 2459
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.176e-06,
+      "loss": 0.0346,
+      "step": 2460
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.172e-06,
+      "loss": 0.0402,
+      "step": 2461
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1680000000000002e-06,
+      "loss": 0.0222,
+      "step": 2462
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.1640000000000004e-06,
+      "loss": 0.0166,
+      "step": 2463
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.16e-06,
+      "loss": 0.0193,
+      "step": 2464
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.156e-06,
+      "loss": 0.0469,
+      "step": 2465
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.1520000000000003e-06,
+      "loss": 0.0303,
+      "step": 2466
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.148e-06,
+      "loss": 0.0128,
+      "step": 2467
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.144e-06,
+      "loss": 0.0153,
+      "step": 2468
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 0.011,
+      "step": 2469
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.1360000000000004e-06,
+      "loss": 0.0318,
+      "step": 2470
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.132e-06,
+      "loss": 0.0165,
+      "step": 2471
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.128e-06,
+      "loss": 0.0367,
+      "step": 2472
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.1240000000000003e-06,
+      "loss": 0.018,
+      "step": 2473
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.12e-06,
+      "loss": 0.0176,
+      "step": 2474
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.116e-06,
+      "loss": 0.0147,
+      "step": 2475
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.1120000000000003e-06,
+      "loss": 0.0202,
+      "step": 2476
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.108e-06,
+      "loss": 0.0229,
+      "step": 2477
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.104e-06,
+      "loss": 0.0128,
+      "step": 2478
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 0.0164,
+      "step": 2479
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0960000000000003e-06,
+      "loss": 0.0147,
+      "step": 2480
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.092e-06,
+      "loss": 0.0186,
+      "step": 2481
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.088e-06,
+      "loss": 0.0431,
+      "step": 2482
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0840000000000003e-06,
+      "loss": 0.0106,
+      "step": 2483
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.08e-06,
+      "loss": 0.0202,
+      "step": 2484
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.076e-06,
+      "loss": 0.0295,
+      "step": 2485
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0720000000000002e-06,
+      "loss": 0.0122,
+      "step": 2486
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0680000000000004e-06,
+      "loss": 0.0093,
+      "step": 2487
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.064e-06,
+      "loss": 0.0223,
+      "step": 2488
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.06e-06,
+      "loss": 0.0186,
+      "step": 2489
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0560000000000003e-06,
+      "loss": 0.0197,
+      "step": 2490
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.052e-06,
+      "loss": 0.027,
+      "step": 2491
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.048e-06,
+      "loss": 0.0252,
+      "step": 2492
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.0440000000000003e-06,
+      "loss": 0.0177,
+      "step": 2493
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.04e-06,
+      "loss": 0.0167,
+      "step": 2494
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.036e-06,
+      "loss": 0.0307,
+      "step": 2495
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.032e-06,
+      "loss": 0.0173,
+      "step": 2496
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.0280000000000003e-06,
+      "loss": 0.0181,
+      "step": 2497
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.024e-06,
+      "loss": 0.0215,
+      "step": 2498
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.02e-06,
+      "loss": 0.0543,
+      "step": 2499
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.0160000000000003e-06,
+      "loss": 0.0394,
+      "step": 2500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.012e-06,
+      "loss": 0.012,
+      "step": 2501
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.008e-06,
+      "loss": 0.0204,
+      "step": 2502
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.004e-06,
+      "loss": 0.0406,
+      "step": 2503
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.0215,
+      "step": 2504
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.996e-06,
+      "loss": 0.0158,
+      "step": 2505
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.992e-06,
+      "loss": 0.0124,
+      "step": 2506
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.9880000000000003e-06,
+      "loss": 0.0173,
+      "step": 2507
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.984e-06,
+      "loss": 0.0129,
+      "step": 2508
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.98e-06,
+      "loss": 0.025,
+      "step": 2509
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.9760000000000002e-06,
+      "loss": 0.0256,
+      "step": 2510
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.972e-06,
+      "loss": 0.0224,
+      "step": 2511
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.968e-06,
+      "loss": 0.0125,
+      "step": 2512
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.964e-06,
+      "loss": 0.0209,
+      "step": 2513
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 0.0121,
+      "step": 2514
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.956e-06,
+      "loss": 0.0257,
+      "step": 2515
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.952e-06,
+      "loss": 0.0241,
+      "step": 2516
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.9480000000000002e-06,
+      "loss": 0.0125,
+      "step": 2517
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.944e-06,
+      "loss": 0.0138,
+      "step": 2518
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.94e-06,
+      "loss": 0.0217,
+      "step": 2519
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.936e-06,
+      "loss": 0.0184,
+      "step": 2520
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.9320000000000003e-06,
+      "loss": 0.0158,
+      "step": 2521
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.928e-06,
+      "loss": 0.0266,
+      "step": 2522
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.924e-06,
+      "loss": 0.0397,
+      "step": 2523
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 0.018,
+      "step": 2524
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.916e-06,
+      "loss": 0.0267,
+      "step": 2525
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.912e-06,
+      "loss": 0.0314,
+      "step": 2526
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.908e-06,
+      "loss": 0.0198,
+      "step": 2527
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.9040000000000003e-06,
