diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,51155 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999266571809954,
+  "eval_steps": 20000,
+  "global_step": 8521,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8125e-08,
+      "loss": 5.1661,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5625e-07,
+      "loss": 5.1711,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3437500000000003e-07,
+      "loss": 5.0582,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.125e-07,
+      "loss": 5.4139,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.90625e-07,
+      "loss": 5.4144,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6875000000000006e-07,
+      "loss": 5.6606,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.468750000000001e-07,
+      "loss": 4.9452,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.25e-07,
+      "loss": 5.2148,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.03125e-07,
+      "loss": 5.5617,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8125e-07,
+      "loss": 5.2777,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.59375e-07,
+      "loss": 5.1849,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.375000000000001e-07,
+      "loss": 5.4469,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0156250000000001e-06,
+      "loss": 5.1032,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0937500000000001e-06,
+      "loss": 4.7811,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1718750000000001e-06,
+      "loss": 4.7525,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-06,
+      "loss": 4.775,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.328125e-06,
+      "loss": 4.5264,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.40625e-06,
+      "loss": 4.0926,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.484375e-06,
+      "loss": 4.1978,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5625e-06,
+      "loss": 3.8619,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.640625e-06,
+      "loss": 3.3229,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.71875e-06,
+      "loss": 2.9258,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.796875e-06,
+      "loss": 3.1048,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8750000000000003e-06,
+      "loss": 2.8253,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.953125e-06,
+      "loss": 2.2981,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0312500000000002e-06,
+      "loss": 2.2149,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.109375e-06,
+      "loss": 2.0256,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1875000000000002e-06,
+      "loss": 1.7812,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.265625e-06,
+      "loss": 1.4904,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3437500000000002e-06,
+      "loss": 1.4741,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.421875e-06,
+      "loss": 1.3415,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 0.9914,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5781250000000004e-06,
+      "loss": 0.9039,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.65625e-06,
+      "loss": 0.6356,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7343750000000004e-06,
+      "loss": 0.5912,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8125e-06,
+      "loss": 0.4788,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8906250000000004e-06,
+      "loss": 0.3831,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.96875e-06,
+      "loss": 0.4101,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0468750000000004e-06,
+      "loss": 0.3264,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.125e-06,
+      "loss": 0.3691,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2031250000000004e-06,
+      "loss": 0.3132,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.28125e-06,
+      "loss": 0.2518,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3593750000000003e-06,
+      "loss": 0.2709,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4375e-06,
+      "loss": 0.2348,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5156250000000003e-06,
+      "loss": 0.2126,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.59375e-06,
+      "loss": 0.3059,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6718750000000003e-06,
+      "loss": 0.2578,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.3424,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.828125000000001e-06,
+      "loss": 0.2334,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.90625e-06,
+      "loss": 0.2578,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.984375e-06,
+      "loss": 0.2169,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0625000000000005e-06,
+      "loss": 0.2544,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.140625000000001e-06,
+      "loss": 0.2406,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.21875e-06,
+      "loss": 0.2092,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.296875e-06,
+      "loss": 0.1697,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3750000000000005e-06,
+      "loss": 0.2425,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.453125000000001e-06,
+      "loss": 0.1918,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.53125e-06,
+      "loss": 0.29,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.609375e-06,
+      "loss": 0.2139,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 0.2551,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.765625000000001e-06,
+      "loss": 0.4777,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.84375e-06,
+      "loss": 0.2221,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.921875e-06,
+      "loss": 0.1768,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.1522,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.078125000000001e-06,
+      "loss": 0.2115,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.156250000000001e-06,
+      "loss": 0.2131,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.234375e-06,
+      "loss": 0.1978,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3125e-06,
+      "loss": 0.129,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.390625000000001e-06,
+      "loss": 0.1279,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.468750000000001e-06,
+      "loss": 0.2358,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.546875e-06,
+      "loss": 0.176,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.625e-06,
+      "loss": 0.1871,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7031250000000006e-06,
+      "loss": 0.1351,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.781250000000001e-06,
+      "loss": 0.1235,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.859375e-06,
+      "loss": 0.1984,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9375e-06,
+      "loss": 0.1383,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.0156250000000005e-06,
+      "loss": 0.1985,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.093750000000001e-06,
+      "loss": 0.3175,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.171875e-06,
+      "loss": 0.1849,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.25e-06,
+      "loss": 0.1422,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.3281250000000005e-06,
+      "loss": 0.1974,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.406250000000001e-06,
+      "loss": 0.168,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.484375000000001e-06,
+      "loss": 0.1415,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5625e-06,
+      "loss": 0.1816,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6406250000000005e-06,
+      "loss": 0.1021,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.718750000000001e-06,
+      "loss": 0.1749,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.796875000000001e-06,
+      "loss": 0.1395,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.875e-06,
+      "loss": 0.1334,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.9531250000000004e-06,
+      "loss": 0.2039,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.031250000000001e-06,
+      "loss": 0.1982,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.109375000000001e-06,
+      "loss": 0.2098,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1875e-06,
+      "loss": 0.144,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.265625e-06,
+      "loss": 0.1527,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.343750000000001e-06,
+      "loss": 0.1825,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.421875000000001e-06,
+      "loss": 0.1284,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.1041,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.578125e-06,
+      "loss": 0.2881,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.656250000000001e-06,
+      "loss": 0.1525,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.734375e-06,
+      "loss": 0.2714,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.8125e-06,
+      "loss": 0.2018,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.890625e-06,
+      "loss": 0.1502,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.96875e-06,
+      "loss": 0.1931,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.046875e-06,
+      "loss": 0.2113,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 0.1148,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.203125000000001e-06,
+      "loss": 0.1344,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.281250000000001e-06,
+      "loss": 0.1616,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.359375e-06,
+      "loss": 0.1898,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.4375e-06,
+      "loss": 0.1047,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.515625e-06,
+      "loss": 0.136,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.59375e-06,
+      "loss": 0.0986,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.671875e-06,
+      "loss": 0.2556,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 0.2162,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.828125000000001e-06,
+      "loss": 0.1387,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.906250000000001e-06,
+      "loss": 0.2065,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.984375000000002e-06,
+      "loss": 0.153,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.0625e-06,
+      "loss": 0.1532,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.140625e-06,
+      "loss": 0.0836,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.21875e-06,
+      "loss": 0.1208,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.296875e-06,
+      "loss": 0.1737,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 0.0687,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.453125000000001e-06,
+      "loss": 0.1087,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.531250000000001e-06,
+      "loss": 0.1218,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.609375000000001e-06,
+      "loss": 0.1238,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.6875e-06,
+      "loss": 0.1471,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.765625e-06,
+      "loss": 0.1937,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.84375e-06,
+      "loss": 0.2338,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.921875e-06,
+      "loss": 0.1941,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.189,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0078125000000001e-05,
+      "loss": 0.2974,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0156250000000001e-05,
+      "loss": 0.25,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0234375000000001e-05,
+      "loss": 0.4139,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0312500000000002e-05,
+      "loss": 0.093,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0390625e-05,
+      "loss": 0.223,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.046875e-05,
+      "loss": 0.1608,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0546875e-05,
+      "loss": 0.3657,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0625e-05,
+      "loss": 0.2213,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0703125000000001e-05,
+      "loss": 0.1636,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0781250000000001e-05,
+      "loss": 0.1398,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0859375000000001e-05,
+      "loss": 0.2118,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0937500000000002e-05,
+      "loss": 0.098,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1015625e-05,
+      "loss": 0.0647,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.109375e-05,
+      "loss": 0.1525,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1171875e-05,
+      "loss": 0.1794,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.125e-05,
+      "loss": 0.2174,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1328125000000001e-05,
+      "loss": 0.1515,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1406250000000001e-05,
+      "loss": 0.1093,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1484375000000001e-05,
+      "loss": 0.0997,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1562500000000002e-05,
+      "loss": 0.1733,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1640625000000002e-05,
+      "loss": 0.1806,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.171875e-05,
+      "loss": 0.2219,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1796875e-05,
+      "loss": 0.1371,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1875e-05,
+      "loss": 0.0783,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1953125000000001e-05,
+      "loss": 0.2568,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2031250000000001e-05,
+      "loss": 0.0582,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2109375000000001e-05,
+      "loss": 0.0853,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2187500000000001e-05,
+      "loss": 0.0721,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2265625000000002e-05,
+      "loss": 0.192,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.234375e-05,
+      "loss": 0.0739,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2421875e-05,
+      "loss": 0.1692,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0703,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2578125e-05,
+      "loss": 0.1524,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2656250000000001e-05,
+      "loss": 0.1098,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2734375000000001e-05,
+      "loss": 0.1759,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2812500000000001e-05,
+      "loss": 0.0505,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2890625000000002e-05,
+      "loss": 0.2064,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2968750000000002e-05,
+      "loss": 0.0404,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3046875e-05,
+      "loss": 0.1686,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3125e-05,
+      "loss": 0.123,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3203125e-05,
+      "loss": 0.1752,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3281250000000001e-05,
+      "loss": 0.1819,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3359375000000001e-05,
+      "loss": 0.2658,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3437500000000001e-05,
+      "loss": 0.1655,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3515625000000002e-05,
+      "loss": 0.1523,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3593750000000002e-05,
+      "loss": 0.1736,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3671875e-05,
+      "loss": 0.069,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.375e-05,
+      "loss": 0.1327,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3828125e-05,
+      "loss": 0.0923,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3906250000000001e-05,
+      "loss": 0.1089,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3984375000000001e-05,
+      "loss": 0.0903,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4062500000000001e-05,
+      "loss": 0.1081,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4140625000000002e-05,
+      "loss": 0.0935,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4218750000000002e-05,
+      "loss": 0.0879,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4296875000000002e-05,
+      "loss": 0.0748,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4375e-05,
+      "loss": 0.0996,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4453125e-05,
+      "loss": 0.2156,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.453125e-05,
+      "loss": 0.1647,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4609375000000001e-05,
+      "loss": 0.1136,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4687500000000001e-05,
+      "loss": 0.24,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4765625000000001e-05,
+      "loss": 0.4129,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4843750000000002e-05,
+      "loss": 0.2836,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4921875000000002e-05,
+      "loss": 0.0976,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.0796,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5078125e-05,
+      "loss": 0.1061,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.515625e-05,
+      "loss": 0.1227,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5234375000000001e-05,
+      "loss": 0.2164,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5312500000000003e-05,
+      "loss": 0.2468,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5390625e-05,
+      "loss": 0.1233,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.546875e-05,
+      "loss": 0.1624,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5546875e-05,
+      "loss": 0.1205,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5625e-05,
+      "loss": 0.1384,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5703125e-05,
+      "loss": 0.0596,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.578125e-05,
+      "loss": 0.0293,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5859375e-05,
+      "loss": 0.2201,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.59375e-05,
+      "loss": 0.1698,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6015625e-05,
+      "loss": 0.0844,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.609375e-05,
+      "loss": 0.0711,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6171875000000002e-05,
+      "loss": 0.1594,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 0.0653,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6328125000000002e-05,
+      "loss": 0.0948,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6406250000000002e-05,
+      "loss": 0.2328,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6484375000000003e-05,
+      "loss": 0.123,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6562500000000003e-05,
+      "loss": 0.1349,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6640625000000003e-05,
+      "loss": 0.1757,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.671875e-05,
+      "loss": 0.0995,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6796875e-05,
+      "loss": 0.0672,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6875e-05,
+      "loss": 0.1238,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6953125e-05,
+      "loss": 0.1398,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.703125e-05,
+      "loss": 0.0635,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7109375e-05,
+      "loss": 0.1241,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.71875e-05,
+      "loss": 0.188,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7265625e-05,
+      "loss": 0.0598,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.734375e-05,
+      "loss": 0.0744,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7421875e-05,
+      "loss": 0.1128,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.0406,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7578125000000002e-05,
+      "loss": 0.1623,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7656250000000002e-05,
+      "loss": 0.1335,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7734375000000002e-05,
+      "loss": 0.1786,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7812500000000003e-05,
+      "loss": 0.0305,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7890625000000003e-05,
+      "loss": 0.26,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7968750000000003e-05,
+      "loss": 0.0459,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8046875e-05,
+      "loss": 0.0273,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8125e-05,
+      "loss": 0.2942,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8203125e-05,
+      "loss": 0.1223,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.828125e-05,
+      "loss": 0.288,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8359375e-05,
+      "loss": 0.1668,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.84375e-05,
+      "loss": 0.0417,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8515625e-05,
+      "loss": 0.1054,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.859375e-05,
+      "loss": 0.105,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8671875e-05,
+      "loss": 0.1402,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.0449,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8828125000000002e-05,
+      "loss": 0.1129,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8906250000000002e-05,
+      "loss": 0.0843,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8984375000000002e-05,
+      "loss": 0.1905,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9062500000000003e-05,
+      "loss": 0.0958,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9140625000000003e-05,
+      "loss": 0.1101,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9218750000000003e-05,
+      "loss": 0.0758,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9296875000000003e-05,
+      "loss": 0.0817,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9375e-05,
+      "loss": 0.1035,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9453125e-05,
+      "loss": 0.0673,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.953125e-05,
+      "loss": 0.2356,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9609375e-05,
+      "loss": 0.0545,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.96875e-05,
+      "loss": 0.068,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9765625e-05,
+      "loss": 0.1083,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.984375e-05,
+      "loss": 0.1186,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9921875e-05,
+      "loss": 0.0246,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999277589538e-05,
+      "loss": 0.1634,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999997110358246e-05,
+      "loss": 0.1234,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999993498306443e-05,
+      "loss": 0.0827,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999988441434648e-05,
+      "loss": 0.2709,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999981939743593e-05,
+      "loss": 0.2479,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997399323422e-05,
+      "loss": 0.0683,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996460190767e-05,
+      "loss": 0.1409,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999953765765306e-05,
+      "loss": 0.1417,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999941484808693e-05,
+      "loss": 0.133,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999927759039606e-05,
+      "loss": 0.2035,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999912588460024e-05,
+      "loss": 0.0824,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999989597307214e-05,
+      "loss": 0.0864,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999987791287836e-05,
+      "loss": 0.1627,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999858407881282e-05,
+      "loss": 0.1106,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999837458083738e-05,
+      "loss": 0.2007,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999815063488747e-05,
+      "loss": 0.0957,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999791224099544e-05,
+      "loss": 0.1925,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999765939919575e-05,
+      "loss": 0.12,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999739210952493e-05,
+      "loss": 0.0367,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999711037202162e-05,
+      "loss": 0.1353,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999968141867265e-05,
+      "loss": 0.0727,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999965035536824e-05,
+      "loss": 0.1366,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999617847293414e-05,
+      "loss": 0.1649,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999583894452875e-05,
+      "loss": 0.1474,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999548496851524e-05,
+      "loss": 0.0764,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999511654494475e-05,
+      "loss": 0.0175,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999473367387056e-05,
+      "loss": 0.0817,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999433635534794e-05,
+      "loss": 0.0682,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999392458943432e-05,
+      "loss": 0.1109,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999934983761892e-05,
+      "loss": 0.1524,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999930577156741e-05,
+      "loss": 0.0786,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999926026079528e-05,
+      "loss": 0.0749,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999213305309094e-05,
+      "loss": 0.0944,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999916490511564e-05,
+      "loss": 0.08,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999911506022192e-05,
+      "loss": 0.0583,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999063770635117e-05,
+      "loss": 0.154,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999901103636266e-05,
+      "loss": 0.056,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999895685741216e-05,
+      "loss": 0.0575,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998901233791443e-05,
+      "loss": 0.035,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999884416550855e-05,
+      "loss": 0.055,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999878565257172e-05,
+      "loss": 0.1218,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998725694989416e-05,
+      "loss": 0.056,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998664292770294e-05,
+      "loss": 0.0088,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998601445923227e-05,
+      "loss": 0.0054,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998537154457298e-05,
+      "loss": 0.2185,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999847141838179e-05,
+      "loss": 0.0822,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999840423770621e-05,
+      "loss": 0.0234,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998335612440254e-05,
+      "loss": 0.0191,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998265542593845e-05,
+      "loss": 0.2999,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998194028177105e-05,
+      "loss": 0.1184,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998121069200364e-05,
+      "loss": 0.2558,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998046665674164e-05,
+      "loss": 0.1622,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997970817609254e-05,
+      "loss": 0.1003,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997893525016596e-05,
+      "loss": 0.1106,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997814787907356e-05,
+      "loss": 0.0688,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997734606292912e-05,
+      "loss": 0.1008,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999765298018484e-05,
+      "loss": 0.1216,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997569909594948e-05,
+      "loss": 0.1739,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997485394535226e-05,
+      "loss": 0.2076,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997399435017893e-05,
+      "loss": 0.0716,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999731203105536e-05,
+      "loss": 0.1751,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997223182660265e-05,
+      "loss": 0.0992,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997132889845438e-05,
+      "loss": 0.0422,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997041152623927e-05,
+      "loss": 0.033,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996947971008986e-05,
+      "loss": 0.141,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996853345014076e-05,
+      "loss": 0.069,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996757274652876e-05,
+      "loss": 0.109,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999665975993926e-05,
+      "loss": 0.0143,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996560800887316e-05,
+      "loss": 0.1592,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996460397511347e-05,
+      "loss": 0.0282,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996358549825856e-05,
+      "loss": 0.0321,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999625525784556e-05,
+      "loss": 0.1808,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996150521585382e-05,
+      "loss": 0.0408,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996044341060456e-05,
+      "loss": 0.1066,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999593671628612e-05,
+      "loss": 0.1287,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995827647277926e-05,
+      "loss": 0.0292,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999571713405163e-05,
+      "loss": 0.0731,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995605176623204e-05,
+      "loss": 0.1373,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999549177500882e-05,
+      "loss": 0.0854,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995376929224863e-05,
+      "loss": 0.0426,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999526063928793e-05,
+      "loss": 0.0833,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995142905214816e-05,
+      "loss": 0.1431,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995023727022538e-05,
+      "loss": 0.1064,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994903104728308e-05,
+      "loss": 0.0405,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999478103834956e-05,
+      "loss": 0.0652,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994657527903932e-05,
+      "loss": 0.0702,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999453257340926e-05,
+      "loss": 0.0721,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999440617488361e-05,
+      "loss": 0.0274,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999427833234523e-05,
+      "loss": 0.1046,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99941490458126e-05,
+      "loss": 0.0182,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99940183153044e-05,
+      "loss": 0.0975,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993886140839516e-05,
+      "loss": 0.1907,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993752522437044e-05,
