diff --git "a/checkpoint-1000/trainer_state.json" "b/checkpoint-1000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1000/trainer_state.json"
@@ -0,0 +1,6025 @@
+{
+  "best_metric": 21.42453819472453,
+  "best_model_checkpoint": "./whisper-small-swedish-Test-3000/checkpoint-1000",
+  "epoch": 1.2936610608020698,
+  "global_step": 1000,
+  "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
+    }
+  ],
+  "max_steps": 3000,
+  "num_train_epochs": 4,
+  "total_flos": 4.61505772191744e+18,
+  "trial_name": null,
+  "trial_params": null
+}