+      "loss": 0.0224,
+      "step": 2528
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 0.0333,
+      "step": 2529
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.8960000000000001e-06,
+      "loss": 0.0179,
+      "step": 2530
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.8920000000000003e-06,
+      "loss": 0.0128,
+      "step": 2531
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8880000000000002e-06,
+      "loss": 0.0116,
+      "step": 2532
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8840000000000003e-06,
+      "loss": 0.0299,
+      "step": 2533
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 0.0168,
+      "step": 2534
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8760000000000001e-06,
+      "loss": 0.0109,
+      "step": 2535
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8720000000000002e-06,
+      "loss": 0.0251,
+      "step": 2536
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8680000000000002e-06,
+      "loss": 0.0268,
+      "step": 2537
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8640000000000003e-06,
+      "loss": 0.022,
+      "step": 2538
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 0.0174,
+      "step": 2539
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.856e-06,
+      "loss": 0.0301,
+      "step": 2540
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8520000000000002e-06,
+      "loss": 0.0152,
+      "step": 2541
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8480000000000001e-06,
+      "loss": 0.0151,
+      "step": 2542
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8440000000000003e-06,
+      "loss": 0.0486,
+      "step": 2543
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 0.0185,
+      "step": 2544
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8360000000000003e-06,
+      "loss": 0.0209,
+      "step": 2545
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8320000000000002e-06,
+      "loss": 0.0153,
+      "step": 2546
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8280000000000001e-06,
+      "loss": 0.0321,
+      "step": 2547
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8240000000000002e-06,
+      "loss": 0.0145,
+      "step": 2548
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 0.0323,
+      "step": 2549
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8160000000000003e-06,
+      "loss": 0.0144,
+      "step": 2550
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8120000000000002e-06,
+      "loss": 0.0194,
+      "step": 2551
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.808e-06,
+      "loss": 0.026,
+      "step": 2552
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8040000000000002e-06,
+      "loss": 0.0179,
+      "step": 2553
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 0.0275,
+      "step": 2554
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7960000000000003e-06,
+      "loss": 0.0219,
+      "step": 2555
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7920000000000002e-06,
+      "loss": 0.0231,
+      "step": 2556
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.788e-06,
+      "loss": 0.0213,
+      "step": 2557
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7840000000000002e-06,
+      "loss": 0.0192,
+      "step": 2558
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.0139,
+      "step": 2559
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7760000000000002e-06,
+      "loss": 0.0305,
+      "step": 2560
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7720000000000001e-06,
+      "loss": 0.0155,
+      "step": 2561
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7680000000000003e-06,
+      "loss": 0.0259,
+      "step": 2562
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7640000000000002e-06,
+      "loss": 0.012,
+      "step": 2563
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.76e-06,
+      "loss": 0.0224,
+      "step": 2564
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7560000000000002e-06,
+      "loss": 0.0158,
+      "step": 2565
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7520000000000001e-06,
+      "loss": 0.0186,
+      "step": 2566
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7480000000000002e-06,
+      "loss": 0.0118,
+      "step": 2567
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7440000000000002e-06,
+      "loss": 0.0327,
+      "step": 2568
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.74e-06,
+      "loss": 0.0503,
+      "step": 2569
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7360000000000002e-06,
+      "loss": 0.0135,
+      "step": 2570
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.732e-06,
+      "loss": 0.0186,
+      "step": 2571
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7280000000000002e-06,
+      "loss": 0.0125,
+      "step": 2572
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7240000000000001e-06,
+      "loss": 0.0185,
+      "step": 2573
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.72e-06,
+      "loss": 0.0122,
+      "step": 2574
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7160000000000002e-06,
+      "loss": 0.0219,
+      "step": 2575
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.712e-06,
+      "loss": 0.0114,
+      "step": 2576
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7080000000000002e-06,
+      "loss": 0.0129,
+      "step": 2577
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7040000000000001e-06,
+      "loss": 0.0143,
+      "step": 2578
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 0.0134,
+      "step": 2579
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6960000000000002e-06,
+      "loss": 0.0192,
+      "step": 2580
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.692e-06,
+      "loss": 0.0439,
+      "step": 2581
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6880000000000002e-06,
+      "loss": 0.0116,
+      "step": 2582
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.684e-06,
+      "loss": 0.0242,
+      "step": 2583
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 0.0244,
+      "step": 2584
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6760000000000001e-06,
+      "loss": 0.0174,
+      "step": 2585
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.672e-06,
+      "loss": 0.0389,
+      "step": 2586
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6680000000000002e-06,
+      "loss": 0.0144,
+      "step": 2587
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.664e-06,
+      "loss": 0.018,
+      "step": 2588
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 0.0246,
+      "step": 2589
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6560000000000001e-06,
+      "loss": 0.0109,
+      "step": 2590
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6520000000000002e-06,
+      "loss": 0.0187,
+      "step": 2591
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6480000000000001e-06,
+      "loss": 0.0252,
+      "step": 2592
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.644e-06,
+      "loss": 0.0172,
+      "step": 2593
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 0.0239,
+      "step": 2594
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.636e-06,