+      "loss": 0.0162,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999361746011629e-05,
+      "loss": 0.0834,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999348095389677e-05,
+      "loss": 0.0758,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993343003798206e-05,
+      "loss": 0.0556,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999320360984053e-05,
+      "loss": 0.1594,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993062772043873e-05,
+      "loss": 0.0561,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992920490428598e-05,
+      "loss": 0.1235,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992776765015252e-05,
+      "loss": 0.1266,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999263159582461e-05,
+      "loss": 0.051,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992484982877633e-05,
+      "loss": 0.0603,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992336926195515e-05,
+      "loss": 0.0275,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992187425799643e-05,
+      "loss": 0.1097,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992036481711618e-05,
+      "loss": 0.0246,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991884093953247e-05,
+      "loss": 0.1624,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991730262546548e-05,
+      "loss": 0.0454,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999157498751375e-05,
+      "loss": 0.047,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991418268877286e-05,
+      "loss": 0.1681,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991260106659798e-05,
+      "loss": 0.0696,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991100500884137e-05,
+      "loss": 0.2095,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990939451573365e-05,
+      "loss": 0.0424,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999077695875075e-05,
+      "loss": 0.0068,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999061302243977e-05,
+      "loss": 0.0759,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990447642664107e-05,
+      "loss": 0.0973,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990280819447662e-05,
+      "loss": 0.0206,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999011255281453e-05,
+      "loss": 0.0922,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998994284278903e-05,
+      "loss": 0.0069,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989771689395678e-05,
+      "loss": 0.0028,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989599092659207e-05,
+      "loss": 0.073,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989425052604545e-05,
+      "loss": 0.0906,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989249569256846e-05,
+      "loss": 0.089,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989072642641463e-05,
+      "loss": 0.1174,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988894272783957e-05,
+      "loss": 0.0511,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99887144597101e-05,
+      "loss": 0.0454,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998853320344587e-05,
+      "loss": 0.0999,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988350504017455e-05,
+      "loss": 0.1237,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988166361451256e-05,
+      "loss": 0.0875,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987980775773876e-05,
+      "loss": 0.0335,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987793747012125e-05,
+      "loss": 0.0837,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998760527519303e-05,
+      "loss": 0.1084,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987415360343823e-05,
+      "loss": 0.0425,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998722400249194e-05,
+      "loss": 0.0607,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998703120166503e-05,
+      "loss": 0.1076,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986836957890947e-05,
+      "loss": 0.009,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998664127119776e-05,
+      "loss": 0.0617,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998644414161374e-05,
+      "loss": 0.0918,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998624556916737e-05,
+      "loss": 0.0229,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986045553887334e-05,
+      "loss": 0.062,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985844095802538e-05,
+      "loss": 0.0237,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985641194942088e-05,
+      "loss": 0.091,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985436851335294e-05,
+      "loss": 0.0214,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985231065011687e-05,
+      "loss": 0.0564,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985023836000996e-05,
+      "loss": 0.1525,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984815164333163e-05,
+      "loss": 0.0243,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984605050038338e-05,
+      "loss": 0.021,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984393493146873e-05,
+      "loss": 0.0935,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984180493689345e-05,
+      "loss": 0.0411,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998396605169652e-05,
+      "loss": 0.0692,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983750167199377e-05,
+      "loss": 0.0523,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998353284022912e-05,
+      "loss": 0.159,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983314070817142e-05,
+      "loss": 0.0945,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998309385899505e-05,
+      "loss": 0.0353,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982872204794662e-05,
+      "loss": 0.1256,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998264910824801e-05,
+      "loss": 0.0671,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982424569387313e-05,
+      "loss": 0.0538,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982198588245027e-05,
+      "loss": 0.1605,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998197116485379e-05,
+      "loss": 0.0239,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998174229924647e-05,
+      "loss": 0.1258,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998151199145613e-05,
+      "loss": 0.0807,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998128024151605e-05,
+      "loss": 0.1558,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981047049459703e-05,
+      "loss": 0.1576,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980812415320793e-05,
+      "loss": 0.0298,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980576339133212e-05,
+      "loss": 0.0888,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980338820931074e-05,
+      "loss": 0.0691,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998009986074869e-05,
+      "loss": 0.0207,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979859458620595e-05,
+      "loss": 0.1205,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979617614581513e-05,
+      "loss": 0.0252,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979374328666398e-05,
+      "loss": 0.1097,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979129600910386e-05,
+      "loss": 0.0829,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978883431348845e-05,
+      "loss": 0.0882,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997863582001734e-05,
+      "loss": 0.0852,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978386766951647e-05,
+      "loss": 0.0126,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978136272187745e-05,
+      "loss": 0.0398,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977884335761834e-05,
+      "loss": 0.0441,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977630957710313e-05,
+      "loss": 0.022,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977376138069783e-05,
+      "loss": 0.0688,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977119876877066e-05,
+      "loss": 0.02,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976862174169186e-05,
+      "loss": 0.0444,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976603029983383e-05,
+      "loss": 0.166,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976342444357088e-05,
+      "loss": 0.0546,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976080417327958e-05,
+      "loss": 0.0173,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997581694893385e-05,
+      "loss": 0.0329,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997555203921283e-05,
+      "loss": 0.0148,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975285688203167e-05,
+      "loss": 0.1425,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975017895943355e-05,
+      "loss": 0.0066,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974748662472076e-05,
+      "loss": 0.0891,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974477987828235e-05,
+      "loss": 0.0955,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997420587205094e-05,
+      "loss": 0.0123,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973932315179502e-05,
+      "loss": 0.2527,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973657317253447e-05,
+      "loss": 0.1233,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973380878312505e-05,
+      "loss": 0.0405,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973102998396623e-05,
+      "loss": 0.0362,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972823677545946e-05,
+      "loss": 0.0159,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997254291580083e-05,
+      "loss": 0.1289,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972260713201842e-05,
+      "loss": 0.2164,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971977069789753e-05,
+      "loss": 0.1499,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971691985605544e-05,
+      "loss": 0.2886,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997140546069041e-05,
+      "loss": 0.055,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997111749508574e-05,
+      "loss": 0.0089,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970828088833148e-05,
+      "loss": 0.1842,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970537241974442e-05,
+      "loss": 0.0074,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970244954551648e-05,
+      "loss": 0.0313,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969951226606995e-05,
+      "loss": 0.0978,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969656058182926e-05,
+      "loss": 0.0784,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969359449322078e-05,
+      "loss": 0.0195,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996906140006731e-05,
+      "loss": 0.0454,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968761910461686e-05,
+      "loss": 0.0552,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996846098054848e-05,
+      "loss": 0.0065,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968158610371164e-05,
+      "loss": 0.0671,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996785479997343e-05,
+      "loss": 0.0182,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967549549399172e-05,
+      "loss": 0.0346,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967242858692494e-05,
+      "loss": 0.0664,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966934727897702e-05,
+      "loss": 0.0937,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966625157059324e-05,
+      "loss": 0.15,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966314146222082e-05,
+      "loss": 0.089,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966001695430907e-05,
+      "loss": 0.0052,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965687804730955e-05,
+      "loss": 0.1132,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965372474167567e-05,
+      "loss": 0.0041,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965055703786306e-05,
+      "loss": 0.1212,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964737493632943e-05,
+      "loss": 0.0638,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996441784375345e-05,
+      "loss": 0.0686,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996409675419401e-05,
+      "loss": 0.1729,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963774225001017e-05,
+      "loss": 0.0061,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963450256221066e-05,
+      "loss": 0.0287,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996312484790097e-05,
+      "loss": 0.1806,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996279800008775e-05,
+      "loss": 0.1119,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962469712828613e-05,
+      "loss": 0.0462,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962139986171004e-05,
+      "loss": 0.0722,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996180882016256e-05,
+      "loss": 0.0693,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961476214851126e-05,
+      "loss": 0.0514,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961142170284762e-05,
+      "loss": 0.0034,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960806686511723e-05,
+      "loss": 0.088,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996046976358049e-05,
+      "loss": 0.1488,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960131401539737e-05,
+      "loss": 0.1055,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959791600438353e-05,
+      "loss": 0.1519,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995945036032543e-05,
+      "loss": 0.0524,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959107681250276e-05,
+      "loss": 0.0915,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99587635632624e-05,
+      "loss": 0.1413,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958418006411522e-05,
+      "loss": 0.1352,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958071010747563e-05,
+      "loss": 0.0289,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957722576320666e-05,
+      "loss": 0.0671,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957372703181165e-05,
+      "loss": 0.0266,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995702139137962e-05,
+      "loss": 0.0431,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995666864096678e-05,
+      "loss": 0.0647,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995631445199362e-05,
+      "loss": 0.0837,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955958824511302e-05,
+      "loss": 0.0679,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995560175857122e-05,
+      "loss": 0.1183,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955243254224958e-05,
+      "loss": 0.0403,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954883311524316e-05,
+      "loss": 0.1841,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954521930521296e-05,
+      "loss": 0.0516,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995415911126811e-05,
+      "loss": 0.0319,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953794853817185e-05,
+      "loss": 0.0577,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953429158221144e-05,
+      "loss": 0.0205,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995306202453283e-05,
+      "loss": 0.0042,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952693452805278e-05,
+      "loss": 0.0642,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952323443091742e-05,
+      "loss": 0.1136,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995195199544569e-05,
+      "loss": 0.1435,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951579109920783e-05,
+      "loss": 0.0325,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951204786570896e-05,
+      "loss": 0.0034,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950829025450116e-05,
+      "loss": 0.0503,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995045182661273e-05,
+      "loss": 0.014,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950073190113238e-05,
+      "loss": 0.0588,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949693116006343e-05,
+      "loss": 0.0879,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949311604346965e-05,
+      "loss": 0.0505,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994892865519022e-05,
+      "loss": 0.0198,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948544268591445e-05,
+      "loss": 0.1788,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948158444606163e-05,
+      "loss": 0.0084,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947771183290133e-05,
+      "loss": 0.0198,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99473824846993e-05,
+      "loss": 0.0733,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946992348889827e-05,
+      "loss": 0.076,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994660077591808e-05,
+      "loss": 0.2511,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946207765840633e-05,
+      "loss": 0.0341,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994581331871427e-05,
+      "loss": 0.0772,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945417434595983e-05,
+      "loss": 0.1358,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994502011354297e-05,
+      "loss": 0.1628,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944621355612637e-05,
+      "loss": 0.064,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944221160862594e-05,
+      "loss": 0.1209,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943819529350664e-05,
+      "loss": 0.1566,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943416461134877e-05,
+      "loss": 0.0724,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943011956273467e-05,
+      "loss": 0.1821,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994260601482488e-05,
+      "loss": 0.0817,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942198636847764e-05,
+      "loss": 0.1427,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994178982240098e-05,
+      "loss": 0.1279,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941379571543597e-05,
+      "loss": 0.0118,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940967884334886e-05,
+      "loss": 0.0574,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994055476083433e-05,
+      "loss": 0.036,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994014020110161e-05,
+      "loss": 0.1102,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939724205196632e-05,
+      "loss": 0.0665,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939306773179498e-05,
+      "loss": 0.0925,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938887905110514e-05,
+      "loss": 0.119,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938467601050207e-05,
+      "loss": 0.0468,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938045861059302e-05,
+      "loss": 0.0489,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937622685198724e-05,
+      "loss": 0.0229,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937198073529626e-05,
+      "loss": 0.0289,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993677202611335e-05,
+      "loss": 0.0297,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936344543011455e-05,
+      "loss": 0.0074,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935915624285706e-05,
+      "loss": 0.0951,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935485269998065e-05,
+      "loss": 0.0868,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993505348021072e-05,
+      "loss": 0.0415,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934620254986054e-05,
+      "loss": 0.0978,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993418559438666e-05,
+      "loss": 0.0206,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993374949847534e-05,
+      "loss": 0.0761,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99333119673151e-05,
+      "loss": 0.0215,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993287300096916e-05,
+      "loss": 0.1075,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993243259950093e-05,
+      "loss": 0.068,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993199076297406e-05,
+      "loss": 0.0703,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993154749145237e-05,
+      "loss": 0.0038,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931102784999915e-05,
+      "loss": 0.0076,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993065664368094e-05,
+      "loss": 0.0546,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930209067559913e-05,
+      "loss": 0.1339,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929760056701494e-05,
+      "loss": 0.0153,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992930961117056e-05,
+      "loss": 0.0055,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992885773103219e-05,
+      "loss": 0.0602,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992840441635167e-05,
+      "loss": 0.0366,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927949667194503e-05,
+      "loss": 0.2887,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927493483626386e-05,
+      "loss": 0.1248,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992703586571324e-05,
+      "loss": 0.076,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926576813521167e-05,
+      "loss": 0.0399,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99261163271165e-05,
+      "loss": 0.0792,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925654406565774e-05,
+      "loss": 0.0477,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925191051935722e-05,
+      "loss": 0.1051,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924726263293297e-05,
+      "loss": 0.2026,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924260040705644e-05,
+      "loss": 0.1756,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923792384240134e-05,
+      "loss": 0.0311,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923323293964325e-05,
+      "loss": 0.03,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922852769945997e-05,
+      "loss": 0.0316,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922380812253134e-05,
+      "loss": 0.1106,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921907420953923e-05,
+      "loss": 0.0138,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921432596116763e-05,
+      "loss": 0.0958,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920956337810255e-05,
+      "loss": 0.0126,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920478646103207e-05,
+      "loss": 0.1146,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919999521064647e-05,
+      "loss": 0.0948,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919518962763788e-05,
+      "loss": 0.0579,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991903697127007e-05,
+      "loss": 0.0575,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991855354665313e-05,
+      "loss": 0.0025,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918068688982814e-05,
+      "loss": 0.1746,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991758239832917e-05,
+      "loss": 0.1026,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917094674762468e-05,
+      "loss": 0.0313,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991660551835317e-05,
+      "loss": 0.0786,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916114929171954e-05,
+      "loss": 0.0119,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915622907289695e-05,
+      "loss": 0.0083,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915129452777485e-05,
+      "loss": 0.1235,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991463456570662e-05,
+      "loss": 0.1105,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99141382461486e-05,
+      "loss": 0.0373,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991364049417514e-05,
+      "loss": 0.0801,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913141309858147e-05,
+      "loss": 0.0075,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912640693269754e-05,
+      "loss": 0.0526,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912138644482283e-05,
+      "loss": 0.0412,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911635163568276e-05,
+      "loss": 0.0543,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911130250600473e-05,
+      "loss": 0.03,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991062390565183e-05,
+      "loss": 0.0214,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99101161287955e-05,
+      "loss": 0.0637,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909606920104853e-05,
+      "loss": 0.06,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909096279653455e-05,
+      "loss": 0.0815,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908584207515084e-05,
+      "loss": 0.037,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908070703763732e-05,
+      "loss": 0.0164,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907555768473585e-05,
+      "loss": 0.0303,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907039401719044e-05,
+      "loss": 0.052,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906521603574713e-05,
+      "loss": 0.0368,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990600237411541e-05,
+      "loss": 0.103,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990548171341615e-05,
+      "loss": 0.0879,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904959621552158e-05,
+      "loss": 0.0441,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904436098598867e-05,
+      "loss": 0.0934,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903911144631924e-05,
+      "loss": 0.0686,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903384759727164e-05,
+      "loss": 0.0048,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990285694396065e-05,
+      "loss": 0.0194,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902327697408634e-05,
+      "loss": 0.0454,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901797020147588e-05,
+      "loss": 0.125,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901264912254186e-05,
+      "loss": 0.0348,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900731373805305e-05,
+      "loss": 0.1577,
+      "step": 627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990019640487803e-05,
+      "loss": 0.0234,
+      "step": 628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989966000554966e-05,
+      "loss": 0.1174,
+      "step": 629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899122175897695e-05,
+      "loss": 0.0837,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898582915999834e-05,
+      "loss": 0.1262,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898042225933997e-05,
+      "loss": 0.135,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98975001057783e-05,
+      "loss": 0.1001,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896956555611078e-05,
+      "loss": 0.0042,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896411575510856e-05,
+      "loss": 0.217,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895865165556375e-05,
+      "loss": 0.0631,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989531732582658e-05,
+      "loss": 0.163,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894768056400635e-05,
+      "loss": 0.0347,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894217357357884e-05,
+      "loss": 0.2271,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893665228777903e-05,
+      "loss": 0.243,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893111670740462e-05,
+      "loss": 0.2631,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989255668332554e-05,
+      "loss": 0.1568,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892000266613322e-05,
+      "loss": 0.0824,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891442420684204e-05,
+      "loss": 0.0756,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890883145618784e-05,
+      "loss": 0.0775,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890322441497857e-05,
+      "loss": 0.1153,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988976030840245e-05,
+      "loss": 0.0514,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889196746413773e-05,
+      "loss": 0.0754,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988863175561325e-05,
+      "loss": 0.1049,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988806533608252e-05,
+      "loss": 0.0579,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887497487903406e-05,
+      "loss": 0.1164,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886928211157962e-05,
+      "loss": 0.0788,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886357505928438e-05,
+      "loss": 0.0842,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885785372297293e-05,
+      "loss": 0.1392,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885211810347185e-05,
+      "loss": 0.0354,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988463682016098e-05,
+      "loss": 0.0434,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884060401821763e-05,
+      "loss": 0.0788,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988348255541281e-05,
+      "loss": 0.0101,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988290328101761e-05,
+      "loss": 0.0637,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988232257871986e-05,
+      "loss": 0.109,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988174044860346e-05,
+      "loss": 0.0198,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881156890752517e-05,
+      "loss": 0.1791,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880571905251346e-05,
+      "loss": 0.0374,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879985492184467e-05,
+      "loss": 0.0061,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879397651636605e-05,
+      "loss": 0.0065,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987880838369269e-05,
+      "loss": 0.0765,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878217688437865e-05,
+      "loss": 0.0527,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987762556595747e-05,
+      "loss": 0.0132,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877032016337064e-05,
+      "loss": 0.0088,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876437039662395e-05,
+      "loss": 0.2192,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875840636019435e-05,
+      "loss": 0.0045,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875242805494347e-05,
+      "loss": 0.1567,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987464354817351e-05,
+      "loss": 0.0794,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987404286414351e-05,
+      "loss": 0.0048,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873440753491122e-05,
+      "loss": 0.0024,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872837216303353e-05,
+      "loss": 0.0126,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872232252667397e-05,
+      "loss": 0.0214,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871625862670666e-05,
+      "loss": 0.1031,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987101804640077e-05,
+      "loss": 0.0265,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870408803945524e-05,
+      "loss": 0.0505,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869798135392953e-05,
+      "loss": 0.0028,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869186040831292e-05,