+      "loss": 0.0094,
+      "step": 2595
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6320000000000002e-06,
+      "loss": 0.025,
+      "step": 2596
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6280000000000001e-06,
+      "loss": 0.0154,
+      "step": 2597
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.624e-06,
+      "loss": 0.0128,
+      "step": 2598
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 0.0102,
+      "step": 2599
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.616e-06,
+      "loss": 0.011,
+      "step": 2600
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6120000000000002e-06,
+      "loss": 0.0092,
+      "step": 2601
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.608e-06,
+      "loss": 0.0268,
+      "step": 2602
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.604e-06,
+      "loss": 0.015,
+      "step": 2603
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.0294,
+      "step": 2604
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.596e-06,
+      "loss": 0.0067,
+      "step": 2605
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5920000000000002e-06,
+      "loss": 0.0462,
+      "step": 2606
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.588e-06,
+      "loss": 0.032,
+      "step": 2607
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5840000000000002e-06,
+      "loss": 0.0173,
+      "step": 2608
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.0143,
+      "step": 2609
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.576e-06,
+      "loss": 0.0256,
+      "step": 2610
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.5720000000000002e-06,
+      "loss": 0.0349,
+      "step": 2611
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.568e-06,
+      "loss": 0.0308,
+      "step": 2612
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.5640000000000002e-06,
+      "loss": 0.0243,
+      "step": 2613
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.56e-06,
+      "loss": 0.013,
+      "step": 2614
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.556e-06,
+      "loss": 0.0409,
+      "step": 2615
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.5520000000000001e-06,
+      "loss": 0.0244,
+      "step": 2616
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.548e-06,
+      "loss": 0.0197,
+      "step": 2617
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.5440000000000002e-06,
+      "loss": 0.0182,
+      "step": 2618
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.54e-06,
+      "loss": 0.0129,
+      "step": 2619
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.536e-06,
+      "loss": 0.0159,
+      "step": 2620
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.5320000000000001e-06,
+      "loss": 0.0192,
+      "step": 2621
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.528e-06,
+      "loss": 0.0153,
+      "step": 2622
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.5240000000000001e-06,
+      "loss": 0.027,
+      "step": 2623
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.52e-06,
+      "loss": 0.0135,
+      "step": 2624
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.5160000000000002e-06,
+      "loss": 0.0143,
+      "step": 2625
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.512e-06,
+      "loss": 0.0138,
+      "step": 2626
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.508e-06,
+      "loss": 0.0214,
+      "step": 2627
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.5040000000000001e-06,
+      "loss": 0.0141,
+      "step": 2628
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.5e-06,
+      "loss": 0.0337,
+      "step": 2629
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.4960000000000002e-06,
+      "loss": 0.0299,
+      "step": 2630
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.492e-06,
+      "loss": 0.0129,
+      "step": 2631
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.488e-06,
+      "loss": 0.0189,
+      "step": 2632
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4840000000000001e-06,
+      "loss": 0.0187,
+      "step": 2633
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.48e-06,
+      "loss": 0.0133,
+      "step": 2634
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4760000000000001e-06,
+      "loss": 0.0209,
+      "step": 2635
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.472e-06,
+      "loss": 0.0366,
+      "step": 2636
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4680000000000002e-06,
+      "loss": 0.0102,
+      "step": 2637
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.464e-06,
+      "loss": 0.0175,
+      "step": 2638
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.46e-06,
+      "loss": 0.0188,
+      "step": 2639
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4560000000000001e-06,
+      "loss": 0.0136,
+      "step": 2640
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.452e-06,
+      "loss": 0.0297,
+      "step": 2641
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4480000000000002e-06,
+      "loss": 0.0258,
+      "step": 2642
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.444e-06,
+      "loss": 0.0327,
+      "step": 2643
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.44e-06,
+      "loss": 0.0212,
+      "step": 2644
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.436e-06,
+      "loss": 0.018,
+      "step": 2645
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.432e-06,
+      "loss": 0.0214,
+      "step": 2646
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4280000000000001e-06,
+      "loss": 0.0136,
+      "step": 2647
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.424e-06,
+      "loss": 0.0169,
+      "step": 2648
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.42e-06,
+      "loss": 0.0192,
+      "step": 2649
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.416e-06,
+      "loss": 0.0129,
+      "step": 2650
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.412e-06,
+      "loss": 0.0154,
+      "step": 2651
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.4080000000000001e-06,
+      "loss": 0.0141,
+      "step": 2652
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.404e-06,
+      "loss": 0.0185,
+      "step": 2653
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 0.0086,
+      "step": 2654
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.396e-06,
+      "loss": 0.0156,
+      "step": 2655
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.392e-06,
+      "loss": 0.0165,
+      "step": 2656
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.388e-06,
+      "loss": 0.0148,
+      "step": 2657
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.384e-06,
+      "loss": 0.0218,
+      "step": 2658
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.012,
+      "step": 2659
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.376e-06,
+      "loss": 0.0148,
+      "step": 2660
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.372e-06,
+      "loss": 0.0212,
+      "step": 2661
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.368e-06,
+      "loss": 0.0215,
+      "step": 2662
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.364e-06,