+      "loss": 0.0517,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868572520348973e-05,
+      "loss": 0.0019,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867957574034644e-05,
+      "loss": 0.028,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867341201977152e-05,
+      "loss": 0.1463,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866723404265546e-05,
+      "loss": 0.0093,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866104180989098e-05,
+      "loss": 0.089,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986548353223726e-05,
+      "loss": 0.0924,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986486145809972e-05,
+      "loss": 0.0466,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864237958666347e-05,
+      "loss": 0.0225,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863613034027224e-05,
+      "loss": 0.125,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986298668427265e-05,
+      "loss": 0.0066,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862358909493113e-05,
+      "loss": 0.0871,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986172970977932e-05,
+      "loss": 0.0971,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986109908522218e-05,
+      "loss": 0.0224,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98604670359128e-05,
+      "loss": 0.0261,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985983356194251e-05,
+      "loss": 0.0049,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859198663402827e-05,
+      "loss": 0.054,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858562340385488e-05,
+      "loss": 0.0117,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857924592982424e-05,
+      "loss": 0.0852,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857285421285785e-05,
+      "loss": 0.0748,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856644825387917e-05,
+      "loss": 0.146,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856002805381372e-05,
+      "loss": 0.0738,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855359361358913e-05,
+      "loss": 0.033,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985471449341351e-05,
+      "loss": 0.0483,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854068201638327e-05,
+      "loss": 0.1048,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853420486126747e-05,
+      "loss": 0.0586,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852771346972354e-05,
+      "loss": 0.0396,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985212078426893e-05,
+      "loss": 0.0602,
+      "step": 709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985146879811048e-05,
+      "loss": 0.1609,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850815388591195e-05,
+      "loss": 0.0886,
+      "step": 711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850160555805485e-05,
+      "loss": 0.1304,
+      "step": 712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849504299847964e-05,
+      "loss": 0.3725,
+      "step": 713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848846620813444e-05,
+      "loss": 0.2474,
+      "step": 714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848187518796952e-05,
+      "loss": 0.084,
+      "step": 715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847526993893712e-05,
+      "loss": 0.0264,
+      "step": 716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984686504619916e-05,
+      "loss": 0.0398,
+      "step": 717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846201675808943e-05,
+      "loss": 0.1136,
+      "step": 718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845536882818893e-05,
+      "loss": 0.12,
+      "step": 719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844870667325073e-05,
+      "loss": 0.0327,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844203029423733e-05,
+      "loss": 0.0597,
+      "step": 721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843533969211332e-05,
+      "loss": 0.0735,
+      "step": 722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842863486784544e-05,
+      "loss": 0.0106,
+      "step": 723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842191582240233e-05,
+      "loss": 0.0259,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841518255675485e-05,
+      "loss": 0.0862,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984084350718758e-05,
+      "loss": 0.0287,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840167336874012e-05,
+      "loss": 0.0304,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839489744832468e-05,
+      "loss": 0.0983,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983881073116085e-05,
+      "loss": 0.0665,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983813029595727e-05,
+      "loss": 0.0219,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837448439320027e-05,
+      "loss": 0.051,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836765161347647e-05,
+      "loss": 0.1723,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836080462138852e-05,
+      "loss": 0.0451,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983539434179256e-05,
+      "loss": 0.0814,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983470680040791e-05,
+      "loss": 0.0699,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834017838084234e-05,
+      "loss": 0.038,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983332745492108e-05,
+      "loss": 0.1006,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832635651018196e-05,
+      "loss": 0.0313,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983194242647553e-05,
+      "loss": 0.0868,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831247781393246e-05,
+      "loss": 0.0693,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830551715871703e-05,
+      "loss": 0.0685,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829854230011477e-05,
+      "loss": 0.1309,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829155323913334e-05,
+      "loss": 0.1035,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828454997678257e-05,
+      "loss": 0.0244,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827753251407434e-05,
+      "loss": 0.1488,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982705008520225e-05,
+      "loss": 0.0303,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826345499164302e-05,
+      "loss": 0.0161,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982563949339539e-05,
+      "loss": 0.2095,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824932067997516e-05,
+      "loss": 0.0782,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824223223072895e-05,
+      "loss": 0.0253,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823512958723942e-05,
+      "loss": 0.0921,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822801275053274e-05,
+      "loss": 0.0851,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822088172163717e-05,
+      "loss": 0.0338,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821373650158305e-05,
+      "loss": 0.0613,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820657709140273e-05,
+      "loss": 0.2068,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981994034921306e-05,
+      "loss": 0.0876,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981922157048031e-05,
+      "loss": 0.3167,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981850137304588e-05,
+      "loss": 0.0759,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817779757013817e-05,
+      "loss": 0.0525,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817056722488387e-05,
+      "loss": 0.0769,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816332269574056e-05,
+      "loss": 0.0129,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815606398375492e-05,
+      "loss": 0.0774,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981487910899757e-05,
+      "loss": 0.0158,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814150401545375e-05,
+      "loss": 0.0375,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813420276124183e-05,
+      "loss": 0.0245,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812688732839497e-05,
+      "loss": 0.0656,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811955771797e-05,
+      "loss": 0.0275,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98112213931026e-05,
+      "loss": 0.1297,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981048559686239e-05,
+      "loss": 0.0193,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98097483831827e-05,
+      "loss": 0.0724,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809009752170022e-05,
+      "loss": 0.0368,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808269703931088e-05,
+      "loss": 0.0944,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980752823857282e-05,
+      "loss": 0.0632,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806785356202342e-05,
+      "loss": 0.0708,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980604105692699e-05,
+      "loss": 0.0197,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805295340854303e-05,
+      "loss": 0.0253,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804548208092022e-05,
+      "loss": 0.1329,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803799658748096e-05,
+      "loss": 0.1052,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803049692930675e-05,
+      "loss": 0.1627,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802298310748117e-05,
+      "loss": 0.0548,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801545512308982e-05,
+      "loss": 0.0111,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800791297722036e-05,
+      "loss": 0.1245,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980003566709625e-05,
+      "loss": 0.0181,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97992786205408e-05,
+      "loss": 0.0582,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798520158165068e-05,
+      "loss": 0.0332,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979776028007863e-05,
+      "loss": 0.103,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979699898639128e-05,
+      "loss": 0.0339,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796236277213013e-05,
+      "loss": 0.1253,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795472152654027e-05,
+      "loss": 0.0073,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979470661282472e-05,
+      "loss": 0.0288,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97939396578357e-05,
+      "loss": 0.0809,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979317128779778e-05,
+      "loss": 0.0338,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979240150282198e-05,
+      "loss": 0.0328,
+      "step": 793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791630303019512e-05,
+      "loss": 0.0531,
+      "step": 794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97908576885018e-05,
+      "loss": 0.0619,
+      "step": 795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979008365938048e-05,
+      "loss": 0.0409,
+      "step": 796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789308215767385e-05,
+      "loss": 0.0386,
+      "step": 797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788531357774547e-05,
+      "loss": 0.0162,
+      "step": 798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978775308551421e-05,
+      "loss": 0.0242,
+      "step": 799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786973399098823e-05,
+      "loss": 0.0802,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978619229864103e-05,
+      "loss": 0.0843,
+      "step": 801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785409784253694e-05,
+      "loss": 0.0651,
+      "step": 802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784625856049876e-05,
+      "loss": 0.1293,
+      "step": 803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978384051414283e-05,
+      "loss": 0.0807,
+      "step": 804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978305375864603e-05,
+      "loss": 0.0554,
+      "step": 805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782265589673146e-05,
+      "loss": 0.0502,
+      "step": 806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 0.0562,
+      "step": 807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780685011754843e-05,
+      "loss": 0.0088,
+      "step": 808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779892603037783e-05,
+      "loss": 0.0715,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779098781301374e-05,
+      "loss": 0.0807,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778303546660304e-05,
+      "loss": 0.015,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777506899229474e-05,
+      "loss": 0.0993,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776708839123975e-05,
+      "loss": 0.0115,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775909366459128e-05,
+      "loss": 0.1635,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977510848135043e-05,
+      "loss": 0.0198,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774306183913602e-05,
+      "loss": 0.0953,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773502474264557e-05,
+      "loss": 0.0992,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772697352519422e-05,
+      "loss": 0.0635,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771890818794514e-05,
+      "loss": 0.2273,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771082873206372e-05,
+      "loss": 0.0635,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770273515871724e-05,
+      "loss": 0.0965,
+      "step": 821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769462746907507e-05,
+      "loss": 0.0863,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768650566430867e-05,
+      "loss": 0.0206,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767836974559147e-05,
+      "loss": 0.0512,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767021971409896e-05,
+      "loss": 0.0817,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976620555710087e-05,
+      "loss": 0.1018,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976538773175002e-05,
+      "loss": 0.0131,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764568495475517e-05,
+      "loss": 0.0569,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763747848395715e-05,
+      "loss": 0.0346,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976292579062919e-05,
+      "loss": 0.1006,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762102322294716e-05,
+      "loss": 0.01,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761277443511263e-05,
+      "loss": 0.1015,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760451154398014e-05,
+      "loss": 0.0185,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759623455074356e-05,
+      "loss": 0.0075,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758794345659873e-05,
+      "loss": 0.1354,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757963826274357e-05,
+      "loss": 0.0982,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757131897037807e-05,
+      "loss": 0.0038,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756298558070416e-05,
+      "loss": 0.1275,
+      "step": 838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975546380949259e-05,
+      "loss": 0.0347,
+      "step": 839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975462765142493e-05,
+      "loss": 0.0332,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753790083988256e-05,
+      "loss": 0.0798,
+      "step": 841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752951107303575e-05,
+      "loss": 0.1238,
+      "step": 842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97521107214921e-05,
+      "loss": 0.08,
+      "step": 843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975126892667526e-05,
+      "loss": 0.116,
+      "step": 844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975042572297468e-05,
+      "loss": 0.0651,
+      "step": 845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974958111051218e-05,
+      "loss": 0.0781,
+      "step": 846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748735089409795e-05,
+      "loss": 0.0207,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974788765978976e-05,
+      "loss": 0.008,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747038821774516e-05,
+      "loss": 0.0346,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97461885754867e-05,
+      "loss": 0.1693,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974533692104916e-05,
+      "loss": 0.0699,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744483858584942e-05,
+      "loss": 0.1983,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743629388217305e-05,
+      "loss": 0.0935,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97427735100697e-05,
+      "loss": 0.0061,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741916224265788e-05,
+      "loss": 0.2259,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741057530929425e-05,
+      "loss": 0.147,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740197430184686e-05,
+      "loss": 0.0369,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973933592215583e-05,
+      "loss": 0.0338,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973847300696734e-05,
+      "loss": 0.0887,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973760868474389e-05,
+      "loss": 0.0338,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973674295561035e-05,
+      "loss": 0.112,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973587581969181e-05,
+      "loss": 0.0805,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735007277113557e-05,
+      "loss": 0.0452,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734137328001077e-05,
+      "loss": 0.0168,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973326597248006e-05,
+      "loss": 0.098,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732393210676405e-05,
+      "loss": 0.1056,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731519042716208e-05,
+      "loss": 0.2076,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973064346872577e-05,
+      "loss": 0.0231,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97297664888316e-05,
+      "loss": 0.0522,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728888103160404e-05,
+      "loss": 0.0476,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972800831183909e-05,
+      "loss": 0.0213,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727127114994772e-05,
+      "loss": 0.051,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726244512754774e-05,
+      "loss": 0.0784,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725360505246608e-05,
+      "loss": 0.012,
+      "step": 874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724475092598003e-05,
+      "loss": 0.0714,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723588274936882e-05,
+      "loss": 0.1401,
+      "step": 876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722700052391376e-05,
+      "loss": 0.0039,
+      "step": 877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972181042508982e-05,
+      "loss": 0.0978,
+      "step": 878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720919393160742e-05,
+      "loss": 0.1108,
+      "step": 879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720026956732884e-05,
+      "loss": 0.176,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719133115935188e-05,
+      "loss": 0.0297,
+      "step": 881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718237870896794e-05,
+      "loss": 0.106,
+      "step": 882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717341221747056e-05,
+      "loss": 0.0571,
+      "step": 883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971644316861552e-05,
+      "loss": 0.0365,
+      "step": 884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715543711631938e-05,
+      "loss": 0.0807,
+      "step": 885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714642850926264e-05,
+      "loss": 0.0256,
+      "step": 886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713740586628656e-05,
+      "loss": 0.0632,
+      "step": 887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712836918869483e-05,
+      "loss": 0.0309,
+      "step": 888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97119318477793e-05,
+      "loss": 0.036,
+      "step": 889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711025373488875e-05,
+      "loss": 0.0586,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971011749612918e-05,
+      "loss": 0.0968,
+      "step": 891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709208215831387e-05,
+      "loss": 0.0817,
+      "step": 892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970829753272687e-05,
+      "loss": 0.0636,
+      "step": 893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97073854469472e-05,
+      "loss": 0.0548,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970647195862417e-05,
+      "loss": 0.0475,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970555706788975e-05,
+      "loss": 0.0196,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704640774876135e-05,
+      "loss": 0.0966,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970372307971571e-05,
+      "loss": 0.0689,
+      "step": 898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970280398254106e-05,
+      "loss": 0.0055,
+      "step": 899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701883483484985e-05,
+      "loss": 0.0375,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700961582680476e-05,
+      "loss": 0.0178,
+      "step": 901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700038280260734e-05,
+      "loss": 0.012,
+      "step": 902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699113576359162e-05,
+      "loss": 0.0031,
+      "step": 903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698187471109356e-05,
+      "loss": 0.0605,
+      "step": 904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969725996464513e-05,
+      "loss": 0.0594,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696331057100484e-05,
+      "loss": 0.0652,
+      "step": 906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969540074860963e-05,
+      "loss": 0.0306,
+      "step": 907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969446903930699e-05,
+      "loss": 0.0034,
+      "step": 908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969353592932717e-05,
+      "loss": 0.1147,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969260141880499e-05,
+      "loss": 0.2139,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691665507875467e-05,
+      "loss": 0.0597,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969072819667383e-05,
+      "loss": 0.056,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689789485335502e-05,
+      "loss": 0.0604,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688849373996106e-05,
+      "loss": 0.0345,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687907862791474e-05,
+      "loss": 0.0247,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686964951857636e-05,
+      "loss": 0.0412,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968602064133083e-05,
+      "loss": 0.0213,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685074931347486e-05,
+      "loss": 0.0299,
+      "step": 918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684127822044243e-05,
+      "loss": 0.0399,
+      "step": 919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683179313557947e-05,
+      "loss": 0.0575,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682229406025635e-05,
+      "loss": 0.0591,
+      "step": 921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681278099584553e-05,
+      "loss": 0.0655,
+      "step": 922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968032539437215e-05,
+      "loss": 0.0266,
+      "step": 923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967937129052607e-05,
+      "loss": 0.1212,
+      "step": 924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967841578818417e-05,
+      "loss": 0.0647,
+      "step": 925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96774588874845e-05,
+      "loss": 0.0236,
+      "step": 926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676500588565313e-05,
+      "loss": 0.0442,
+      "step": 927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675540891565065e-05,
+      "loss": 0.0569,
+      "step": 928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967457979662242e-05,
+      "loss": 0.016,
+      "step": 929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673617303876242e-05,
+      "loss": 0.0757,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672653413465584e-05,
+      "loss": 0.0309,
+      "step": 931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671688125529718e-05,
+      "loss": 0.0126,
+      "step": 932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670721440208107e-05,
+      "loss": 0.0969,
+      "step": 933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966975335764042e-05,
+      "loss": 0.0674,
+      "step": 934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668783877966534e-05,
+      "loss": 0.008,
+      "step": 935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667813001326512e-05,
+      "loss": 0.0565,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666840727860635e-05,
+      "loss": 0.0021,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665867057709375e-05,
+      "loss": 0.0094,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664891991013414e-05,
+      "loss": 0.0249,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663915527913628e-05,
+      "loss": 0.0393,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662937668551097e-05,
+      "loss": 0.0054,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966195841306711e-05,
+      "loss": 0.0172,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966097776160315e-05,
+      "loss": 0.0141,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96599957143009e-05,
+      "loss": 0.0597,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659012271302253e-05,
+      "loss": 0.0721,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658027432749293e-05,
+      "loss": 0.1063,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965704119878432e-05,
+      "loss": 0.0206,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656053569549817e-05,
+      "loss": 0.0971,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965506454518849e-05,
+      "loss": 0.0011,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654074125843222e-05,
+      "loss": 0.0036,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653082311657124e-05,
+      "loss": 0.0991,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652089102773487e-05,
+      "loss": 0.0019,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651094499335815e-05,
+      "loss": 0.0544,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965009850148781e-05,
+      "loss": 0.001,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649101109373377e-05,
+      "loss": 0.0338,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964810232313662e-05,
+      "loss": 0.0703,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647102142921844e-05,
+      "loss": 0.1135,
+      "step": 957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964610056887356e-05,
+      "loss": 0.1665,
+      "step": 958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645097601136475e-05,
+      "loss": 0.137,
+      "step": 959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644093239855504e-05,
+      "loss": 0.0471,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643087485175752e-05,
+      "loss": 0.0797,
+      "step": 961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964208033724254e-05,
+      "loss": 0.0275,
+      "step": 962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964107179620138e-05,
+      "loss": 0.011,
+      "step": 963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640061862197986e-05,
+      "loss": 0.0291,
+      "step": 964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963905053537828e-05,
+      "loss": 0.0661,
+      "step": 965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638037815888378e-05,
+      "loss": 0.0259,
+      "step": 966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96370237038746e-05,
+      "loss": 0.0689,
+      "step": 967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636008199483464e-05,
+      "loss": 0.0123,
+      "step": 968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634991302861695e-05,
+      "loss": 0.205,
+      "step": 969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963397301415622e-05,
+      "loss": 0.0975,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963295333351416e-05,
+      "loss": 0.1153,
+      "step": 971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631932261082834e-05,
+      "loss": 0.1999,
+      "step": 972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963090979700978e-05,
+      "loss": 0.0086,
+      "step": 973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962988594144272e-05,
+      "loss": 0.0463,
+      "step": 974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628860694529586e-05,
+      "loss": 0.0561,
+      "step": 975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96278340564185e-05,
+      "loss": 0.0236,
+      "step": 976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626806027257804e-05,
+      "loss": 0.0386,
+      "step": 977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625776607196024e-05,
+      "loss": 0.0263,
+      "step": 978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962474579638189e-05,
+      "loss": 0.0739,
+      "step": 979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623713594964337e-05,
+      "loss": 0.0409,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622680003092503e-05,
+      "loss": 0.0249,
+      "step": 981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621645020915725e-05,
+      "loss": 0.0099,
+      "step": 982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962060864858353e-05,
+      "loss": 0.0027,
+      "step": 983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961957088624567e-05,
+      "loss": 0.0275,
+      "step": 984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618531734052068e-05,
+      "loss": 0.0442,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961749119215287e-05,
+      "loss": 0.0083,
+      "step": 986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616449260698416e-05,
+      "loss": 0.0629,
+      "step": 987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615405939839246e-05,
+      "loss": 0.0604,
+      "step": 988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96143612297261e-05,
+      "loss": 0.1132,
+      "step": 989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961331513050992e-05,
+      "loss": 0.0736,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961226764234185e-05,
+      "loss": 0.0066,
+      "step": 991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611218765373234e-05,
+      "loss": 0.0037,
+      "step": 992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610168499755612e-05,
+      "loss": 0.0981,
+      "step": 993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960911684564073e-05,
+      "loss": 0.0081,