+      "loss": 0.0184,
+      "step": 2663
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 0.012,
+      "step": 2664
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.356e-06,
+      "loss": 0.0163,
+      "step": 2665
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.352e-06,
+      "loss": 0.0217,
+      "step": 2666
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.348e-06,
+      "loss": 0.0112,
+      "step": 2667
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.344e-06,
+      "loss": 0.0154,
+      "step": 2668
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.34e-06,
+      "loss": 0.0153,
+      "step": 2669
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.336e-06,
+      "loss": 0.0108,
+      "step": 2670
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3320000000000003e-06,
+      "loss": 0.0275,
+      "step": 2671
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.328e-06,
+      "loss": 0.0207,
+      "step": 2672
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.324e-06,
+      "loss": 0.0551,
+      "step": 2673
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.32e-06,
+      "loss": 0.0217,
+      "step": 2674
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.316e-06,
+      "loss": 0.0201,
+      "step": 2675
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3120000000000003e-06,
+      "loss": 0.0243,
+      "step": 2676
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.308e-06,
+      "loss": 0.0417,
+      "step": 2677
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.304e-06,
+      "loss": 0.016,
+      "step": 2678
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3e-06,
+      "loss": 0.0182,
+      "step": 2679
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.296e-06,
+      "loss": 0.0153,
+      "step": 2680
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.2920000000000003e-06,
+      "loss": 0.0179,
+      "step": 2681
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.288e-06,
+      "loss": 0.0186,
+      "step": 2682
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.284e-06,
+      "loss": 0.0109,
+      "step": 2683
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.28e-06,
+      "loss": 0.0277,
+      "step": 2684
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.276e-06,
+      "loss": 0.0301,
+      "step": 2685
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.2720000000000003e-06,
+      "loss": 0.0166,
+      "step": 2686
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.268e-06,
+      "loss": 0.0158,
+      "step": 2687
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.2640000000000003e-06,
+      "loss": 0.0142,
+      "step": 2688
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.26e-06,
+      "loss": 0.0271,
+      "step": 2689
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.256e-06,
+      "loss": 0.0326,
+      "step": 2690
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.2520000000000003e-06,
+      "loss": 0.0156,
+      "step": 2691
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.248e-06,
+      "loss": 0.0107,
+      "step": 2692
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.244e-06,
+      "loss": 0.0174,
+      "step": 2693
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 0.0189,
+      "step": 2694
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2360000000000001e-06,
+      "loss": 0.0229,
+      "step": 2695
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2320000000000002e-06,
+      "loss": 0.0143,
+      "step": 2696
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2280000000000001e-06,
+      "loss": 0.0212,
+      "step": 2697
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.224e-06,
+      "loss": 0.0167,
+      "step": 2698
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 0.0348,
+      "step": 2699
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.216e-06,
+      "loss": 0.0108,
+      "step": 2700
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2120000000000002e-06,
+      "loss": 0.0297,
+      "step": 2701
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.2080000000000001e-06,
+      "loss": 0.0245,
+      "step": 2702
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.204e-06,
+      "loss": 0.0162,
+      "step": 2703
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.0248,
+      "step": 2704
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.196e-06,
+      "loss": 0.0196,
+      "step": 2705
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.1920000000000002e-06,
+      "loss": 0.0213,
+      "step": 2706
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.188e-06,
+      "loss": 0.0206,
+      "step": 2707
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.1840000000000002e-06,
+      "loss": 0.0249,
+      "step": 2708
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.015,
+      "step": 2709
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.176e-06,
+      "loss": 0.0122,
+      "step": 2710
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1720000000000002e-06,
+      "loss": 0.0189,
+      "step": 2711
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.168e-06,
+      "loss": 0.0157,
+      "step": 2712
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1640000000000002e-06,
+      "loss": 0.0119,
+      "step": 2713
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 0.0233,
+      "step": 2714
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.156e-06,
+      "loss": 0.0136,
+      "step": 2715
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1520000000000002e-06,
+      "loss": 0.0096,
+      "step": 2716
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.148e-06,
+      "loss": 0.0307,
+      "step": 2717
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.1440000000000002e-06,
+      "loss": 0.0134,
+      "step": 2718
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.14e-06,
+      "loss": 0.0313,
+      "step": 2719
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.1360000000000002e-06,
+      "loss": 0.0072,
+      "step": 2720
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.1320000000000001e-06,
+      "loss": 0.0134,
+      "step": 2721
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.128e-06,
+      "loss": 0.0247,
+      "step": 2722
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.1240000000000002e-06,
+      "loss": 0.0231,
+      "step": 2723
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.12e-06,
+      "loss": 0.0234,
+      "step": 2724
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.1160000000000002e-06,
+      "loss": 0.0236,
+      "step": 2725
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.1120000000000001e-06,
+      "loss": 0.0124,
+      "step": 2726
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.108e-06,
+      "loss": 0.0146,
+      "step": 2727
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.1040000000000001e-06,
+      "loss": 0.0123,
+      "step": 2728
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.1e-06,
+      "loss": 0.0166,
+      "step": 2729
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0960000000000002e-06,
+      "loss": 0.0282,
+      "step": 2730
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.092e-06,