+      "step": 994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608063803180535e-05,
+      "loss": 0.0266,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607009372527172e-05,
+      "loss": 0.0757,
+      "step": 996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605953553832987e-05,
+      "loss": 0.0659,
+      "step": 997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960489634725053e-05,
+      "loss": 0.0798,
+      "step": 998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603837752932545e-05,
+      "loss": 0.1987,
+      "step": 999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602777771031976e-05,
+      "loss": 0.089,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601716401701976e-05,
+      "loss": 0.019,
+      "step": 1001
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96006536450959e-05,
+      "loss": 0.0557,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599589501367286e-05,
+      "loss": 0.0477,
+      "step": 1003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598523970669893e-05,
+      "loss": 0.0465,
+      "step": 1004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597457053157662e-05,
+      "loss": 0.0786,
+      "step": 1005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959638874898475e-05,
+      "loss": 0.0147,
+      "step": 1006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959531905830551e-05,
+      "loss": 0.0345,
+      "step": 1007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594247981274484e-05,
+      "loss": 0.01,
+      "step": 1008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959317551804643e-05,
+      "loss": 0.0536,
+      "step": 1009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95921016687763e-05,
+      "loss": 0.1153,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591026433619246e-05,
+      "loss": 0.0804,
+      "step": 1011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589949812730614e-05,
+      "loss": 0.0471,
+      "step": 1012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588871806265962e-05,
+      "loss": 0.0147,
+      "step": 1013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587792414381048e-05,
+      "loss": 0.086,
+      "step": 1014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958671163723181e-05,
+      "loss": 0.0419,
+      "step": 1015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585629474974413e-05,
+      "loss": 0.0116,
+      "step": 1016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584545927765208e-05,
+      "loss": 0.2126,
+      "step": 1017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958346099576075e-05,
+      "loss": 0.0528,
+      "step": 1018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582374679117783e-05,
+      "loss": 0.1044,
+      "step": 1019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581286977993267e-05,
+      "loss": 0.1201,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580197892544354e-05,
+      "loss": 0.0275,
+      "step": 1021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579107422928402e-05,
+      "loss": 0.0517,
+      "step": 1022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957801556930296e-05,
+      "loss": 0.0911,
+      "step": 1023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576922331825776e-05,
+      "loss": 0.0334,
+      "step": 1024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575827710654812e-05,
+      "loss": 0.0057,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957473170594822e-05,
+      "loss": 0.0196,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957363431786435e-05,
+      "loss": 0.0342,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572535546561757e-05,
+      "loss": 0.0819,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571435392199188e-05,
+      "loss": 0.0079,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570333854935607e-05,
+      "loss": 0.045,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569230934930157e-05,
+      "loss": 0.0588,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568126632342192e-05,
+      "loss": 0.1124,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956702094733127e-05,
+      "loss": 0.0042,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956591388005713e-05,
+      "loss": 0.0125,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564805430679733e-05,
+      "loss": 0.0053,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563695599359233e-05,
+      "loss": 0.0051,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956258438625597e-05,
+      "loss": 0.1216,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561471791530503e-05,
+      "loss": 0.1438,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560357815343577e-05,
+      "loss": 0.1168,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559242457856147e-05,
+      "loss": 0.0172,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558125719229357e-05,
+      "loss": 0.0034,
+      "step": 1041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955700759962456e-05,
+      "loss": 0.0046,
+      "step": 1042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95558880992033e-05,
+      "loss": 0.019,
+      "step": 1043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955476721812733e-05,
+      "loss": 0.0515,
+      "step": 1044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955364495655859e-05,
+      "loss": 0.0134,
+      "step": 1045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955252131465923e-05,
+      "loss": 0.066,
+      "step": 1046
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95513962925916e-05,
+      "loss": 0.0057,
+      "step": 1047
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550269890518243e-05,
+      "loss": 0.033,
+      "step": 1048
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95491421086019e-05,
+      "loss": 0.0553,
+      "step": 1049
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954801294700552e-05,
+      "loss": 0.2068,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546882405892247e-05,
+      "loss": 0.1521,
+      "step": 1051
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545750485425422e-05,
+      "loss": 0.0433,
+      "step": 1052
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544617185768585e-05,
+      "loss": 0.0482,
+      "step": 1053
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543482507085484e-05,
+      "loss": 0.0208,
+      "step": 1054
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954234644954005e-05,
+      "loss": 0.0203,
+      "step": 1055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541209013296435e-05,
+      "loss": 0.1304,
+      "step": 1056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540070198518967e-05,
+      "loss": 0.0016,
+      "step": 1057
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538930005372194e-05,
+      "loss": 0.0904,
+      "step": 1058
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537788434020843e-05,
+      "loss": 0.0329,
+      "step": 1059
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953664548462986e-05,
+      "loss": 0.0699,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535501157364375e-05,
+      "loss": 0.0585,
+      "step": 1061
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534355452389727e-05,
+      "loss": 0.0241,
+      "step": 1062
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533208369871448e-05,
+      "loss": 0.07,
+      "step": 1063
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953205990997527e-05,
+      "loss": 0.0568,
+      "step": 1064
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9530910072867124e-05,
+      "loss": 0.0164,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952975885871314e-05,
+      "loss": 0.05,
+      "step": 1066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528606267679655e-05,
+      "loss": 0.0377,
+      "step": 1067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527452299933192e-05,
+      "loss": 0.0549,
+      "step": 1068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952629695564048e-05,
+      "loss": 0.0184,
+      "step": 1069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525140234968444e-05,
+      "loss": 0.1184,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952398213808421e-05,
+      "loss": 0.05,
+      "step": 1071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522822665155104e-05,
+      "loss": 0.0239,
+      "step": 1072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521661816348647e-05,
+      "loss": 0.0456,
+      "step": 1073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520499591832564e-05,
+      "loss": 0.1084,
+      "step": 1074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951933599177477e-05,
+      "loss": 0.0196,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951817101634339e-05,
+      "loss": 0.0384,
+      "step": 1076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517004665706738e-05,
+      "loss": 0.0525,
+      "step": 1077
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515836940033335e-05,
+      "loss": 0.004,
+      "step": 1078
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514667839491892e-05,
+      "loss": 0.0069,
+      "step": 1079
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513497364251327e-05,
+      "loss": 0.0801,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951232551448075e-05,
+      "loss": 0.0813,
+      "step": 1081
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511152290349473e-05,
+      "loss": 0.0508,
+      "step": 1082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950997769202701e-05,
+      "loss": 0.0113,
+      "step": 1083
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950880171968306e-05,
+      "loss": 0.008,
+      "step": 1084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507624373487538e-05,
+      "loss": 0.0343,
+      "step": 1085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506445653610548e-05,
+      "loss": 0.2273,
+      "step": 1086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950526556022239e-05,
+      "loss": 0.0485,
+      "step": 1087
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950408409349357e-05,
+      "loss": 0.0646,
+      "step": 1088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950290125359479e-05,
+      "loss": 0.0025,
+      "step": 1089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501717040696945e-05,
+      "loss": 0.1106,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500531454971136e-05,
+      "loss": 0.0273,
+      "step": 1091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499344496588656e-05,
+      "loss": 0.0349,
+      "step": 1092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498156165721002e-05,
+      "loss": 0.0505,
+      "step": 1093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496966462539868e-05,
+      "loss": 0.0171,
+      "step": 1094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495775387217136e-05,
+      "loss": 0.0014,
+      "step": 1095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494582939924908e-05,
+      "loss": 0.0132,
+      "step": 1096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493389120835462e-05,
+      "loss": 0.0266,
+      "step": 1097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492193930121283e-05,
+      "loss": 0.0325,
+      "step": 1098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949099736795506e-05,
+      "loss": 0.168,
+      "step": 1099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489799434509672e-05,
+      "loss": 0.0822,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94886001299582e-05,
+      "loss": 0.0987,
+      "step": 1101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948739945447392e-05,
+      "loss": 0.0085,
+      "step": 1102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486197408230316e-05,
+      "loss": 0.1147,
+      "step": 1103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948499399140105e-05,
+      "loss": 0.0012,
+      "step": 1104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948378920416e-05,
+      "loss": 0.0032,
+      "step": 1105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482583046681238e-05,
+      "loss": 0.045,
+      "step": 1106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481375519139027e-05,
+      "loss": 0.0366,
+      "step": 1107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480166621707842e-05,
+      "loss": 0.0035,
+      "step": 1108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478956354562334e-05,
+      "loss": 0.0056,
+      "step": 1109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947774471787738e-05,
+      "loss": 0.061,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476531711828027e-05,
+      "loss": 0.0044,
+      "step": 1111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475317336589537e-05,
+      "loss": 0.0245,
+      "step": 1112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947410159233737e-05,
+      "loss": 0.0675,
+      "step": 1113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472884479247173e-05,
+      "loss": 0.0035,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94716659974948e-05,
+      "loss": 0.1143,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94704461472563e-05,
+      "loss": 0.0475,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946922492870792e-05,
+      "loss": 0.0509,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94680023420261e-05,
+      "loss": 0.0096,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466778387387487e-05,
+      "loss": 0.1888,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465553064968915e-05,
+      "loss": 0.0044,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464326374947432e-05,
+      "loss": 0.1284,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946309831750026e-05,
+      "loss": 0.0585,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461868892804836e-05,
+      "loss": 0.1912,
+      "step": 1123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460638101038795e-05,
+      "loss": 0.0363,
+      "step": 1124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459405942379955e-05,
+      "loss": 0.023,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458172417006347e-05,
+      "loss": 0.0781,
+      "step": 1126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456937525096196e-05,
+      "loss": 0.0421,
+      "step": 1127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455701266827915e-05,
+      "loss": 0.0383,
+      "step": 1128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454463642380125e-05,
+      "loss": 0.0743,
+      "step": 1129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453224651931638e-05,
+      "loss": 0.0808,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451984295661467e-05,
+      "loss": 0.103,
+      "step": 1131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945074257374882e-05,
+      "loss": 0.027,
+      "step": 1132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449499486373113e-05,
+      "loss": 0.0645,
+      "step": 1133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448255033713936e-05,
+      "loss": 0.0634,
+      "step": 1134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447009215951097e-05,
+      "loss": 0.0311,
+      "step": 1135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445762033264595e-05,
+      "loss": 0.059,
+      "step": 1136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444513485834624e-05,
+      "loss": 0.0238,
+      "step": 1137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944326357384158e-05,
+      "loss": 0.1081,
+      "step": 1138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442012297466046e-05,
+      "loss": 0.1518,
+      "step": 1139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440759656888812e-05,
+      "loss": 0.0433,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439505652290867e-05,
+      "loss": 0.0928,
+      "step": 1141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438250283853385e-05,
+      "loss": 0.0431,
+      "step": 1142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943699355175775e-05,
+      "loss": 0.0146,
+      "step": 1143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435735456185535e-05,
+      "loss": 0.0621,
+      "step": 1144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943447599731851e-05,
+      "loss": 0.0057,
+      "step": 1145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943321517533865e-05,
+      "loss": 0.0182,
+      "step": 1146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431952990428112e-05,
+      "loss": 0.0618,
+      "step": 1147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943068944276927e-05,
+      "loss": 0.0155,
+      "step": 1148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429424532544677e-05,
+      "loss": 0.0178,
+      "step": 1149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942815825993709e-05,
+      "loss": 0.0502,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426890625129466e-05,
+      "loss": 0.0936,
+      "step": 1151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425621628304957e-05,
+      "loss": 0.008,
+      "step": 1152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94243512696469e-05,
+      "loss": 0.1416,
+      "step": 1153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942307954933885e-05,
+      "loss": 0.0091,
+      "step": 1154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421806467564546e-05,
+      "loss": 0.0194,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942053202450792e-05,
+      "loss": 0.0444,
+      "step": 1156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419256220353113e-05,
+      "loss": 0.1117,
+      "step": 1157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941797905528445e-05,
+      "loss": 0.0753,
+      "step": 1158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941670052948646e-05,
+      "loss": 0.0265,
+      "step": 1159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415420643143875e-05,
+      "loss": 0.0266,
+      "step": 1160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414139396441602e-05,
+      "loss": 0.055,
+      "step": 1161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941285678956477e-05,
+      "loss": 0.01,
+      "step": 1162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411572822698686e-05,
+      "loss": 0.0487,
+      "step": 1163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410287496028857e-05,
+      "loss": 0.0068,
+      "step": 1164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409000809741e-05,
+      "loss": 0.047,
+      "step": 1165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407712764021015e-05,
+      "loss": 0.0431,
+      "step": 1166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406423359054997e-05,
+      "loss": 0.0408,
+      "step": 1167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940513259502924e-05,
+      "loss": 0.0157,
+      "step": 1168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403840472130247e-05,
+      "loss": 0.0366,
+      "step": 1169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402546990544694e-05,
+      "loss": 0.1244,
+      "step": 1170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401252150459478e-05,
+      "loss": 0.0102,
+      "step": 1171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399955952061673e-05,
+      "loss": 0.0993,
+      "step": 1172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398658395538558e-05,
+      "loss": 0.0621,
+      "step": 1173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397359481077603e-05,
+      "loss": 0.1497,
+      "step": 1174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396059208866483e-05,
+      "loss": 0.0506,
+      "step": 1175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394757579093063e-05,
+      "loss": 0.1699,
+      "step": 1176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393454591945405e-05,
+      "loss": 0.0377,
+      "step": 1177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392150247611766e-05,
+      "loss": 0.1443,
+      "step": 1178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390844546280602e-05,
+      "loss": 0.05,
+      "step": 1179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389537488140563e-05,
+      "loss": 0.0106,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388229073380495e-05,
+      "loss": 0.0814,
+      "step": 1181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938691930218944e-05,
+      "loss": 0.0609,
+      "step": 1182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385608174756637e-05,
+      "loss": 0.005,
+      "step": 1183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384295691271523e-05,
+      "loss": 0.0417,
+      "step": 1184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382981851923724e-05,
+      "loss": 0.0133,
+      "step": 1185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381666656903068e-05,
+      "loss": 0.0048,
+      "step": 1186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380350106399578e-05,
+      "loss": 0.0505,
+      "step": 1187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379032200603473e-05,
+      "loss": 0.0446,
+      "step": 1188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377712939705164e-05,
+      "loss": 0.0374,
+      "step": 1189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376392323895263e-05,
+      "loss": 0.0091,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375070353364573e-05,
+      "loss": 0.0083,
+      "step": 1191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373747028304096e-05,
+      "loss": 0.0029,
+      "step": 1192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937242234890503e-05,
+      "loss": 0.0345,
+      "step": 1193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371096315358763e-05,
+      "loss": 0.0744,
+      "step": 1194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936976892785689e-05,
+      "loss": 0.0814,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368440186591192e-05,
+      "loss": 0.0433,
+      "step": 1196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367110091753645e-05,
+      "loss": 0.0098,
+      "step": 1197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936577864353643e-05,
+      "loss": 0.0092,
+      "step": 1198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364445842131906e-05,
+      "loss": 0.0418,
+      "step": 1199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363111687732655e-05,
+      "loss": 0.0518,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361776180531428e-05,
+      "loss": 0.0197,
+      "step": 1201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936043932072118e-05,
+      "loss": 0.063,
+      "step": 1202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359101108495073e-05,
+      "loss": 0.0387,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357761544046446e-05,
+      "loss": 0.0688,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935642062756885e-05,
+      "loss": 0.1236,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935507835925601e-05,
+      "loss": 0.0101,
+      "step": 1206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353734739301876e-05,
+      "loss": 0.1364,
+      "step": 1207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352389767900566e-05,
+      "loss": 0.0781,
+      "step": 1208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935104344524641e-05,
+      "loss": 0.0428,
+      "step": 1209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934969577153392e-05,
+      "loss": 0.1371,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934834674695782e-05,
+      "loss": 0.0599,
+      "step": 1211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346996371713015e-05,
+      "loss": 0.0065,
+      "step": 1212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934564464599461e-05,
+      "loss": 0.0644,
+      "step": 1213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934429156999791e-05,
+      "loss": 0.0622,
+      "step": 1214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342937143918398e-05,
+      "loss": 0.0376,
+      "step": 1215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934158136795178e-05,
+      "loss": 0.0233,
+      "step": 1216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934022424229393e-05,
+      "loss": 0.0418,
+      "step": 1217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933886576714093e-05,
+      "loss": 0.0078,
+      "step": 1218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337505942689065e-05,
+      "loss": 0.1558,
+      "step": 1219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336144769134796e-05,
+      "loss": 0.0703,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334782246674786e-05,
+      "loss": 0.0917,
+      "step": 1221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333418375505904e-05,
+      "loss": 0.0376,
+      "step": 1222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332053155825194e-05,
+      "loss": 0.0126,
+      "step": 1223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330686587829918e-05,
+      "loss": 0.0275,
+      "step": 1224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932931867171751e-05,
+      "loss": 0.1289,
+      "step": 1225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327949407685617e-05,
+      "loss": 0.0356,
+      "step": 1226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326578795932068e-05,
+      "loss": 0.0459,
+      "step": 1227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325206836654895e-05,
+      "loss": 0.0148,
+      "step": 1228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932383353005232e-05,
+      "loss": 0.0427,
+      "step": 1229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322458876322762e-05,
+      "loss": 0.06,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321082875664834e-05,
+      "loss": 0.1969,
+      "step": 1231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931970552827734e-05,
+      "loss": 0.1039,
+      "step": 1232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931832683435929e-05,
+      "loss": 0.0453,
+      "step": 1233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316946794109875e-05,
+      "loss": 0.0531,
+      "step": 1234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315565407728484e-05,
+      "loss": 0.0432,
+      "step": 1235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314182675414704e-05,
+      "loss": 0.0684,
+      "step": 1236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931279859736832e-05,
+      "loss": 0.0341,
+      "step": 1237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93114131737893e-05,
+      "loss": 0.0737,
+      "step": 1238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310026404877816e-05,
+      "loss": 0.0654,
+      "step": 1239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308638290834235e-05,
+      "loss": 0.0277,
+      "step": 1240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307248831859106e-05,
+      "loss": 0.0571,
+      "step": 1241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305858028153186e-05,
+      "loss": 0.0193,
+      "step": 1242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930446587991742e-05,
+      "loss": 0.0946,
+      "step": 1243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303072387352952e-05,
+      "loss": 0.0111,
+      "step": 1244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301677550661115e-05,
+      "loss": 0.0966,
+      "step": 1245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300281370043433e-05,
+      "loss": 0.0857,
+      "step": 1246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298883845701638e-05,
+      "loss": 0.0441,
+      "step": 1247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929748497783764e-05,
+      "loss": 0.0581,
+      "step": 1248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929608476665355e-05,
+      "loss": 0.0217,
+      "step": 1249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929468321235168e-05,
+      "loss": 0.0897,
+      "step": 1250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293280315134524e-05,
+      "loss": 0.0318,
+      "step": 1251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291876075204774e-05,
+      "loss": 0.1634,
+      "step": 1252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290470492765326e-05,
+      "loss": 0.0117,
+      "step": 1253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289063568019254e-05,
+      "loss": 0.2278,
+      "step": 1254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928765530116983e-05,
+      "loss": 0.0071,
+      "step": 1255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928624569242054e-05,
+      "loss": 0.0766,
+      "step": 1256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284834741975026e-05,
+      "loss": 0.0613,
+      "step": 1257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283422450037163e-05,
+      "loss": 0.0632,
+      "step": 1258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928200881681099e-05,
+      "loss": 0.0531,
+      "step": 1259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280593842500756e-05,
+      "loss": 0.0096,
+      "step": 1260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92791775273109e-05,
+      "loss": 0.003,
+      "step": 1261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277759871446048e-05,
+      "loss": 0.0087,
+      "step": 1262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927634087511104e-05,
+      "loss": 0.0839,
+      "step": 1263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927492053851088e-05,
+      "loss": 0.136,
+      "step": 1264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927349886185079e-05,
+      "loss": 0.0219,
+      "step": 1265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272075845336177e-05,
+      "loss": 0.0995,
+      "step": 1266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927065148917264e-05,
+      "loss": 0.0189,
+      "step": 1267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926922579356597e-05,
+      "loss": 0.0234,
+      "step": 1268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267798758722156e-05,
+      "loss": 0.1009,
+      "step": 1269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926637038484738e-05,
+      "loss": 0.0158,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264940672148018e-05,
+      "loss": 0.0036,
+      "step": 1271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263509620830634e-05,
+      "loss": 0.0406,
+      "step": 1272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262077231101993e-05,
+      "loss": 0.0656,
+      "step": 1273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260643503169047e-05,
+      "loss": 0.0173,
+      "step": 1274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259208437238948e-05,
+      "loss": 0.0833,
+      "step": 1275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257772033519032e-05,
+      "loss": 0.0496,
+      "step": 1276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256334292216836e-05,
+      "loss": 0.0966,
+      "step": 1277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254895213540086e-05,
+      "loss": 0.0445,
+      "step": 1278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253454797696706e-05,
+      "loss": 0.0331,
+      "step": 1279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925201304489481e-05,