+      "loss": 0.0268,
+      "step": 2731
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.088e-06,
+      "loss": 0.0112,
+      "step": 2732
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0840000000000001e-06,
+      "loss": 0.0084,
+      "step": 2733
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.08e-06,
+      "loss": 0.0171,
+      "step": 2734
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0760000000000002e-06,
+      "loss": 0.0216,
+      "step": 2735
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.072e-06,
+      "loss": 0.0261,
+      "step": 2736
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0680000000000002e-06,
+      "loss": 0.0224,
+      "step": 2737
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.064e-06,
+      "loss": 0.0172,
+      "step": 2738
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.06e-06,
+      "loss": 0.0164,
+      "step": 2739
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0560000000000001e-06,
+      "loss": 0.0171,
+      "step": 2740
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.052e-06,
+      "loss": 0.0074,
+      "step": 2741
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0480000000000002e-06,
+      "loss": 0.0381,
+      "step": 2742
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.044e-06,
+      "loss": 0.0162,
+      "step": 2743
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.04e-06,
+      "loss": 0.0175,
+      "step": 2744
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0360000000000001e-06,
+      "loss": 0.0281,
+      "step": 2745
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.032e-06,
+      "loss": 0.0137,
+      "step": 2746
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0280000000000002e-06,
+      "loss": 0.0892,
+      "step": 2747
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.024e-06,
+      "loss": 0.0271,
+      "step": 2748
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.02e-06,
+      "loss": 0.0099,
+      "step": 2749
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.016e-06,
+      "loss": 0.0168,
+      "step": 2750
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.012e-06,
+      "loss": 0.0278,
+      "step": 2751
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0080000000000001e-06,
+      "loss": 0.0204,
+      "step": 2752
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.004e-06,
+      "loss": 0.025,
+      "step": 2753
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.0211,
+      "step": 2754
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 9.96e-07,
+      "loss": 0.0169,
+      "step": 2755
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.92e-07,
+      "loss": 0.0187,
+      "step": 2756
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.880000000000001e-07,
+      "loss": 0.0162,
+      "step": 2757
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.84e-07,
+      "loss": 0.0122,
+      "step": 2758
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.0221,
+      "step": 2759
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.76e-07,
+      "loss": 0.0115,
+      "step": 2760
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.72e-07,
+      "loss": 0.0158,
+      "step": 2761
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.68e-07,
+      "loss": 0.0222,
+      "step": 2762
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.64e-07,
+      "loss": 0.0606,
+      "step": 2763
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 0.0124,
+      "step": 2764
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.56e-07,
+      "loss": 0.0204,
+      "step": 2765
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.520000000000002e-07,
+      "loss": 0.0148,
+      "step": 2766
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.480000000000001e-07,
+      "loss": 0.0215,
+      "step": 2767
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.440000000000001e-07,
+      "loss": 0.013,
+      "step": 2768
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 0.0252,
+      "step": 2769
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.360000000000001e-07,
+      "loss": 0.0192,
+      "step": 2770
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.320000000000001e-07,
+      "loss": 0.0126,
+      "step": 2771
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.28e-07,
+      "loss": 0.0176,
+      "step": 2772
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.240000000000001e-07,
+      "loss": 0.0181,
+      "step": 2773
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 0.0141,
+      "step": 2774
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.160000000000001e-07,
+      "loss": 0.0113,
+      "step": 2775
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.120000000000001e-07,
+      "loss": 0.0058,
+      "step": 2776
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.080000000000001e-07,
+      "loss": 0.0159,
+      "step": 2777
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.04e-07,
+      "loss": 0.0457,
+      "step": 2778
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 0.018,
+      "step": 2779
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.960000000000001e-07,
+      "loss": 0.0126,
+      "step": 2780
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.920000000000001e-07,
+      "loss": 0.0107,
+      "step": 2781
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.880000000000001e-07,
+      "loss": 0.0223,
+      "step": 2782
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.840000000000001e-07,
+      "loss": 0.0267,
+      "step": 2783
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.8e-07,
+      "loss": 0.0238,
+      "step": 2784
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.760000000000001e-07,
+      "loss": 0.017,
+      "step": 2785
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.720000000000001e-07,
+      "loss": 0.014,
+      "step": 2786
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.680000000000001e-07,
+      "loss": 0.0246,
+      "step": 2787
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.640000000000001e-07,
+      "loss": 0.0179,
+      "step": 2788
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.6e-07,
+      "loss": 0.0115,
+      "step": 2789
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.56e-07,
+      "loss": 0.0274,
+      "step": 2790
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.520000000000001e-07,
+      "loss": 0.0175,
+      "step": 2791
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.480000000000001e-07,
+      "loss": 0.0172,
+      "step": 2792
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.440000000000001e-07,
+      "loss": 0.0102,
+      "step": 2793
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 0.0148,
+      "step": 2794
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.36e-07,
+      "loss": 0.0083,
+      "step": 2795
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.32e-07,
+      "loss": 0.0248,
+      "step": 2796
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.280000000000001e-07,
+      "loss": 0.0207,
+      "step": 2797
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.240000000000001e-07,
+      "loss": 0.0269,
+      "step": 2798