+      "loss": 0.0855,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250569955342705e-05,
+      "loss": 0.1124,
+      "step": 1281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924912552924889e-05,
+      "loss": 0.0106,
+      "step": 1282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924767976682206e-05,
+      "loss": 0.0348,
+      "step": 1283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246232668271102e-05,
+      "loss": 0.1418,
+      "step": 1284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244784233805097e-05,
+      "loss": 0.0084,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243334463633315e-05,
+      "loss": 0.0129,
+      "step": 1286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241883357965224e-05,
+      "loss": 0.0406,
+      "step": 1287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924043091701048e-05,
+      "loss": 0.0623,
+      "step": 1288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923897714097894e-05,
+      "loss": 0.1673,
+      "step": 1289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237522030080644e-05,
+      "loss": 0.0289,
+      "step": 1290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923606558452583e-05,
+      "loss": 0.1258,
+      "step": 1291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923460780452493e-05,
+      "loss": 0.0522,
+      "step": 1292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233148690288567e-05,
+      "loss": 0.043,
+      "step": 1293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231688242027557e-05,
+      "loss": 0.034,
+      "step": 1294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230226459952902e-05,
+      "loss": 0.0154,
+      "step": 1295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228763344275813e-05,
+      "loss": 0.0349,
+      "step": 1296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922729889520768e-05,
+      "loss": 0.0061,
+      "step": 1297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225833112960092e-05,
+      "loss": 0.0725,
+      "step": 1298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224365997744824e-05,
+      "loss": 0.0024,
+      "step": 1299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922289754977385e-05,
+      "loss": 0.0323,
+      "step": 1300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221427769259333e-05,
+      "loss": 0.0791,
+      "step": 1301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219956656413633e-05,
+      "loss": 0.1366,
+      "step": 1302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218484211449295e-05,
+      "loss": 0.0059,
+      "step": 1303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921701043457907e-05,
+      "loss": 0.039,
+      "step": 1304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215535326015878e-05,
+      "loss": 0.0538,
+      "step": 1305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921405888597286e-05,
+      "loss": 0.0025,
+      "step": 1306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212581114663325e-05,
+      "loss": 0.041,
+      "step": 1307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921110201230079e-05,
+      "loss": 0.0618,
+      "step": 1308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209621579098957e-05,
+      "loss": 0.0429,
+      "step": 1309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208139815271726e-05,
+      "loss": 0.0061,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206656721033178e-05,
+      "loss": 0.0178,
+      "step": 1311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92051722965976e-05,
+      "loss": 0.1531,
+      "step": 1312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920368654217946e-05,
+      "loss": 0.0034,
+      "step": 1313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202199457993423e-05,
+      "loss": 0.129,
+      "step": 1314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920071104425435e-05,
+      "loss": 0.007,
+      "step": 1315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199221301177287e-05,
+      "loss": 0.0016,
+      "step": 1316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197730228977478e-05,
+      "loss": 0.0045,
+      "step": 1317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196237827870354e-05,
+      "loss": 0.0075,
+      "step": 1318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919474409807154e-05,
+      "loss": 0.1297,
+      "step": 1319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193249039796856e-05,
+      "loss": 0.1171,
+      "step": 1320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919175265326231e-05,
+      "loss": 0.1069,
+      "step": 1321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91902549386841e-05,
+      "loss": 0.1541,
+      "step": 1322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188755896278618e-05,
+      "loss": 0.009,
+      "step": 1323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187255526262456e-05,
+      "loss": 0.0079,
+      "step": 1324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185753828852384e-05,
+      "loss": 0.0942,
+      "step": 1325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184250804265375e-05,
+      "loss": 0.0383,
+      "step": 1326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182746452718586e-05,
+      "loss": 0.1237,
+      "step": 1327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181240774429374e-05,
+      "loss": 0.043,
+      "step": 1328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179733769615273e-05,
+      "loss": 0.0996,
+      "step": 1329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178225438494026e-05,
+      "loss": 0.0488,
+      "step": 1330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176715781283556e-05,
+      "loss": 0.0069,
+      "step": 1331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175204798201982e-05,
+      "loss": 0.0078,
+      "step": 1332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173692489467618e-05,
+      "loss": 0.1951,
+      "step": 1333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917217885529896e-05,
+      "loss": 0.0328,
+      "step": 1334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170663895914705e-05,
+      "loss": 0.0337,
+      "step": 1335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169147611533737e-05,
+      "loss": 0.0059,
+      "step": 1336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916763000237513e-05,
+      "loss": 0.0662,
+      "step": 1337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166111068658152e-05,
+      "loss": 0.0182,
+      "step": 1338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164590810602262e-05,
+      "loss": 0.1101,
+      "step": 1339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163069228427108e-05,
+      "loss": 0.0171,
+      "step": 1340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916154632235254e-05,
+      "loss": 0.0142,
+      "step": 1341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160022092598577e-05,
+      "loss": 0.0611,
+      "step": 1342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158496539385453e-05,
+      "loss": 0.0614,
+      "step": 1343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915696966293358e-05,
+      "loss": 0.1338,
+      "step": 1344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155441463463562e-05,
+      "loss": 0.049,
+      "step": 1345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91539119411962e-05,
+      "loss": 0.0094,
+      "step": 1346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152381096352486e-05,
+      "loss": 0.0702,
+      "step": 1347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915084892915359e-05,
+      "loss": 0.0327,
+      "step": 1348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914931543982089e-05,
+      "loss": 0.1008,
+      "step": 1349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147780628575945e-05,
+      "loss": 0.0343,
+      "step": 1350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914624449564051e-05,
+      "loss": 0.0041,
+      "step": 1351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144707041236524e-05,
+      "loss": 0.0547,
+      "step": 1352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143168265586126e-05,
+      "loss": 0.0625,
+      "step": 1353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914162816891164e-05,
+      "loss": 0.0069,
+      "step": 1354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140086751435584e-05,
+      "loss": 0.0121,
+      "step": 1355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913854401338066e-05,
+      "loss": 0.0746,
+      "step": 1356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136999954969773e-05,
+      "loss": 0.0202,
+      "step": 1357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.0409,
+      "step": 1358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133907877972647e-05,
+      "loss": 0.1332,
+      "step": 1359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132359859833158e-05,
+      "loss": 0.0114,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91308105222312e-05,
+      "loss": 0.0343,
+      "step": 1361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912925986539063e-05,
+      "loss": 0.0229,
+      "step": 1362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127707889535484e-05,
+      "loss": 0.0406,
+      "step": 1363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912615459489e-05,
+      "loss": 0.127,
+      "step": 1364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91245999816786e-05,
+      "loss": 0.033,
+      "step": 1365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123044050125892e-05,
+      "loss": 0.0171,
+      "step": 1366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121486800456684e-05,
+      "loss": 0.0381,
+      "step": 1367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119928232895977e-05,
+      "loss": 0.069,
+      "step": 1368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911836834766895e-05,
+      "loss": 0.0046,
+      "step": 1369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9116807145000977e-05,
+      "loss": 0.0567,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115244625117627e-05,
+      "loss": 0.0101,
+      "step": 1371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113680788244654e-05,
+      "loss": 0.0118,
+      "step": 1372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112115634608008e-05,
+      "loss": 0.0274,
+      "step": 1373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110549164433822e-05,
+      "loss": 0.0519,
+      "step": 1374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108981377948426e-05,
+      "loss": 0.0036,
+      "step": 1375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107412275378335e-05,
+      "loss": 0.0045,
+      "step": 1376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105841856950254e-05,
+      "loss": 0.0554,
+      "step": 1377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104270122891085e-05,
+      "loss": 0.0026,
+      "step": 1378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910269707342791e-05,
+      "loss": 0.0948,
+      "step": 1379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101122708788015e-05,
+      "loss": 0.0583,
+      "step": 1380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909954702919886e-05,
+      "loss": 0.0142,
+      "step": 1381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90979700348881e-05,
+      "loss": 0.0992,
+      "step": 1382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096391726083592e-05,
+      "loss": 0.0564,
+      "step": 1383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909481210301337e-05,
+      "loss": 0.0611,
+      "step": 1384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909323116590566e-05,
+      "loss": 0.005,
+      "step": 1385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091648914988873e-05,
+      "loss": 0.1153,
+      "step": 1386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909006535049163e-05,
+      "loss": 0.0182,
+      "step": 1387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908848047264271e-05,
+      "loss": 0.0207,
+      "step": 1388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086894281671115e-05,
+      "loss": 0.0571,
+      "step": 1389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085306777806013e-05,
+      "loss": 0.0549,
+      "step": 1390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083717961276775e-05,
+      "loss": 0.0131,
+      "step": 1391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082127832312953e-05,
+      "loss": 0.1139,
+      "step": 1392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908053639114429e-05,
+      "loss": 0.0434,
+      "step": 1393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078943638000726e-05,
+      "loss": 0.0334,
+      "step": 1394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907734957311238e-05,
+      "loss": 0.0287,
+      "step": 1395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075754196709574e-05,
+      "loss": 0.0025,
+      "step": 1396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90741575090228e-05,
+      "loss": 0.0536,
+      "step": 1397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907255951028276e-05,
+      "loss": 0.0029,
+      "step": 1398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070960200720332e-05,
+      "loss": 0.15,
+      "step": 1399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906935958056659e-05,
+      "loss": 0.0697,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067757650052792e-05,
+      "loss": 0.0121,
+      "step": 1401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066154409410388e-05,
+      "loss": 0.1701,
+      "step": 1402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064549858871023e-05,
+      "loss": 0.0059,
+      "step": 1403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906294399866652e-05,
+      "loss": 0.1501,
+      "step": 1404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90613368290289e-05,
+      "loss": 0.0175,
+      "step": 1405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.905972835019037e-05,
+      "loss": 0.0529,
+      "step": 1406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058118562383326e-05,
+      "loss": 0.0389,
+      "step": 1407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056507465840354e-05,
+      "loss": 0.0237,
+      "step": 1408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054895060794224e-05,
+      "loss": 0.0107,
+      "step": 1409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053281347477907e-05,
+      "loss": 0.0507,
+      "step": 1410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905166632612455e-05,
+      "loss": 0.0747,
+      "step": 1411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90500499969675e-05,
+      "loss": 0.0119,
+      "step": 1412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904843236024028e-05,
+      "loss": 0.0924,
+      "step": 1413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904681341617662e-05,
+      "loss": 0.0248,
+      "step": 1414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045193165010417e-05,
+      "loss": 0.0243,
+      "step": 1415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043571606975776e-05,
+      "loss": 0.0516,
+      "step": 1416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041948742306982e-05,
+      "loss": 0.0177,
+      "step": 1417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904032457123851e-05,
+      "loss": 0.0508,
+      "step": 1418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038699094005016e-05,
+      "loss": 0.008,
+      "step": 1419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037072310841364e-05,
+      "loss": 0.0037,
+      "step": 1420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903544422198259e-05,
+      "loss": 0.1236,
+      "step": 1421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033814827663925e-05,
+      "loss": 0.0844,
+      "step": 1422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032184128120777e-05,
+      "loss": 0.0066,
+      "step": 1423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903055212358877e-05,
+      "loss": 0.08,
+      "step": 1424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902891881430369e-05,
+      "loss": 0.0654,
+      "step": 1425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902728420050152e-05,
+      "loss": 0.0503,
+      "step": 1426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025648282418438e-05,
+      "loss": 0.0762,
+      "step": 1427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90240110602908e-05,
+      "loss": 0.0612,
+      "step": 1428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022372534355152e-05,
+      "loss": 0.0871,
+      "step": 1429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902073270484824e-05,
+      "loss": 0.0021,
+      "step": 1430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019091572006988e-05,
+      "loss": 0.0009,
+      "step": 1431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017449136068507e-05,
+      "loss": 0.1017,
+      "step": 1432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90158053972701e-05,
+      "loss": 0.0792,
+      "step": 1433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901416035584926e-05,
+      "loss": 0.0035,
+      "step": 1434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012514012043663e-05,
+      "loss": 0.0073,
+      "step": 1435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010866366091182e-05,
+      "loss": 0.1066,
+      "step": 1436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9009217418229867e-05,
+      "loss": 0.0021,
+      "step": 1437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900756716869796e-05,
+      "loss": 0.0039,
+      "step": 1438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90059156177339e-05,
+      "loss": 0.0064,
+      "step": 1439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90042627655763e-05,
+      "loss": 0.0842,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900260861246397e-05,
+      "loss": 0.0179,
+      "step": 1441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000953158635903e-05,
+      "loss": 0.0018,
+      "step": 1442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8999296404331288e-05,
+      "loss": 0.0842,
+      "step": 1443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899763834978949e-05,
+      "loss": 0.1271,
+      "step": 1444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899597899525007e-05,
+      "loss": 0.0355,
+      "step": 1445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899431834095278e-05,
+      "loss": 0.0011,
+      "step": 1446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992656387137552e-05,
+      "loss": 0.0311,
+      "step": 1447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89909931340445e-05,
+      "loss": 0.037,
+      "step": 1448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989328581913947e-05,
+      "loss": 0.0033,
+      "step": 1449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987662730986388e-05,
+      "loss": 0.0191,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89859955815025e-05,
+      "loss": 0.0278,
+      "step": 1451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898432713370317e-05,
+      "loss": 0.1207,
+      "step": 1452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982657387829447e-05,
+      "loss": 0.226,
+      "step": 1453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898098634412258e-05,
+      "loss": 0.0147,
+      "step": 1454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979314002824013e-05,
+      "loss": 0.0286,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977640364175367e-05,
+      "loss": 0.2398,
+      "step": 1456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975965428418447e-05,
+      "loss": 0.1473,
+      "step": 1457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974289195795256e-05,
+      "loss": 0.0881,
+      "step": 1458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897261166654798e-05,
+      "loss": 0.0138,
+      "step": 1459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970932840918988e-05,
+      "loss": 0.0856,
+      "step": 1460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969252719150847e-05,
+      "loss": 0.091,
+      "step": 1461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89675713014863e-05,
+      "loss": 0.0905,
+      "step": 1462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896588858816828e-05,
+      "loss": 0.0169,
+      "step": 1463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896420457943991e-05,
+      "loss": 0.0071,
+      "step": 1464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962519275544502e-05,
+      "loss": 0.0056,
+      "step": 1465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896083267672555e-05,
+      "loss": 0.0073,
+      "step": 1466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959144783226738e-05,
+      "loss": 0.0781,
+      "step": 1467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957455595291936e-05,
+      "loss": 0.0228,
+      "step": 1468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89557651131652e-05,
+      "loss": 0.0661,
+      "step": 1469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895407333709078e-05,
+      "loss": 0.0032,
+      "step": 1470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89523802673131e-05,
+      "loss": 0.0025,
+      "step": 1471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895068590407678e-05,
+      "loss": 0.0044,
+      "step": 1472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894899024762663e-05,
+      "loss": 0.0052,
+      "step": 1473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947293298207637e-05,
+      "loss": 0.0601,
+      "step": 1474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945595056064982e-05,
+      "loss": 0.0022,
+      "step": 1475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894389552144403e-05,
+      "loss": 0.0835,
+      "step": 1476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942194694590334e-05,
+      "loss": 0.0939,
+      "step": 1477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894049257574963e-05,
+      "loss": 0.061,
+      "step": 1478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893878916516785e-05,
+      "loss": 0.0423,
+      "step": 1479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937084463091095e-05,
+      "loss": 0.0031,
+      "step": 1480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893537846976568e-05,
+      "loss": 0.0085,
+      "step": 1481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893367118543808e-05,
+      "loss": 0.005,
+      "step": 1482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931962610354964e-05,
+      "loss": 0.0042,
+      "step": 1483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930252744763198e-05,
+      "loss": 0.0038,
+      "step": 1484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928541588909827e-05,
+      "loss": 0.029,
+      "step": 1485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926829143042075e-05,
+      "loss": 0.0451,
+      "step": 1486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925115407407368e-05,
+      "loss": 0.2109,
+      "step": 1487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923400382253307e-05,
+      "loss": 0.0545,
+      "step": 1488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921684067827676e-05,
+      "loss": 0.0535,
+      "step": 1489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919966464378463e-05,
+      "loss": 0.0085,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918247572153822e-05,
+      "loss": 0.0606,
+      "step": 1491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916527391402112e-05,
+      "loss": 0.0245,
+      "step": 1492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914805922371857e-05,
+      "loss": 0.0132,
+      "step": 1493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913083165311785e-05,
+      "loss": 0.1416,
+      "step": 1494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911359120470804e-05,
+      "loss": 0.0331,
+      "step": 1495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909633788098004e-05,
+      "loss": 0.0549,
+      "step": 1496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907907168442664e-05,
+      "loss": 0.0231,
+      "step": 1497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906179261754253e-05,
+      "loss": 0.0098,
+      "step": 1498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904450068282424e-05,
+      "loss": 0.0344,
+      "step": 1499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890271958827701e-05,
+      "loss": 0.02,
+      "step": 1500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900987821988038e-05,
+      "loss": 0.0045,
+      "step": 1501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889925476966571e-05,
+      "loss": 0.0736,
+      "step": 1502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897520431560435e-05,
+      "loss": 0.0024,
+      "step": 1503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889578480792278e-05,
+      "loss": 0.0124,
+      "step": 1504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889404789900352e-05,
+      "loss": 0.0518,
+      "step": 1505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88923097050536e-05,
+      "loss": 0.047,
+      "step": 1506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890570226324164e-05,
+      "loss": 0.014,
+      "step": 1507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888829463066533e-05,
+      "loss": 0.0833,
+      "step": 1508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887087415532218e-05,
+      "loss": 0.0194,
+      "step": 1509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885344083972912e-05,
+      "loss": 0.0536,
+      "step": 1510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883599468640498e-05,
+      "loss": 0.1942,
+      "step": 1511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881853569787037e-05,
+      "loss": 0.1239,
+      "step": 1512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880106387664783e-05,
+      "loss": 0.0038,
+      "step": 1513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878357922526173e-05,
+      "loss": 0.018,
+      "step": 1514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887660817462382e-05,
+      "loss": 0.0026,
+      "step": 1515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887485714421055e-05,
+      "loss": 0.0103,
+      "step": 1516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887310483153934e-05,
+      "loss": 0.0321,
+      "step": 1517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887135123686337e-05,
+      "loss": 0.0614,
+      "step": 1518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886959636043601e-05,
+      "loss": 0.1244,
+      "step": 1519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88678402025108e-05,
+      "loss": 0.1905,
+      "step": 1520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886608276334148e-05,
+      "loss": 0.0018,
+      "step": 1521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864324043181964e-05,
+      "loss": 0.0007,
+      "step": 1522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862564042286358e-05,
+      "loss": 0.0847,
+      "step": 1523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886080276090895e-05,
+      "loss": 0.0435,
+      "step": 1524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885904019930421e-05,
+      "loss": 0.0033,
+      "step": 1525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857276357726802e-05,
+      "loss": 0.0022,
+      "step": 1526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855511236431567e-05,
+      "loss": 0.1308,
+      "step": 1527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853744835673537e-05,
+      "loss": 0.0544,
+      "step": 1528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851977155707917e-05,
+      "loss": 0.0048,
+      "step": 1529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850208196790114e-05,
+      "loss": 0.0442,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848437959175703e-05,
+      "loss": 0.009,
+      "step": 1531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884666644312046e-05,
+      "loss": 0.0229,
+      "step": 1532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884489364888033e-05,
+      "loss": 0.0703,
+      "step": 1533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843119576711457e-05,
+      "loss": 0.0644,
+      "step": 1534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841344226870157e-05,
+      "loss": 0.0346,
+      "step": 1535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839567599612934e-05,
+      "loss": 0.0015,
+      "step": 1536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837789695196492e-05,
+      "loss": 0.0108,
+      "step": 1537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883601051387769e-05,
+      "loss": 0.0267,
+      "step": 1538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834230055913596e-05,
+      "loss": 0.1796,
+      "step": 1539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832448321561453e-05,
+      "loss": 0.072,
+      "step": 1540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830665311078694e-05,
+      "loss": 0.0652,
+      "step": 1541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828881024722923e-05,
+      "loss": 0.0037,
+      "step": 1542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827095462751945e-05,
+      "loss": 0.0338,
+      "step": 1543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882530862542374e-05,
+      "loss": 0.014,
+      "step": 1544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823520512996473e-05,
+      "loss": 0.0182,
+      "step": 1545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821731125728493e-05,
+      "loss": 0.0109,
+      "step": 1546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819940463878338e-05,
+      "loss": 0.0428,
+      "step": 1547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881814852770472e-05,
+      "loss": 0.0516,
+      "step": 1548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881635531746655e-05,
+      "loss": 0.1418,
+      "step": 1549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814560833422915e-05,
+      "loss": 0.0075,
+      "step": 1550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812765075833076e-05,
+      "loss": 0.0366,
+      "step": 1551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810968044956493e-05,
+      "loss": 0.0132,
+      "step": 1552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880916974105281e-05,
+      "loss": 0.0068,
+      "step": 1553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880737016438184e-05,
+      "loss": 0.1852,
+      "step": 1554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88055693152036e-05,
+      "loss": 0.0062,
+      "step": 1555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803767193778273e-05,
+      "loss": 0.0031,
+      "step": 1556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801963800366236e-05,
+      "loss": 0.0054,
+      "step": 1557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800159135228046e-05,
+      "loss": 0.0505,
+      "step": 1558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8798353198624443e-05,
+      "loss": 0.0098,
+      "step": 1559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796545990816355e-05,
+      "loss": 0.0025,
+      "step": 1560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879473751206489e-05,
+      "loss": 0.02,
+      "step": 1561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792927762631347e-05,
+      "loss": 0.0019,
+      "step": 1562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791116742777192e-05,
+      "loss": 0.0683,
+      "step": 1563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789304452764095e-05,
+      "loss": 0.0406,
+      "step": 1564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878749089285389e-05,
+      "loss": 0.0519,
+      "step": 1565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878567606330861e-05,
+      "loss": 0.0975,
+      "step": 1566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783859964390466e-05,
+      "loss": 0.0187,
+      "step": 1567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782042596361846e-05,
+      "loss": 0.0384,
+      "step": 1568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780223959485332e-05,
+      "loss": 0.0089,
+      "step": 1569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778404054023687e-05,
+      "loss": 0.0795,
+      "step": 1570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776582880239847e-05,
+      "loss": 0.032,
+      "step": 1571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774760438396944e-05,
+      "loss": 0.0729,
+      "step": 1572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772936728758284e-05,
+      "loss": 0.0031,
+      "step": 1573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771111751587368e-05,
+      "loss": 0.0078,
+      "step": 1574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769285507147867e-05,
+      "loss": 0.0186,
+      "step": 1575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876745799570364e-05,
+      "loss": 0.1051,
+      "step": 1576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765629217518733e-05,
+      "loss": 0.1167,
+      "step": 1577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876379917285737e-05,
+      "loss": 0.0354,
+      "step": 1578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761967861983957e-05,
+      "loss": 0.0664,
+      "step": 1579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760135285163094e-05,
+      "loss": 0.0012,
+      "step": 1580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758301442659546e-05,
+      "loss": 0.0876,
+      "step": 1581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756466334738276e-05,
+      "loss": 0.0092,
+      "step": 1582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754629961664417e-05,
+      "loss": 0.0063,
+      "step": 1583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752792323703303e-05,
+      "loss": 0.0795,
+      "step": 1584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750953421120436e-05,
+      "loss": 0.0603,
+      "step": 1585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749113254181498e-05,
+      "loss": 0.103,
+      "step": 1586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747271823152368e-05,
+      "loss": 0.0847,
+      "step": 1587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8745429128299096e-05,
+      "loss": 0.0057,
+      "step": 1588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743585169887918e-05,
+      "loss": 0.1065,
+      "step": 1589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741739948185256e-05,
+      "loss": 0.0497,
+      "step": 1590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873989346345771e-05,
+      "loss": 0.0589,
+      "step": 1591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738045715972063e-05,
+      "loss": 0.0016,
+      "step": 1592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873619670599528e-05,
+      "loss": 0.06,
+      "step": 1593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734346433794517e-05,
+      "loss": 0.0151,
+      "step": 1594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732494899637096e-05,
+      "loss": 0.1618,
+      "step": 1595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730642103790537e-05,
+      "loss": 0.0228,
+      "step": 1596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872878804652253e-05,
+      "loss": 0.0063,
+      "step": 1597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872693272810096e-05,
+      "loss": 0.0278,
+      "step": 1598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725076148793882e-05,
+      "loss": 0.0097,
+      "step": 1599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872321830886954e-05,
+      "loss": 0.0048,
+      "step": 1600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872135920859636e-05,
+      "loss": 0.0345,
+      "step": 1601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719498848242948e-05,
+      "loss": 0.0712,
+      "step": 1602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871763722807809e-05,
+      "loss": 0.0007,
+      "step": 1603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715774348370763e-05,
+      "loss": 0.0242,
+      "step": 1604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713910209390113e-05,
+      "loss": 0.0056,
+      "step": 1605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871204481140548e-05,
+      "loss": 0.0405,
+      "step": 1606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710178154686376e-05,
+      "loss": 0.0456,
+      "step": 1607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8708310239502502e-05,
+      "loss": 0.0871,
+      "step": 1608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870644106612374e-05,
+      "loss": 0.0292,
+      "step": 1609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704570634820147e-05,
+      "loss": 0.0747,
+      "step": 1610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870269894586197e-05,
+      "loss": 0.0311,
+      "step": 1611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700825999519637e-05,
+      "loss": 0.1334,
+      "step": 1612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698951796063754e-05,
+      "loss": 0.0463,
+      "step": 1613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697076335765106e-05,
+      "loss": 0.0292,
+      "step": 1614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695199618894663e-05,
+      "loss": 0.0276,
+      "step": 1615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693321645723584e-05,
+      "loss": 0.0014,
+      "step": 1616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691442416523197e-05,
+      "loss": 0.0162,
+      "step": 1617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689561931565017e-05,
+      "loss": 0.0391,
+      "step": 1618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687680191120746e-05,
+      "loss": 0.1518,
+      "step": 1619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685797195462253e-05,
+      "loss": 0.0162,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683912944861606e-05,
+      "loss": 0.0106,
+      "step": 1621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682027439591033e-05,
+      "loss": 0.0429,
+      "step": 1622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680140679922973e-05,
+      "loss": 0.002,
+      "step": 1623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678252666130016e-05,
+      "loss": 0.0748,
+      "step": 1624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676363398484948e-05,
+      "loss": 0.0474,
+      "step": 1625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867447287726074e-05,
+      "loss": 0.0017,
+      "step": 1626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672581102730537e-05,
+      "loss": 0.0517,
+      "step": 1627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867068807516766e-05,
+      "loss": 0.0063,
+      "step": 1628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668793794845622e-05,
+      "loss": 0.0037,
+      "step": 1629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666898262038113e-05,
+      "loss": 0.0651,
+      "step": 1630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665001477019006e-05,
+      "loss": 0.0008,
+      "step": 1631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663103440062347e-05,
+      "loss": 0.0163,
+      "step": 1632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8661204151442374e-05,
+      "loss": 0.097,
+      "step": 1633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659303611433492e-05,
+      "loss": 0.0718,
+      "step": 1634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657401820310306e-05,
+      "loss": 0.0011,
+      "step": 1635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865549877834758e-05,
+      "loss": 0.0055,
+      "step": 1636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653594485820278e-05,
+      "loss": 0.0809,
+      "step": 1637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865168894300353e-05,
+      "loss": 0.0331,
+      "step": 1638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649782150172656e-05,
+      "loss": 0.0013,
+      "step": 1639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647874107603156e-05,
+      "loss": 0.0023,
+      "step": 1640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86459648155707e-05,
+      "loss": 0.0256,
+      "step": 1641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644054274351153e-05,
+      "loss": 0.0389,
+      "step": 1642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864214248422055e-05,
+      "loss": 0.0547,
+      "step": 1643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864022944545512e-05,
+      "loss": 0.0089,
+      "step": 1644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863831515833125e-05,
+      "loss": 0.1317,
+      "step": 1645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636399623125524e-05,
+      "loss": 0.0015,
+      "step": 1646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863448284011471e-05,
+      "loss": 0.0115,
+      "step": 1647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863256480957574e-05,
+      "loss": 0.0741,
+      "step": 1648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630645531785737e-05,
+      "loss": 0.0027,
+      "step": 1649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8628725007022007e-05,
+      "loss": 0.0768,
+      "step": 1650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626803235562025e-05,
+      "loss": 0.0757,
+      "step": 1651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862488021768346e-05,
+      "loss": 0.0743,
+      "step": 1652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622955953664144e-05,
+      "loss": 0.0055,
+      "step": 1653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8621030443782105e-05,
+      "loss": 0.0167,
+      "step": 1654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619103688315542e-05,
+      "loss": 0.0814,
+      "step": 1655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617175687542844e-05,
+      "loss": 0.0745,
+      "step": 1656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861524644174256e-05,
+      "loss": 0.0453,
+      "step": 1657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613315951193444e-05,
+      "loss": 0.0041,
+      "step": 1658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611384216174408e-05,
+      "loss": 0.1189,
+      "step": 1659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609451236964556e-05,
+      "loss": 0.0559,
+      "step": 1660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8607517013843172e-05,
+      "loss": 0.001,
+      "step": 1661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605581547089715e-05,
+      "loss": 0.092,
+      "step": 1662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860364483698382e-05,
+      "loss": 0.0364,
+      "step": 1663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8601706883805316e-05,
+      "loss": 0.0093,
+      "step": 1664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599767687834197e-05,
+      "loss": 0.1057,
+      "step": 1665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859782724935064e-05,
+      "loss": 0.0036,
+      "step": 1666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859588556863501e-05,
+      "loss": 0.0116,
+      "step": 1667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859394264596784e-05,
+      "loss": 0.0153,
+      "step": 1668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859199848162985e-05,
+      "loss": 0.0121,
+      "step": 1669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590053075901936e-05,
+      "loss": 0.0627,
+      "step": 1670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8588106429065175e-05,
+      "loss": 0.0745,
+      "step": 1671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8586158541400824e-05,
+      "loss": 0.1019,
+      "step": 1672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584209413190314e-05,
+      "loss": 0.0072,
+      "step": 1673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858225904471526e-05,
+      "loss": 0.0087,
+      "step": 1674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858030743625746e-05,
+      "loss": 0.0295,
+      "step": 1675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578354588098877e-05,
+      "loss": 0.0036,
+      "step": 1676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576400500521673e-05,
+      "loss": 0.1128,
+      "step": 1677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574445173808172e-05,
+      "loss": 0.0023,
+      "step": 1678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572488608240888e-05,
+      "loss": 0.0185,
+      "step": 1679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570530804102505e-05,
+      "loss": 0.0035,
+      "step": 1680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568571761675893e-05,
+      "loss": 0.1163,
+      "step": 1681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85666114812441e-05,
+      "loss": 0.0724,
+      "step": 1682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856464996309035e-05,
+      "loss": 0.2106,
+      "step": 1683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856268720749805e-05,
+      "loss": 0.007,
+      "step": 1684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560723214750777e-05,
+      "loss": 0.0176,
+      "step": 1685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558757985132296e-05,
+      "loss": 0.0415,
+      "step": 1686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855679151892655e-05,
+      "loss": 0.0164,
+      "step": 1687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554823816417656e-05,
+      "loss": 0.118,
+      "step": 1688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855285487788991e-05,
+      "loss": 0.0151,
+      "step": 1689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855088470362779e-05,
+      "loss": 0.0074,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548913293915954e-05,
+      "loss": 0.0502,
+      "step": 1691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8546940649039226e-05,
+      "loss": 0.1067,
+      "step": 1692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854496676928263e-05,
+      "loss": 0.0074,
+      "step": 1693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542991654931347e-05,
+      "loss": 0.0118,
+      "step": 1694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854101530627075e-05,
+      "loss": 0.0729,
+      "step": 1695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539037723586387e-05,
+      "loss": 0.0085,
+      "step": 1696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853705890716398e-05,
+      "loss": 0.0655,
+      "step": 1697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535078857289433e-05,
+      "loss": 0.0013,
+      "step": 1698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853309757424883e-05,
+      "loss": 0.0061,
+      "step": 1699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853111505832843e-05,
+      "loss": 0.1077,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529131309814667e-05,
+      "loss": 0.0198,
+      "step": 1701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527146328994166e-05,
+      "loss": 0.0863,
+      "step": 1702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525160116153713e-05,
+      "loss": 0.0513,
+      "step": 1703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523172671580283e-05,
+      "loss": 0.0398,
+      "step": 1704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521183995561027e-05,
+      "loss": 0.0204,
+      "step": 1705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851919408838327e-05,
+      "loss": 0.0019,
+      "step": 1706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517202950334522e-05,
+      "loss": 0.0232,
+      "step": 1707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851521058170247e-05,
+      "loss": 0.1428,
+      "step": 1708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513216982774966e-05,
+      "loss": 0.0504,
+      "step": 1709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511222153840055e-05,
+      "loss": 0.0062,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850922609518595e-05,
+      "loss": 0.0563,
+      "step": 1711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850722880710105e-05,
+      "loss": 0.0346,
+      "step": 1712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850523028987393e-05,
+      "loss": 0.0786,
+      "step": 1713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503230543793334e-05,
+      "loss": 0.1091,
+      "step": 1714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501229569148188e-05,
+      "loss": 0.0207,
+      "step": 1715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499227366227605e-05,
+      "loss": 0.1459,
+      "step": 1716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8497223935320864e-05,
+      "loss": 0.0401,
+      "step": 1717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849521927671742e-05,
+      "loss": 0.0895,
+      "step": 1718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849321339070692e-05,
+      "loss": 0.0176,
+      "step": 1719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491206277579165e-05,
+      "loss": 0.0271,
+      "step": 1720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489197937624162e-05,
+      "loss": 0.0895,
+      "step": 1721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848718837113207e-05,
+      "loss": 0.0199,
+      "step": 1722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485177578393237e-05,
+      "loss": 0.0275,
+      "step": 1723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848316555969819e-05,
+      "loss": 0.0133,
+      "step": 1724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481152315337627e-05,
+      "loss": 0.0706,
+      "step": 1725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8479137845602426e-05,
+      "loss": 0.0063,
+      "step": 1726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847712215078364e-05,
+      "loss": 0.0457,
+      "step": 1727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847510523117251e-05,
+      "loss": 0.0073,
+      "step": 1728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847308708706043e-05,
+      "loss": 0.1812,
+      "step": 1729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471067718739e-05,
+      "loss": 0.0077,
+      "step": 1730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469047126499967e-05,
+      "loss": 0.0382,
+      "step": 1731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467025310635286e-05,
+      "loss": 0.0094,
+      "step": 1732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465002271437064e-05,
+      "loss": 0.0439,
+      "step": 1733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462978009197598e-05,
+      "loss": 0.0058,
+      "step": 1734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460952524209355e-05,
+      "loss": 0.0262,
+      "step": 1735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458925816764983e-05,
+      "loss": 0.028,
+      "step": 1736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8456897887157306e-05,
+      "loss": 0.0091,
+      "step": 1737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845486873567932e-05,
+      "loss": 0.0011,
+      "step": 1738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84528383626242e-05,
+      "loss": 0.0044,
+      "step": 1739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450806768285307e-05,
+      "loss": 0.0051,
+      "step": 1740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448773952956164e-05,
+      "loss": 0.0018,
+      "step": 1741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446739916930473e-05,
+      "loss": 0.103,
+      "step": 1742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844470466050212e-05,
+      "loss": 0.04,
+      "step": 1743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8442668183965166e-05,
+      "loss": 0.136,
+      "step": 1744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844063048761384e-05,
+      "loss": 0.0031,
+      "step": 1745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438591571742556e-05,
+      "loss": 0.1246,
+      "step": 1746
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436551436645896e-05,
+      "loss": 0.0543,
+      "step": 1747
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434510082618633e-05,
+      "loss": 0.0686,
+      "step": 1748
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432467509955694e-05,
+      "loss": 0.0576,
+      "step": 1749
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430423718952203e-05,
+      "loss": 0.0742,
+      "step": 1750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428378709903447e-05,
+      "loss": 0.0024,
+      "step": 1751
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426332483104893e-05,
+      "loss": 0.0068,
+      "step": 1752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842428503885219e-05,
+      "loss": 0.0752,
+      "step": 1753
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8422236377441147e-05,
+      "loss": 0.0023,
+      "step": 1754
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420186499167766e-05,
+      "loss": 0.076,
+      "step": 1755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418135404328218e-05,
+      "loss": 0.0078,
+      "step": 1756
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416083093218847e-05,
+      "loss": 0.0327,
+      "step": 1757
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414029566136174e-05,
+      "loss": 0.0053,
+      "step": 1758
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84119748233769e-05,
+      "loss": 0.1415,
+      "step": 1759
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84099188652379e-05,
+      "loss": 0.0039,
+      "step": 1760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407861692016215e-05,
+      "loss": 0.1964,
+      "step": 1761
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405803304009077e-05,
+      "loss": 0.0199,
+      "step": 1762
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403743701513882e-05,
+      "loss": 0.0262,
+      "step": 1763
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401682884828212e-05,
+      "loss": 0.0235,
+      "step": 1764
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399620854249807e-05,
+      "loss": 0.2576,
+      "step": 1765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839755761007661e-05,
+      "loss": 0.0572,
+      "step": 1766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839549315260671e-05,
+      "loss": 0.0739,
+      "step": 1767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393427482138383e-05,
+      "loss": 0.0025,
+      "step": 1768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391360598970088e-05,
+      "loss": 0.0879,
+      "step": 1769
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838929250340045e-05,
+      "loss": 0.0152,
+      "step": 1770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387223195728276e-05,
+      "loss": 0.0295,
+      "step": 1771
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385152676252538e-05,
+      "loss": 0.0264,
+      "step": 1772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838308094527239e-05,
+      "loss": 0.0911,
+      "step": 1773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838100800308716e-05,
+      "loss": 0.0026,
+      "step": 1774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8378933849996358e-05,
+      "loss": 0.0095,
+      "step": 1775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837685848629965e-05,
+      "loss": 0.0801,
+      "step": 1776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374781912296894e-05,
+      "loss": 0.0017,
+      "step": 1777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372704128288123e-05,
+      "loss": 0.0012,
+      "step": 1778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370625134573537e-05,
+      "loss": 0.0383,
+      "step": 1779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368544931453508e-05,
+      "loss": 0.0554,
+      "step": 1780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836646351922859e-05,
+      "loss": 0.0801,
+      "step": 1781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364380898199516e-05,
+      "loss": 0.0049,
+      "step": 1782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836229706866718e-05,
+      "loss": 0.0233,
+      "step": 1783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8360212030932664e-05,
+      "loss": 0.0411,
+      "step": 1784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358125785297213e-05,
+      "loss": 0.0034,
+      "step": 1785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356038332062258e-05,
+      "loss": 0.0088,
+      "step": 1786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353949671529392e-05,
+      "loss": 0.0437,
+      "step": 1787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351859804000395e-05,
+      "loss": 0.0039,
+      "step": 1788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834976872977721e-05,
+      "loss": 0.0876,
+      "step": 1789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834767644916197e-05,
+      "loss": 0.0044,
+      "step": 1790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8345582962456955e-05,
+      "loss": 0.1644,
+      "step": 1791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343488269964652e-05,
+      "loss": 0.0012,
+      "step": 1792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83413923719877e-05,
+      "loss": 0.1632,
+      "step": 1793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833929526882892e-05,
+      "loss": 0.0192,
+      "step": 1794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337196960791305e-05,
+      "loss": 0.0502,
+      "step": 1795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833509744817802e-05,
+      "loss": 0.0667,
+      "step": 1796
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332996731292415e-05,
+      "loss": 0.1149,
+      "step": 1797
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8330894810437998e-05,
+      "loss": 0.0046,
+      "step": 1798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328791685918463e-05,
+      "loss": 0.07,
+      "step": 1799
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8326687358037672e-05,
+      "loss": 0.0683,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324581827099665e-05,
+      "loss": 0.0534,
+      "step": 1801
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322475093408653e-05,
+      "loss": 0.0107,
+      "step": 1802
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832036715726902e-05,
+      "loss": 0.1202,
+      "step": 1803
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318258018985324e-05,
+      "loss": 0.0539,
+      "step": 1804
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83161476788623e-05,
+      "loss": 0.1199,
+      "step": 1805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314036137204853e-05,
+      "loss": 0.0647,
+      "step": 1806
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831192339431806e-05,
+      "loss": 0.0152,
+      "step": 1807
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309809450507185e-05,
+      "loss": 0.0241,
+      "step": 1808
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307694306077644e-05,
+      "loss": 0.0313,
+      "step": 1809
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305577961335043e-05,
+      "loss": 0.0102,
+      "step": 1810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8303460416585155e-05,
+      "loss": 0.016,
+      "step": 1811
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301341672133924e-05,
+      "loss": 0.0519,
+      "step": 1812
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299221728287476e-05,
+      "loss": 0.0546,
+      "step": 1813
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82971005853521e-05,
+      "loss": 0.1204,
+      "step": 1814
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829497824363427e-05,
+      "loss": 0.0034,
+      "step": 1815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829285470344062e-05,
+      "loss": 0.0035,
+      "step": 1816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290729965077963e-05,
+      "loss": 0.0804,
+      "step": 1817
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8288604028853287e-05,
+      "loss": 0.0083,
+      "step": 1818
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286476895073755e-05,
+      "loss": 0.0521,
+      "step": 1819
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284348564046695e-05,
+      "loss": 0.0365,
+      "step": 1820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828221903607962e-05,
+      "loss": 0.0028,
+      "step": 1821
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280088311480203e-05,
+      "loss": 0.0103,
+      "step": 1822
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277956390556294e-05,
+      "loss": 0.0087,
+      "step": 1823
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275823273615923e-05,
+      "loss": 0.0049,
+      "step": 1824
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827368896096728e-05,
+      "loss": 0.0013,
+      "step": 1825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827155345291874e-05,
+      "loss": 0.0019,
+      "step": 1826
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269416749778846e-05,
+      "loss": 0.0562,
+      "step": 1827
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267278851856312e-05,
+      "loss": 0.0162,
+      "step": 1828
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265139759460022e-05,
+      "loss": 0.038,
+      "step": 1829
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262999472899045e-05,
+      "loss": 0.0601,
+      "step": 1830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826085799248261e-05,
+      "loss": 0.0064,
+      "step": 1831
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.825871531852012e-05,
+      "loss": 0.0078,
+      "step": 1832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256571451321155e-05,
+      "loss": 0.0107,
+      "step": 1833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254426391195465e-05,
+      "loss": 0.0046,
+      "step": 1834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8252280138452977e-05,