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 0.0261,
+      "step": 2799
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.160000000000001e-07,
+      "loss": 0.0124,
+      "step": 2800
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.12e-07,
+      "loss": 0.0249,
+      "step": 2801
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.08e-07,
+      "loss": 0.0116,
+      "step": 2802
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.04e-07,
+      "loss": 0.0186,
+      "step": 2803
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.0166,
+      "step": 2804
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.960000000000001e-07,
+      "loss": 0.027,
+      "step": 2805
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.920000000000001e-07,
+      "loss": 0.0269,
+      "step": 2806
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.88e-07,
+      "loss": 0.0124,
+      "step": 2807
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.84e-07,
+      "loss": 0.0134,
+      "step": 2808
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.8e-07,
+      "loss": 0.0128,
+      "step": 2809
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.760000000000001e-07,
+      "loss": 0.0199,
+      "step": 2810
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.720000000000001e-07,
+      "loss": 0.0325,
+      "step": 2811
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.68e-07,
+      "loss": 0.014,
+      "step": 2812
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.64e-07,
+      "loss": 0.0203,
+      "step": 2813
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.6e-07,
+      "loss": 0.0375,
+      "step": 2814
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.56e-07,
+      "loss": 0.0259,
+      "step": 2815
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.520000000000001e-07,
+      "loss": 0.0205,
+      "step": 2816
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.480000000000001e-07,
+      "loss": 0.0142,
+      "step": 2817
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.44e-07,
+      "loss": 0.0243,
+      "step": 2818
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.4e-07,
+      "loss": 0.0162,
+      "step": 2819
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.36e-07,
+      "loss": 0.0116,
+      "step": 2820
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.32e-07,
+      "loss": 0.014,
+      "step": 2821
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.280000000000001e-07,
+      "loss": 0.0154,
+      "step": 2822
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.240000000000001e-07,
+      "loss": 0.0173,
+      "step": 2823
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.2e-07,
+      "loss": 0.0134,
+      "step": 2824
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.16e-07,
+      "loss": 0.0107,
+      "step": 2825
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 7.12e-07,
+      "loss": 0.0086,
+      "step": 2826
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 7.08e-07,
+      "loss": 0.0121,
+      "step": 2827
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 7.040000000000001e-07,
+      "loss": 0.0232,
+      "step": 2828
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 0.0484,
+      "step": 2829
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.96e-07,
+      "loss": 0.0126,
+      "step": 2830
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.92e-07,
+      "loss": 0.0183,
+      "step": 2831
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.88e-07,
+      "loss": 0.0093,
+      "step": 2832
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.84e-07,
+      "loss": 0.0133,
+      "step": 2833
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 0.0102,
+      "step": 2834
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.76e-07,
+      "loss": 0.0508,
+      "step": 2835
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.72e-07,
+      "loss": 0.0274,
+      "step": 2836
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.68e-07,
+      "loss": 0.0188,
+      "step": 2837
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.64e-07,
+      "loss": 0.0077,
+      "step": 2838
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.6e-07,
+      "loss": 0.0405,
+      "step": 2839
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.560000000000002e-07,
+      "loss": 0.0285,
+      "step": 2840
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.52e-07,
+      "loss": 0.0093,
+      "step": 2841
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.48e-07,
+      "loss": 0.0414,
+      "step": 2842
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.44e-07,
+      "loss": 0.0124,
+      "step": 2843
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.4e-07,
+      "loss": 0.0195,
+      "step": 2844
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.360000000000001e-07,
+      "loss": 0.0098,
+      "step": 2845
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.320000000000002e-07,
+      "loss": 0.0316,
+      "step": 2846
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.28e-07,
+      "loss": 0.014,
+      "step": 2847
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.24e-07,
+      "loss": 0.0336,
+      "step": 2848
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 0.0276,
+      "step": 2849
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.160000000000001e-07,
+      "loss": 0.0285,
+      "step": 2850
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.12e-07,
+      "loss": 0.018,
+      "step": 2851
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.08e-07,
+      "loss": 0.0252,
+      "step": 2852
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.040000000000001e-07,
+      "loss": 0.0189,
+      "step": 2853
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 0.0128,
+      "step": 2854
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.960000000000001e-07,
+      "loss": 0.0068,
+      "step": 2855
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.920000000000001e-07,
+      "loss": 0.0304,
+      "step": 2856
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.88e-07,
+      "loss": 0.0301,
+      "step": 2857
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.84e-07,
+      "loss": 0.0245,
+      "step": 2858
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.0124,
+      "step": 2859
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.760000000000001e-07,
+      "loss": 0.0116,
+      "step": 2860
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.720000000000001e-07,
+      "loss": 0.0325,
+      "step": 2861
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.680000000000001e-07,
+      "loss": 0.0133,
+      "step": 2862
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.64e-07,
+      "loss": 0.0096,
+      "step": 2863
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.6e-07,
+      "loss": 0.0431,
+      "step": 2864
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.560000000000001e-07,
+      "loss": 0.012,
+      "step": 2865
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.520000000000001e-07,
+      "loss": 0.008,
+      "step": 2866
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.480000000000001e-07,
+      "loss": 0.0085,
+      "step": 2867