+      "loss": 0.0799,
+      "step": 1835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825013269340378e-05,
+      "loss": 0.0816,
+      "step": 1836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247984056358142e-05,
+      "loss": 0.0662,
+      "step": 1837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824583422762651e-05,
+      "loss": 0.0022,
+      "step": 1838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243683207519484e-05,
+      "loss": 0.0277,
+      "step": 1839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241530996347854e-05,
+      "loss": 0.1052,
+      "step": 1840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823937759442258e-05,
+      "loss": 0.0203,
+      "step": 1841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237223002054785e-05,
+      "loss": 0.0074,
+      "step": 1842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823506721955577e-05,
+      "loss": 0.083,
+      "step": 1843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232910247237008e-05,
+      "loss": 0.0467,
+      "step": 1844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230752085410137e-05,
+      "loss": 0.0681,
+      "step": 1845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228592734386983e-05,
+      "loss": 0.0015,
+      "step": 1846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226432194479527e-05,
+      "loss": 0.0125,
+      "step": 1847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822427046599993e-05,
+      "loss": 0.0026,
+      "step": 1848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822210754926052e-05,
+      "loss": 0.0035,
+      "step": 1849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8219943444573803e-05,
+      "loss": 0.0729,
+      "step": 1850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821777815225245e-05,
+      "loss": 0.016,
+      "step": 1851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215611672609316e-05,
+      "loss": 0.0028,
+      "step": 1852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821344400595741e-05,
+      "loss": 0.0364,
+      "step": 1853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821127515260992e-05,
+      "loss": 0.0743,
+      "step": 1854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209105112880214e-05,
+      "loss": 0.0316,
+      "step": 1855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206933887081817e-05,
+      "loss": 0.0112,
+      "step": 1856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204761475528436e-05,
+      "loss": 0.0475,
+      "step": 1857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202587878533942e-05,
+      "loss": 0.041,
+      "step": 1858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200413096412385e-05,
+      "loss": 0.0003,
+      "step": 1859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198237129477977e-05,
+      "loss": 0.1126,
+      "step": 1860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8196059978045117e-05,
+      "loss": 0.0097,
+      "step": 1861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819388164242835e-05,
+      "loss": 0.0347,
+      "step": 1862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8191702122942417e-05,
+      "loss": 0.0171,
+      "step": 1863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189521419902215e-05,
+      "loss": 0.0018,
+      "step": 1864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187339533622816e-05,
+      "loss": 0.023,
+      "step": 1865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185156464419467e-05,
+      "loss": 0.0547,
+      "step": 1866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818297221260758e-05,
+      "loss": 0.0732,
+      "step": 1867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818078677850274e-05,
+      "loss": 0.0687,
+      "step": 1868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178600162420705e-05,
+      "loss": 0.1571,
+      "step": 1869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176412364677398e-05,
+      "loss": 0.0362,
+      "step": 1870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817422338558892e-05,
+      "loss": 0.0063,
+      "step": 1871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817203322547154e-05,
+      "loss": 0.0094,
+      "step": 1872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169841884641694e-05,
+      "loss": 0.092,
+      "step": 1873
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167649363415992e-05,
+      "loss": 0.003,
+      "step": 1874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816545566211122e-05,
+      "loss": 0.0023,
+      "step": 1875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163260781044317e-05,
+      "loss": 0.0049,
+      "step": 1876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161064720532415e-05,
+      "loss": 0.0187,
+      "step": 1877
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158867480892798e-05,
+      "loss": 0.0018,
+      "step": 1878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156669062442934e-05,
+      "loss": 0.0008,
+      "step": 1879
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154469465500447e-05,
+      "loss": 0.0033,
+      "step": 1880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152268690383146e-05,
+      "loss": 0.0104,
+      "step": 1881
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150066737409006e-05,
+      "loss": 0.0046,
+      "step": 1882
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8147863606896164e-05,
+      "loss": 0.0361,
+      "step": 1883
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814565929916293e-05,
+      "loss": 0.0649,
+      "step": 1884
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143453814527795e-05,
+      "loss": 0.0014,
+      "step": 1885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814124715330941e-05,
+      "loss": 0.0017,
+      "step": 1886
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139039315826596e-05,
+      "loss": 0.0213,
+      "step": 1887
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136830302398342e-05,
+      "loss": 0.0277,
+      "step": 1888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813462011334382e-05,
+      "loss": 0.0039,
+      "step": 1889
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132408748982356e-05,
+      "loss": 0.0012,
+      "step": 1890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8130196209633455e-05,
+      "loss": 0.0011,
+      "step": 1891
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812798249561679e-05,
+      "loss": 0.0018,
+      "step": 1892
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125767607252202e-05,
+      "loss": 0.001,
+      "step": 1893
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81235515448597e-05,
+      "loss": 0.0408,
+      "step": 1894
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121334308759473e-05,
+      "loss": 0.0015,
+      "step": 1895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8119115899271863e-05,
+      "loss": 0.0037,
+      "step": 1896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116896316717394e-05,
+      "loss": 0.0964,
+      "step": 1897
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811467556141676e-05,
+      "loss": 0.1661,
+      "step": 1898
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112453633690818e-05,
+      "loss": 0.0019,
+      "step": 1899
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8110230533860588e-05,
+      "loss": 0.0909,
+      "step": 1900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108006262247282e-05,
+      "loss": 0.0244,
+      "step": 1901
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810578081917226e-05,
+      "loss": 0.1302,
+      "step": 1902
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810355420495706e-05,
+      "loss": 0.1315,
+      "step": 1903
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101326419923384e-05,
+      "loss": 0.0455,
+      "step": 1904
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099097464393114e-05,
+      "loss": 0.0668,
+      "step": 1905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809686733868829e-05,
+      "loss": 0.0031,
+      "step": 1906
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094636043131127e-05,
+      "loss": 0.016,
+      "step": 1907
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8092403578044006e-05,
+      "loss": 0.0107,
+      "step": 1908
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.0166,
+      "step": 1909
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808793514057026e-05,
+      "loss": 0.1195,
+      "step": 1910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808569916882925e-05,
+      "loss": 0.0496,
+      "step": 1911
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083462028849498e-05,
+      "loss": 0.1171,
+      "step": 1912
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8081223720954236e-05,
+      "loss": 0.0136,
+      "step": 1913
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078984245466852e-05,
+      "loss": 0.0548,
+      "step": 1914
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807674360271092e-05,
+      "loss": 0.0076,
+      "step": 1915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074501793010166e-05,
+      "loss": 0.1349,
+      "step": 1916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807225881668849e-05,
+      "loss": 0.0534,
+      "step": 1917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8070014674069968e-05,
+      "loss": 0.0125,
+      "step": 1918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067769365478832e-05,
+      "loss": 0.0228,
+      "step": 1919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8065522891239497e-05,
+      "loss": 0.0052,
+      "step": 1920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806327525167653e-05,
+      "loss": 0.118,
+      "step": 1921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806102644711468e-05,
+      "loss": 0.0385,
+      "step": 1922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058776477878853e-05,
+      "loss": 0.0083,
+      "step": 1923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056525344294137e-05,
+      "loss": 0.011,
+      "step": 1924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054273046685776e-05,
+      "loss": 0.045,
+      "step": 1925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805201958537919e-05,
+      "loss": 0.0656,
+      "step": 1926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049764960699957e-05,
+      "loss": 0.0507,
+      "step": 1927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047509172973832e-05,
+      "loss": 0.0065,
+      "step": 1928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045252222526746e-05,
+      "loss": 0.0015,
+      "step": 1929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8042994109684777e-05,
+      "loss": 0.066,
+      "step": 1930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040734834774186e-05,
+      "loss": 0.0305,
+      "step": 1931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803847439812139e-05,
+      "loss": 0.004,
+      "step": 1932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036212800052998e-05,
+      "loss": 0.1459,
+      "step": 1933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033950040895754e-05,
+      "loss": 0.005,
+      "step": 1934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031686120976594e-05,
+      "loss": 0.0542,
+      "step": 1935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029421040622614e-05,
+      "loss": 0.0039,
+      "step": 1936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027154800161075e-05,
+      "loss": 0.041,
+      "step": 1937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802488739991941e-05,
+      "loss": 0.1212,
+      "step": 1938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802261884022522e-05,
+      "loss": 0.0598,
+      "step": 1939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8020349121406267e-05,
+      "loss": 0.004,
+      "step": 1940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018078243790484e-05,
+      "loss": 0.0932,
+      "step": 1941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015806207705978e-05,
+      "loss": 0.2113,
+      "step": 1942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801353301348101e-05,
+      "loss": 0.0943,
+      "step": 1943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801125866144402e-05,
+      "loss": 0.0839,
+      "step": 1944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008983151923613e-05,
+      "loss": 0.1111,
+      "step": 1945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8006706485248558e-05,
+      "loss": 0.0414,
+      "step": 1946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800442866174779e-05,
+      "loss": 0.0835,
+      "step": 1947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8002149681750415e-05,
+      "loss": 0.017,
+      "step": 1948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999869545585707e-05,
+      "loss": 0.0334,
+      "step": 1949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997588253583103e-05,
+      "loss": 0.0066,
+      "step": 1950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799530580607221e-05,
+      "loss": 0.0105,
+      "step": 1951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79930222033828e-05,
+      "loss": 0.0034,
+      "step": 1952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799073744584481e-05,
+      "loss": 0.0081,
+      "step": 1953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798845153378835e-05,
+      "loss": 0.0404,
+      "step": 1954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7986164467543693e-05,
+      "loss": 0.0599,
+      "step": 1955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7983876247441282e-05,
+      "loss": 0.0101,
+      "step": 1956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981586873811717e-05,
+      "loss": 0.0084,
+      "step": 1957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7979296346985774e-05,
+      "loss": 0.0033,
+      "step": 1958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977004667294394e-05,
+      "loss": 0.0225,
+      "step": 1959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974711835068684e-05,
+      "loss": 0.1257,
+      "step": 1960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972417850639916e-05,
+      "loss": 0.0026,
+      "step": 1961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7970122714339533e-05,
+      "loss": 0.0024,
+      "step": 1962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967826426499136e-05,
+      "loss": 0.004,
+      "step": 1963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79655289874505e-05,
+      "loss": 0.1031,
+      "step": 1964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796323039752557e-05,
+      "loss": 0.1583,
+      "step": 1965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796093065705644e-05,
+      "loss": 0.0221,
+      "step": 1966
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958629766375387e-05,
+      "loss": 0.0556,
+      "step": 1967
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7956327725814845e-05,
+      "loss": 0.0625,
+      "step": 1968
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954024535707428e-05,
+      "loss": 0.0139,
+      "step": 1969
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795172019638589e-05,
+      "loss": 0.0025,
+      "step": 1970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794941470818318e-05,
+      "loss": 0.1108,
+      "step": 1971
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947108071432394e-05,
+      "loss": 0.0526,
+      "step": 1972
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944800286466798e-05,
+      "loss": 0.0118,
+      "step": 1973
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7942491353619828e-05,
+      "loss": 0.0021,
+      "step": 1974
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940181273225086e-05,
+      "loss": 0.0025,
+      "step": 1975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937870045616334e-05,
+      "loss": 0.0059,
+      "step": 1976
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7935557671127505e-05,
+      "loss": 0.1292,
+      "step": 1977
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793324415009269e-05,
+      "loss": 0.0029,
+      "step": 1978
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793092948284616e-05,
+      "loss": 0.0403,
+      "step": 1979
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792861366972234e-05,
+      "loss": 0.0612,
+      "step": 1980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926296711055818e-05,
+      "loss": 0.0297,
+      "step": 1981
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923978607181358e-05,
+      "loss": 0.0083,
+      "step": 1982
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921659358433883e-05,
+      "loss": 0.0076,
+      "step": 1983
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791933896514849e-05,
+      "loss": 0.1244,
+      "step": 1984
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917017427660424e-05,
+      "loss": 0.0204,
+      "step": 1985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914694746305107e-05,
+      "loss": 0.0834,
+      "step": 1986
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912370921418133e-05,
+      "loss": 0.0285,
+      "step": 1987
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910045953335246e-05,
+      "loss": 0.0151,
+      "step": 1988
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790771984239236e-05,
+      "loss": 0.0195,
+      "step": 1989
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905392588925563e-05,
+      "loss": 0.013,
+      "step": 1990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79030641932711e-05,
+      "loss": 0.0021,
+      "step": 1991
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900734655765375e-05,
+      "loss": 0.0653,
+      "step": 1992
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7898403976744977e-05,
+      "loss": 0.0413,
+      "step": 1993
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7896072156546637e-05,
+      "loss": 0.0025,
+      "step": 1994
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893739195507267e-05,
+      "loss": 0.0044,
+      "step": 1995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789140509396394e-05,
+      "loss": 0.0036,
+      "step": 1996
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7889069852253885e-05,
+      "loss": 0.2141,
+      "step": 1997
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78867334707145e-05,
+      "loss": 0.1858,
+      "step": 1998
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7884395949683363e-05,
+      "loss": 0.1317,
+      "step": 1999
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7882057289498197e-05,
+      "loss": 0.0024,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879717490496896e-05,
+      "loss": 0.0955,
+      "step": 2001
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787737655301752e-05,
+      "loss": 0.0032,
+      "step": 2002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787503447739829e-05,
+      "loss": 0.0438,
+      "step": 2003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78726912639776e-05,
+      "loss": 0.0159,
+      "step": 2004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787034691309399e-05,
+      "loss": 0.0629,
+      "step": 2005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868001425086193e-05,
+      "loss": 0.0017,
+      "step": 2006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786565480029308e-05,
+      "loss": 0.0006,
+      "step": 2007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7863307039053695e-05,
+      "loss": 0.0034,
+      "step": 2008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860958141707256e-05,
+      "loss": 0.0257,
+      "step": 2009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785860810859313e-05,
+      "loss": 0.0226,
+      "step": 2010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785625694005086e-05,
+      "loss": 0.1361,
+      "step": 2011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785390463642014e-05,
+      "loss": 0.0039,
+      "step": 2012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785155119804084e-05,
+      "loss": 0.0097,
+      "step": 2013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784919662525299e-05,
+      "loss": 0.0113,
+      "step": 2014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784684091839678e-05,
+      "loss": 0.0428,
+      "step": 2015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844484077812575e-05,
+      "loss": 0.002,
+      "step": 2016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784212610384089e-05,
+      "loss": 0.0014,
+      "step": 2017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783976699682241e-05,
+      "loss": 0.1946,
+      "step": 2018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837406757097994e-05,
+      "loss": 0.0644,
+      "step": 2019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783504538500864e-05,
+      "loss": 0.0275,
+      "step": 2020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832682880895526e-05,
+      "loss": 0.0257,
+      "step": 2021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78303192451e-05,
+      "loss": 0.007,
+      "step": 2022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782795447796356e-05,
+      "loss": 0.0052,
+      "step": 2023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782558857982787e-05,
+      "loss": 0.0555,
+      "step": 2024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823221551034766e-05,
+      "loss": 0.0704,
+      "step": 2025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782085339192624e-05,
+      "loss": 0.0623,
+      "step": 2026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818484102844445e-05,
+      "loss": 0.0183,
+      "step": 2027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816113684131702e-05,
+      "loss": 0.1013,
+      "step": 2028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781374213613049e-05,
+      "loss": 0.003,
+      "step": 2029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811369459183468e-05,
+      "loss": 0.0732,
+      "step": 2030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7808995653633433e-05,
+      "loss": 0.0283,
+      "step": 2031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7806620719823362e-05,
+      "loss": 0.0966,
+      "step": 2032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780424465809639e-05,
+      "loss": 0.0018,
+      "step": 2033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780186746879582e-05,
+      "loss": 0.0078,
+      "step": 2034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799489152265104e-05,
+      "loss": 0.0154,
+      "step": 2035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779710970884787e-05,
+      "loss": 0.0456,
+      "step": 2036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7794729138887908e-05,
+      "loss": 0.0957,
+      "step": 2037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792347442729164e-05,
+      "loss": 0.005,
+      "step": 2038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778996462071575e-05,
+      "loss": 0.0699,
+      "step": 2039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787580673191948e-05,
+      "loss": 0.0134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778519560050219e-05,
+      "loss": 0.0134,
+      "step": 2041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782809402991076e-05,
+      "loss": 0.0169,
+      "step": 2042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780422081003367e-05,
+      "loss": 0.0951,
+      "step": 2043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778033634883996e-05,
+      "loss": 0.0849,
+      "step": 2044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7775644064978042e-05,
+      "loss": 0.042,
+      "step": 2045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773253371630762e-05,
+      "loss": 0.0917,
+      "step": 2046
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770861555187565e-05,
+      "loss": 0.0071,
+      "step": 2047
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7768468615994025e-05,
+      "loss": 0.055,
+      "step": 2048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776607455439588e-05,
+      "loss": 0.0142,
+      "step": 2049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776367937073903e-05,
+      "loss": 0.0149,
+      "step": 2050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7761283065369533e-05,
+      "loss": 0.0851,
+      "step": 2051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758885638633615e-05,
+      "loss": 0.0045,
+      "step": 2052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775648709087766e-05,
+      "loss": 0.0006,
+      "step": 2053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754087422448217e-05,
+      "loss": 0.1778,
+      "step": 2054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751686633691992e-05,
+      "loss": 0.0332,
+      "step": 2055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774928472495586e-05,
+      "loss": 0.0013,
+      "step": 2056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746881696586847e-05,
+      "loss": 0.0122,
+      "step": 2057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774447754893216e-05,
+      "loss": 0.0755,
+      "step": 2058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742072282339142e-05,
+      "loss": 0.0041,
+      "step": 2059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773966589715532e-05,
+      "loss": 0.0339,
+      "step": 2060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7737258393728363e-05,
+      "loss": 0.017,
+      "step": 2061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734849772406125e-05,
+      "loss": 0.101,
+      "step": 2062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732440033536602e-05,
+      "loss": 0.1121,
+      "step": 2063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7730029177467955e-05,
+      "loss": 0.0562,
+      "step": 2064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772761720454852e-05,
+      "loss": 0.0333,
+      "step": 2065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7725204115126772e-05,
+      "loss": 0.0849,
+      "step": 2066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722789909551367e-05,
+      "loss": 0.018,
+      "step": 2067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772037458817111e-05,
+      "loss": 0.0023,
+      "step": 2068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7717958151334976e-05,
+      "loss": 0.068,
+      "step": 2069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715540599392094e-05,
+      "loss": 0.1371,
+      "step": 2070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713121932691755e-05,
+      "loss": 0.2429,
+      "step": 2071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771070215158342e-05,
+      "loss": 0.003,
+      "step": 2072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7708281256416696e-05,
+      "loss": 0.0351,
+      "step": 2073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770585924754136e-05,
+      "loss": 0.014,
+      "step": 2074
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703436125307356e-05,
+      "loss": 0.0584,
+      "step": 2075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7701011890064778e-05,
+      "loss": 0.0824,
+      "step": 2076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769858654216388e-05,
+      "loss": 0.0327,
+      "step": 2077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696160081955085e-05,
+      "loss": 0.0545,
+      "step": 2078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693732509788975e-05,
+      "loss": 0.0073,
+      "step": 2079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769130382601629e-05,
+      "loss": 0.0133,
+      "step": 2080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768887403098793e-05,
+      "loss": 0.0139,
+      "step": 2081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7686443125054957e-05,
+      "loss": 0.0532,
+      "step": 2082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7684011108568593e-05,
+      "loss": 0.0819,
+      "step": 2083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768157798188022e-05,
+      "loss": 0.0486,
+      "step": 2084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679143745341383e-05,
+      "loss": 0.0415,
+      "step": 2085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676708399303784e-05,
+      "loss": 0.0125,
+      "step": 2086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7674271944119293e-05,
+      "loss": 0.0377,
+      "step": 2087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671834380139925e-05,
+      "loss": 0.0067,
+      "step": 2088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669395707717868e-05,
+      "loss": 0.0377,
+      "step": 2089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766695592720547e-05,
+      "loss": 0.009,
+      "step": 2090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766451503895523e-05,
+      "loss": 0.1011,
+      "step": 2091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662073043319817e-05,
+      "loss": 0.011,
+      "step": 2092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7659629940652052e-05,
+      "loss": 0.0702,
+      "step": 2093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657185731304923e-05,
+      "loss": 0.0023,
+      "step": 2094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7654740415631574e-05,
+      "loss": 0.0909,
+      "step": 2095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7652293993985304e-05,
+      "loss": 0.001,
+      "step": 2096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7649846466719583e-05,
+      "loss": 0.042,
+      "step": 2097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647397834188036e-05,
+      "loss": 0.0569,
+      "step": 2098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764494809674444e-05,
+      "loss": 0.104,
+      "step": 2099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642497254742744e-05,
+      "loss": 0.0073,
+      "step": 2100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764004530853705e-05,
+      "loss": 0.0043,
+      "step": 2101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7637592258481618e-05,
+      "loss": 0.0534,
+      "step": 2102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635138104930872e-05,
+      "loss": 0.0262,
+      "step": 2103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632682848239392e-05,
+      "loss": 0.0176,
+      "step": 2104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7630226488761918e-05,
+      "loss": 0.0547,
+      "step": 2105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762776902685335e-05,
+      "loss": 0.046,
+      "step": 2106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762531046286875e-05,
+      "loss": 0.0059,
+      "step": 2107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762285079716333e-05,
+      "loss": 0.0108,
+      "step": 2108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762039003009248e-05,
+      "loss": 0.003,
+      "step": 2109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761792816201172e-05,
+      "loss": 0.0309,
+      "step": 2110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761546519327676e-05,