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.44e-07,
+      "loss": 0.0153,
+      "step": 2868
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.4e-07,
+      "loss": 0.0366,
+      "step": 2869
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.36e-07,
+      "loss": 0.0433,
+      "step": 2870
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.32e-07,
+      "loss": 0.0144,
+      "step": 2871
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.280000000000001e-07,
+      "loss": 0.0316,
+      "step": 2872
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.240000000000001e-07,
+      "loss": 0.0228,
+      "step": 2873
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.2e-07,
+      "loss": 0.0257,
+      "step": 2874
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.16e-07,
+      "loss": 0.0184,
+      "step": 2875
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.12e-07,
+      "loss": 0.0132,
+      "step": 2876
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.08e-07,
+      "loss": 0.0278,
+      "step": 2877
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.040000000000001e-07,
+      "loss": 0.0178,
+      "step": 2878
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.0278,
+      "step": 2879
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.96e-07,
+      "loss": 0.0351,
+      "step": 2880
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.92e-07,
+      "loss": 0.0442,
+      "step": 2881
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.88e-07,
+      "loss": 0.0196,
+      "step": 2882
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.84e-07,
+      "loss": 0.0228,
+      "step": 2883
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 0.0213,
+      "step": 2884
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.760000000000001e-07,
+      "loss": 0.0226,
+      "step": 2885
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.7200000000000004e-07,
+      "loss": 0.0145,
+      "step": 2886
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6800000000000006e-07,
+      "loss": 0.0132,
+      "step": 2887
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.64e-07,
+      "loss": 0.0534,
+      "step": 2888
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 0.014,
+      "step": 2889
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5600000000000006e-07,
+      "loss": 0.0189,
+      "step": 2890
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.52e-07,
+      "loss": 0.0201,
+      "step": 2891
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.4800000000000004e-07,
+      "loss": 0.019,
+      "step": 2892
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.4400000000000006e-07,
+      "loss": 0.0417,
+      "step": 2893
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.4e-07,
+      "loss": 0.0274,
+      "step": 2894
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.3600000000000004e-07,
+      "loss": 0.0113,
+      "step": 2895
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.3200000000000006e-07,
+      "loss": 0.0194,
+      "step": 2896
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.28e-07,
+      "loss": 0.0159,
+      "step": 2897
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.2400000000000004e-07,
+      "loss": 0.0114,
+      "step": 2898
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 0.0168,
+      "step": 2899
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.16e-07,
+      "loss": 0.0235,
+      "step": 2900
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.1200000000000004e-07,
+      "loss": 0.0558,
+      "step": 2901
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.0800000000000005e-07,
+      "loss": 0.0215,
+      "step": 2902
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.04e-07,
+      "loss": 0.0097,
+      "step": 2903
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.0219,
+      "step": 2904
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.9600000000000005e-07,
+      "loss": 0.018,
+      "step": 2905
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.92e-07,
+      "loss": 0.0211,
+      "step": 2906
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.8800000000000003e-07,
+      "loss": 0.0167,
+      "step": 2907
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.84e-07,
+      "loss": 0.0132,
+      "step": 2908
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.8e-07,
+      "loss": 0.012,
+      "step": 2909
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.7600000000000003e-07,
+      "loss": 0.0199,
+      "step": 2910
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.72e-07,
+      "loss": 0.0077,
+      "step": 2911
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.68e-07,
+      "loss": 0.0163,
+      "step": 2912
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.6400000000000003e-07,
+      "loss": 0.0261,
+      "step": 2913
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.6e-07,
+      "loss": 0.0119,
+      "step": 2914
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.56e-07,
+      "loss": 0.0223,
+      "step": 2915
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.5200000000000003e-07,
+      "loss": 0.0134,
+      "step": 2916
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.48e-07,
+      "loss": 0.0172,
+      "step": 2917
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.44e-07,
+      "loss": 0.0108,
+      "step": 2918
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 0.017,
+      "step": 2919
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.36e-07,
+      "loss": 0.0173,
+      "step": 2920
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.32e-07,
+      "loss": 0.0221,
+      "step": 2921
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.280000000000001e-07,
+      "loss": 0.0196,
+      "step": 2922
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.24e-07,
+      "loss": 0.0137,
+      "step": 2923
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.2e-07,
+      "loss": 0.0344,
+      "step": 2924
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.160000000000001e-07,
+      "loss": 0.0167,
+      "step": 2925
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.12e-07,
+      "loss": 0.016,
+      "step": 2926
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.0800000000000006e-07,
+      "loss": 0.0151,
+      "step": 2927
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.04e-07,
+      "loss": 0.0101,
+      "step": 2928
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 0.0295,
+      "step": 2929
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9600000000000006e-07,
+      "loss": 0.01,
+      "step": 2930
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.92e-07,
+      "loss": 0.0213,
+      "step": 2931
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8800000000000004e-07,
+      "loss": 0.0113,
+      "step": 2932
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8400000000000005e-07,
+      "loss": 0.0133,
+      "step": 2933
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8e-07,
+      "loss": 0.0451,
+      "step": 2934
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7600000000000004e-07,
+      "loss": 0.0193,
+      "step": 2935
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.72e-07,
+      "loss": 0.0129,
+      "step": 2936
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.68e-07,
+      "loss": 0.0169,
+      "step": 2937
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.6400000000000003e-07,
+      "loss": 0.0124,
+      "step": 2938
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.6e-07,
+      "loss": 0.0533,
+      "step": 2939
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.56e-07,
+      "loss": 0.0217,
+      "step": 2940
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.5200000000000003e-07,
+      "loss": 0.0133,
+      "step": 2941
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.48e-07,
+      "loss": 0.0279,
+      "step": 2942
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.44e-07,
+      "loss": 0.0319,
+      "step": 2943
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 0.0113,
+      "step": 2944
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.3600000000000002e-07,
+      "loss": 0.0183,
+      "step": 2945
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.32e-07,
+      "loss": 0.0222,
+      "step": 2946
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.2800000000000003e-07,
+      "loss": 0.0171,
+      "step": 2947
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.2400000000000002e-07,
+      "loss": 0.0156,
+      "step": 2948
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.2e-07,
+      "loss": 0.0139,
+      "step": 2949
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.1600000000000003e-07,
+      "loss": 0.0187,
+      "step": 2950
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.1200000000000002e-07,
+      "loss": 0.0369,
+      "step": 2951
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.08e-07,
+      "loss": 0.0255,
+      "step": 2952
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.0400000000000003e-07,
+      "loss": 0.0158,
+      "step": 2953
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 0.0109,
+      "step": 2954
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.96e-07,
+      "loss": 0.0136,
+      "step": 2955
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.92e-07,
+      "loss": 0.0202,
+      "step": 2956
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8800000000000002e-07,
+      "loss": 0.0219,
+      "step": 2957
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.84e-07,
+      "loss": 0.0071,
+      "step": 2958
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8e-07,
+      "loss": 0.0184,
+      "step": 2959
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7600000000000001e-07,
+      "loss": 0.0269,
+      "step": 2960
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.72e-07,
+      "loss": 0.0128,
+      "step": 2961
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.68e-07,
+      "loss": 0.0141,
+      "step": 2962
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.6400000000000004e-07,
+      "loss": 0.0101,
+      "step": 2963
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.6e-07,
+      "loss": 0.0279,
+      "step": 2964
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.56e-07,
+      "loss": 0.0173,
+      "step": 2965
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.52e-07,
+      "loss": 0.0162,
+      "step": 2966
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4800000000000003e-07,
+      "loss": 0.0113,
+      "step": 2967
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4400000000000002e-07,
+      "loss": 0.0164,
+      "step": 2968
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4e-07,
+      "loss": 0.0156,
+      "step": 2969
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.36e-07,
+      "loss": 0.015,
+      "step": 2970
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3200000000000002e-07,
+      "loss": 0.04,
+      "step": 2971
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.28e-07,
+      "loss": 0.0109,
+      "step": 2972
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.24e-07,
+      "loss": 0.0213,
+      "step": 2973
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 0.0237,
+      "step": 2974
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.16e-07,
+      "loss": 0.0111,
+      "step": 2975
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1200000000000001e-07,
+      "loss": 0.0191,
+      "step": 2976
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.0800000000000001e-07,
+      "loss": 0.0181,
+      "step": 2977
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.04e-07,
+      "loss": 0.0134,
+      "step": 2978
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 0.0207,
+      "step": 2979
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 9.6e-08,
+      "loss": 0.0148,
+      "step": 2980
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 9.2e-08,
+      "loss": 0.0171,
+      "step": 2981
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.800000000000001e-08,
+      "loss": 0.0132,
+      "step": 2982
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.4e-08,
+      "loss": 0.0144,
+      "step": 2983
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8e-08,
+      "loss": 0.0283,
+      "step": 2984
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.6e-08,
+      "loss": 0.0097,
+      "step": 2985
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.200000000000001e-08,
+      "loss": 0.0208,
+      "step": 2986
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.8e-08,
+      "loss": 0.0149,
+      "step": 2987
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.4e-08,
+      "loss": 0.0234,
+      "step": 2988
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 0.011,
+      "step": 2989
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.6000000000000005e-08,
+      "loss": 0.0175,
+      "step": 2990
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.2e-08,
+      "loss": 0.016,
+      "step": 2991
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.8e-08,
+      "loss": 0.0214,
+      "step": 2992
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.4000000000000004e-08,
+      "loss": 0.0138,
+      "step": 2993
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4e-08,
+      "loss": 0.0276,
+      "step": 2994
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6000000000000005e-08,
+      "loss": 0.0151,
+      "step": 2995
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.2e-08,
+      "loss": 0.0339,
+      "step": 2996
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8000000000000003e-08,
+      "loss": 0.0125,
+      "step": 2997
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4e-08,
+      "loss": 0.0192,
+      "step": 2998
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-08,
+      "loss": 0.0195,
+      "step": 2999
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6e-08,
+      "loss": 0.0233,
+      "step": 3000
+    },
+    {
+      "epoch": 3.88,
+      "eval_loss": 0.2973702549934387,
+      "eval_runtime": 1871.2299,
+      "eval_samples_per_second": 2.709,
+      "eval_steps_per_second": 0.169,
+      "eval_wer": 19.604205318491033,
+      "step": 3000
+    }
+  ],
+  "max_steps": 3000,
+  "num_train_epochs": 4,
+  "total_flos": 1.384517316575232e+19,
+  "trial_name": null,
+  "trial_params": null
+}