+      "loss": 0.0726,
+      "step": 2111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613001124243448e-05,
+      "loss": 0.163,
+      "step": 2112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76105359552678e-05,
+      "loss": 0.0114,
+      "step": 2113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608069686705985e-05,
+      "loss": 0.0046,
+      "step": 2114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7605602318914344e-05,
+      "loss": 0.0384,
+      "step": 2115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603133852249357e-05,
+      "loss": 0.0134,
+      "step": 2116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760066428706768e-05,
+      "loss": 0.0242,
+      "step": 2117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598193623726115e-05,
+      "loss": 0.0551,
+      "step": 2118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595721862581635e-05,
+      "loss": 0.001,
+      "step": 2119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593249003991362e-05,
+      "loss": 0.0089,
+      "step": 2120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590775048312577e-05,
+      "loss": 0.0729,
+      "step": 2121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758829999590273e-05,
+      "loss": 0.0026,
+      "step": 2122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7585823847119416e-05,
+      "loss": 0.0792,
+      "step": 2123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583346602320396e-05,
+      "loss": 0.0982,
+      "step": 2124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580868261863585e-05,
+      "loss": 0.0497,
+      "step": 2125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757838882610706e-05,
+      "loss": 0.1253,
+      "step": 2126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757590829540906e-05,
+      "loss": 0.0832,
+      "step": 2127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7573426670127966e-05,
+      "loss": 0.1042,
+      "step": 2128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757094395062234e-05,
+      "loss": 0.1051,
+      "step": 2129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568460137250888e-05,
+      "loss": 0.0449,
+      "step": 2130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7565975230372473e-05,
+      "loss": 0.0669,
+      "step": 2131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563489230346116e-05,
+      "loss": 0.0703,
+      "step": 2132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756100213753101e-05,
+      "loss": 0.0822,
+      "step": 2133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558513952286488e-05,
+      "loss": 0.044,
+      "step": 2134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556024674972053e-05,
+      "loss": 0.05,
+      "step": 2135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553534305947352e-05,
+      "loss": 0.0917,
+      "step": 2136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755104284557221e-05,
+      "loss": 0.0133,
+      "step": 2137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754855029420659e-05,
+      "loss": 0.1064,
+      "step": 2138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7546056652210627e-05,
+      "loss": 0.016,
+      "step": 2139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543561919944604e-05,
+      "loss": 0.026,
+      "step": 2140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541066097768965e-05,
+      "loss": 0.0406,
+      "step": 2141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7538569186044312e-05,
+      "loss": 0.0863,
+      "step": 2142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536071185131404e-05,
+      "loss": 0.0125,
+      "step": 2143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753357209539116e-05,
+      "loss": 0.0543,
+      "step": 2144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531071917184656e-05,
+      "loss": 0.024,
+      "step": 2145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528570650873115e-05,
+      "loss": 0.0356,
+      "step": 2146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526068296817926e-05,
+      "loss": 0.0342,
+      "step": 2147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752356485538064e-05,
+      "loss": 0.0986,
+      "step": 2148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521060326922955e-05,
+      "loss": 0.0663,
+      "step": 2149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518554711806735e-05,
+      "loss": 0.0309,
+      "step": 2150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516048010393993e-05,
+      "loss": 0.0697,
+      "step": 2151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513540223046903e-05,
+      "loss": 0.081,
+      "step": 2152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7511031350127795e-05,
+      "loss": 0.0358,
+      "step": 2153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7508521391999158e-05,
+      "loss": 0.018,
+      "step": 2154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506010349023635e-05,
+      "loss": 0.011,
+      "step": 2155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503498221564026e-05,
+      "loss": 0.027,
+      "step": 2156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750098500998329e-05,
+      "loss": 0.055,
+      "step": 2157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7498470714644537e-05,
+      "loss": 0.0321,
+      "step": 2158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7495955335911044e-05,
+      "loss": 0.1071,
+      "step": 2159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7493438874146233e-05,
+      "loss": 0.0872,
+      "step": 2160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749092132971369e-05,
+      "loss": 0.0123,
+      "step": 2161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488402702977155e-05,
+      "loss": 0.0853,
+      "step": 2162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485882994300524e-05,
+      "loss": 0.0165,
+      "step": 2163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748336220404785e-05,
+      "loss": 0.0091,
+      "step": 2164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748084033258334e-05,
+      "loss": 0.0645,
+      "step": 2165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7478317380271367e-05,
+      "loss": 0.081,
+      "step": 2166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747579334747644e-05,
+      "loss": 0.0726,
+      "step": 2167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473268234563247e-05,
+      "loss": 0.0027,
+      "step": 2168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7470742041896618e-05,
+      "loss": 0.0084,
+      "step": 2169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7468214769841542e-05,
+      "loss": 0.0206,
+      "step": 2170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746568641876316e-05,
+      "loss": 0.063,
+      "step": 2171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7463156989026785e-05,
+      "loss": 0.0603,
+      "step": 2172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7460626480997865e-05,
+      "loss": 0.0034,
+      "step": 2173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458094895042017e-05,
+      "loss": 0.1088,
+      "step": 2174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455562231525006e-05,
+      "loss": 0.0119,
+      "step": 2175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7453028490812764e-05,
+      "loss": 0.0105,
+      "step": 2176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450493673271364e-05,
+      "loss": 0.0873,
+      "step": 2177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744795777926704e-05,
+      "loss": 0.0801,
+      "step": 2178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7445420809166192e-05,
+      "loss": 0.0453,
+      "step": 2179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744288276333536e-05,
+      "loss": 0.006,
+      "step": 2180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744034364214125e-05,
+      "loss": 0.0032,
+      "step": 2181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743780344595072e-05,
+      "loss": 0.0247,
+      "step": 2182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7435262175130775e-05,
+      "loss": 0.0033,
+      "step": 2183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432719830048594e-05,
+      "loss": 0.016,
+      "step": 2184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430176411071492e-05,
+      "loss": 0.0631,
+      "step": 2185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742763191856695e-05,
+      "loss": 0.1224,
+      "step": 2186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7425086352902605e-05,
+      "loss": 0.0343,
+      "step": 2187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7422539714446244e-05,
+      "loss": 0.0051,
+      "step": 2188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419992003565806e-05,
+      "loss": 0.0179,
+      "step": 2189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417443220629394e-05,
+      "loss": 0.0423,
+      "step": 2190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414893366005263e-05,
+      "loss": 0.0041,
+      "step": 2191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7412342440061816e-05,
+      "loss": 0.0045,
+      "step": 2192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740979044316762e-05,
+      "loss": 0.0016,
+      "step": 2193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7407237375691394e-05,
+      "loss": 0.0095,
+      "step": 2194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404683238002004e-05,
+      "loss": 0.0024,
+      "step": 2195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402128030468487e-05,
+      "loss": 0.03,
+      "step": 2196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399571753460017e-05,
+      "loss": 0.0008,
+      "step": 2197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739701440734593e-05,
+      "loss": 0.1815,
+      "step": 2198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394455992495722e-05,
+      "loss": 0.029,
+      "step": 2199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391896509279038e-05,
+      "loss": 0.251,
+      "step": 2200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738933595806567e-05,
+      "loss": 0.0976,
+      "step": 2201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386774339225582e-05,
+      "loss": 0.2042,
+      "step": 2202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384211653128873e-05,
+      "loss": 0.0765,
+      "step": 2203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381647900145808e-05,
+      "loss": 0.0688,
+      "step": 2204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737908308064681e-05,
+      "loss": 0.0362,
+      "step": 2205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737651719500244e-05,
+      "loss": 0.0258,
+      "step": 2206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7373950243583426e-05,
+      "loss": 0.0652,
+      "step": 2207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371382226760646e-05,
+      "loss": 0.0086,
+      "step": 2208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7368813144905136e-05,
+      "loss": 0.0086,
+      "step": 2209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736624299838808e-05,
+      "loss": 0.0029,
+      "step": 2210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7363671787580818e-05,
+      "loss": 0.0025,
+      "step": 2211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7361099512854844e-05,
+      "loss": 0.0599,
+      "step": 2212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358526174581806e-05,
+      "loss": 0.1191,
+      "step": 2213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355951773133504e-05,
+      "loss": 0.0039,
+      "step": 2214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7353376308881894e-05,
+      "loss": 0.052,
+      "step": 2215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735079978219908e-05,
+      "loss": 0.0103,
+      "step": 2216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348222193457332e-05,
+      "loss": 0.0333,
+      "step": 2217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734564354302906e-05,
+      "loss": 0.0253,
+      "step": 2218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343063831286837e-05,
+      "loss": 0.0647,
+      "step": 2219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340483058603382e-05,
+      "loss": 0.0047,
+      "step": 2220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733790122535157e-05,
+      "loss": 0.0083,
+      "step": 2221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7335318331904428e-05,
+      "loss": 0.0302,
+      "step": 2222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7332734378635146e-05,
+      "loss": 0.1135,
+      "step": 2223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733014936591705e-05,
+      "loss": 0.087,
+      "step": 2224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732756329412363e-05,
+      "loss": 0.0192,
+      "step": 2225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7324976163628533e-05,
+      "loss": 0.052,
+      "step": 2226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322387974805546e-05,
+      "loss": 0.0505,
+      "step": 2227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319798728028617e-05,
+      "loss": 0.028,
+      "step": 2228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731720842367185e-05,
+      "loss": 0.0646,
+      "step": 2229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314617062109492e-05,
+      "loss": 0.0216,
+      "step": 2230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7312024643715958e-05,
+      "loss": 0.0637,
+      "step": 2231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730943116886579e-05,
+      "loss": 0.0372,
+      "step": 2232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306836637933716e-05,
+      "loss": 0.04,
+      "step": 2233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7304241051294586e-05,
+      "loss": 0.0457,
+      "step": 2234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7301644409323424e-05,
+      "loss": 0.0377,
+      "step": 2235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7299046712395394e-05,
+      "loss": 0.0126,
+      "step": 2236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7296447960885818e-05,
+      "loss": 0.0034,
+      "step": 2237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729384815517017e-05,
+      "loss": 0.0702,
+      "step": 2238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291247295624073e-05,
+      "loss": 0.0139,
+      "step": 2239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7288645382623307e-05,
+      "loss": 0.0164,
+      "step": 2240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72860424165438e-05,
+      "loss": 0.0237,
+      "step": 2241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283438397761636e-05,
+      "loss": 0.0047,
+      "step": 2242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280833326653044e-05,
+      "loss": 0.0486,
+      "step": 2243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727822720359442e-05,
+      "loss": 0.0035,
+      "step": 2244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275620028962295e-05,
+      "loss": 0.0346,
+      "step": 2245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273011803133356e-05,
+      "loss": 0.0165,
+      "step": 2246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270402526484454e-05,
+      "loss": 0.0016,
+      "step": 2247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7267792199392578e-05,
+      "loss": 0.0648,
+      "step": 2248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726518082223487e-05,
+      "loss": 0.0106,
+      "step": 2249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726256839538863e-05,
+      "loss": 0.0496,
+      "step": 2250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725995491923131e-05,
+      "loss": 0.0628,
+      "step": 2251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257340394140508e-05,
+      "loss": 0.0647,
+      "step": 2252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7254724820493977e-05,
+      "loss": 0.0022,
+      "step": 2253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725210819866962e-05,
+      "loss": 0.0473,
+      "step": 2254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7249490529045487e-05,
+      "loss": 0.001,
+      "step": 2255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7246871811999793e-05,
+      "loss": 0.0036,
+      "step": 2256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7244252047910893e-05,
+      "loss": 0.0011,
+      "step": 2257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724163123715729e-05,
+      "loss": 0.0038,
+      "step": 2258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239009380117653e-05,
+      "loss": 0.0045,
+      "step": 2259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723638647717079e-05,
+      "loss": 0.0464,
+      "step": 2260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723376252869566e-05,
+      "loss": 0.0023,
+      "step": 2261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723113753507138e-05,
+      "loss": 0.0224,
+      "step": 2262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7228511496677213e-05,
+      "loss": 0.0282,
+      "step": 2263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722588441389258e-05,
+      "loss": 0.0521,
+      "step": 2264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722325628709704e-05,
+      "loss": 0.0878,
+      "step": 2265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7220627116670314e-05,
+      "loss": 0.0495,
+      "step": 2266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721799690299227e-05,
+      "loss": 0.1217,
+      "step": 2267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7215365646442923e-05,
+      "loss": 0.0572,
+      "step": 2268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212733347402448e-05,
+      "loss": 0.1048,
+      "step": 2269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210100006251164e-05,
+      "loss": 0.0046,
+      "step": 2270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207465623369535e-05,
+      "loss": 0.0612,
+      "step": 2271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7204830199138195e-05,
+      "loss": 0.1817,
+      "step": 2272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72021937339379e-05,
+      "loss": 0.0014,
+      "step": 2273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199556228149585e-05,
+      "loss": 0.0328,
+      "step": 2274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7196917682154316e-05,
+      "loss": 0.0741,
+      "step": 2275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719427809633332e-05,
+      "loss": 0.0383,
+      "step": 2276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7191637471067962e-05,
+      "loss": 0.0033,
+      "step": 2277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7188995806739773e-05,
+      "loss": 0.0137,
+      "step": 2278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186353103730423e-05,
+      "loss": 0.0957,
+      "step": 2279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183709362421733e-05,
+      "loss": 0.0333,
+      "step": 2280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718106458319568e-05,
+      "loss": 0.006,
+      "step": 2281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717841876643439e-05,
+      "loss": 0.0943,
+      "step": 2282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175771912520127e-05,
+      "loss": 0.222,
+      "step": 2283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173124021835322e-05,
+      "loss": 0.0458,
+      "step": 2284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7170475094762542e-05,
+      "loss": 0.0116,
+      "step": 2285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167825131684516e-05,
+      "loss": 0.0061,
+      "step": 2286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165174132984107e-05,
+      "loss": 0.0586,
+      "step": 2287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716252209904435e-05,
+      "loss": 0.007,
+      "step": 2288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7159869030248404e-05,
+      "loss": 0.0404,
+      "step": 2289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7157214926979594e-05,
+      "loss": 0.0847,
+      "step": 2290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154559789621394e-05,
+      "loss": 0.114,
+      "step": 2291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715190361855742e-05,
+      "loss": 0.0184,
+      "step": 2292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7149246414171445e-05,
+      "loss": 0.0111,
+      "step": 2293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7146588176847377e-05,
+      "loss": 0.0672,
+      "step": 2294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7143928906969297e-05,
+      "loss": 0.0021,
+      "step": 2295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141268604921414e-05,
+      "loss": 0.0271,
+      "step": 2296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71386072710881e-05,
+      "loss": 0.0568,
+      "step": 2297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135944905853863e-05,
+      "loss": 0.0819,
+      "step": 2298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713328150960337e-05,
+      "loss": 0.0018,
+      "step": 2299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7130617082721436e-05,
+      "loss": 0.0064,
+      "step": 2300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712795162559302e-05,
+      "loss": 0.0468,
+      "step": 2301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125285138603236e-05,
+      "loss": 0.0831,
+      "step": 2302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712261762213734e-05,
+      "loss": 0.0859,
+      "step": 2303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7119949076580743e-05,
+      "loss": 0.0535,
+      "step": 2304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117279502319004e-05,
+      "loss": 0.0692,
+      "step": 2305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711460889973782e-05,
+      "loss": 0.0134,
+      "step": 2306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711193726922306e-05,
+      "loss": 0.0168,
+      "step": 2307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710926461116071e-05,
+      "loss": 0.0875,
+      "step": 2308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710659092593693e-05,
+      "loss": 0.0024,
+      "step": 2309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710391621393802e-05,
+      "loss": 0.086,
+      "step": 2310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7101240475550427e-05,
+      "loss": 0.0364,
+      "step": 2311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7098563711160746e-05,
+      "loss": 0.0476,
+      "step": 2312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7095885921155725e-05,
+      "loss": 0.0039,
+      "step": 2313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7093207105922256e-05,
+      "loss": 0.0301,
+      "step": 2314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090527265847375e-05,
+      "loss": 0.0269,
+      "step": 2315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7087846401318273e-05,
+      "loss": 0.0264,
+      "step": 2316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708516451272229e-05,
+      "loss": 0.126,
+      "step": 2317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708248160044691e-05,
+      "loss": 0.1634,
+      "step": 2318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079797664879763e-05,
+      "loss": 0.0763,
+      "step": 2319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077112706408628e-05,
+      "loss": 0.0083,
+      "step": 2320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7074426725421444e-05,
+      "loss": 0.0119,
+      "step": 2321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071739722306273e-05,
+      "loss": 0.0067,
+      "step": 2322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706905169745135e-05,
+      "loss": 0.1226,
+      "step": 2323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7066362651245038e-05,
+      "loss": 0.0083,
+      "step": 2324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063672584075862e-05,
+      "loss": 0.0666,
+      "step": 2325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706098149633249e-05,
+      "loss": 0.0046,
+      "step": 2326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705828938840373e-05,
+      "loss": 0.0475,
+      "step": 2327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7055596260678546e-05,
+      "loss": 0.1208,
+      "step": 2328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7052902113546047e-05,
+      "loss": 0.0249,
+      "step": 2329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705020694739549e-05,
+      "loss": 0.0054,
+      "step": 2330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7047510762616275e-05,
+      "loss": 0.0037,
+      "step": 2331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044813559597953e-05,
+      "loss": 0.0575,
+      "step": 2332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704211533873023e-05,
+      "loss": 0.0045,
+      "step": 2333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039416100402937e-05,
+      "loss": 0.016,
+      "step": 2334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7036715845006075e-05,
+      "loss": 0.0569,
+      "step": 2335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034014572929777e-05,
+      "loss": 0.0822,
+      "step": 2336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031312284564334e-05,
+      "loss": 0.0322,
+      "step": 2337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7028608980300175e-05,
+      "loss": 0.0635,
+      "step": 2338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702590466052788e-05,
+      "loss": 0.1408,
+      "step": 2339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7023199325638175e-05,
+      "loss": 0.0287,
+      "step": 2340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702049297602193e-05,
+      "loss": 0.0907,
+      "step": 2341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7017785612070166e-05,
+      "loss": 0.026,
+      "step": 2342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7015077234174045e-05,
+      "loss": 0.031,
+      "step": 2343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012367842724887e-05,
+      "loss": 0.0809,
+      "step": 2344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009657438114144e-05,
+      "loss": 0.0194,
+      "step": 2345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7006946020733426e-05,
+      "loss": 0.0377,
+      "step": 2346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700423359097448e-05,
+      "loss": 0.0984,
+      "step": 2347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7001520149229205e-05,
+      "loss": 0.2668,
+      "step": 2348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6998805695889643e-05,
+      "loss": 0.0648,
+      "step": 2349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699609023134799e-05,
+      "loss": 0.0461,
+      "step": 2350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6993373755996572e-05,
+      "loss": 0.0647,
+      "step": 2351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990656270227877e-05,
+      "loss": 0.0133,
+      "step": 2352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6987937774434534e-05,
+      "loss": 0.083,
+      "step": 2353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6985218269009316e-05,
+      "loss": 0.0391,
+      "step": 2354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698249775434514e-05,
+      "loss": 0.0194,
+      "step": 2355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979776230835076e-05,
+      "loss": 0.0416,
+      "step": 2356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977053698872335e-05,
+      "loss": 0.0357,
+      "step": 2357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697433015885027e-05,
+      "loss": 0.0622,
+      "step": 2358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697160561116239e-05,
+      "loss": 0.0433,
+      "step": 2359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696888005620233e-05,
+      "loss": 0.0227,
+      "step": 2360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69661534943639e-05,
+      "loss": 0.138,
+      "step": 2361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6963425926041036e-05,
+      "loss": 0.0646,
+      "step": 2362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696069735162782e-05,
+      "loss": 0.0498,
+      "step": 2363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957967771518475e-05,
+      "loss": 0.0241,
+      "step": 2364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955237186107387e-05,
+      "loss": 0.0778,
+      "step": 2365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695250559578907e-05,
+      "loss": 0.0156,
+      "step": 2366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949773000958195e-05,
+      "loss": 0.1389,
+      "step": 2367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6947039402009574e-05,
+      "loss": 0.0339,
+      "step": 2368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944304799338156e-05,
+      "loss": 0.0348,
+      "step": 2369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694156919333905e-05,
+      "loss": 0.0104,
+      "step": 2370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6938832584407494e-05,
+      "loss": 0.0371,
+      "step": 2371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6936094972938885e-05,
+      "loss": 0.1303,
+      "step": 2372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6933356359328756e-05,
+      "loss": 0.0034,
+      "step": 2373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693061674397279e-05,
+      "loss": 0.0317,
+      "step": 2374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6927876127266807e-05,
+      "loss": 0.0983,
+      "step": 2375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6925134509606786e-05,
+      "loss": 0.1027,
+      "step": 2376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6922391891388835e-05,
+      "loss": 0.0372,
+      "step": 2377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919648273009213e-05,
+      "loss": 0.0079,
+      "step": 2378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916903654864328e-05,
+      "loss": 0.0437,
+      "step": 2379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6914158037350724e-05,
+      "loss": 0.0634,
+      "step": 2380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6911411420865097e-05,
+      "loss": 0.0318,
+      "step": 2381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908663805804284e-05,
+      "loss": 0.0899,
+      "step": 2382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690591519256526e-05,
+      "loss": 0.002,
+      "step": 2383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690316558154516e-05,
+      "loss": 0.0055,
+      "step": 2384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900414973141242e-05,
+      "loss": 0.0727,
+      "step": 2385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689766336775093e-05,
+      "loss": 0.0184,
+      "step": 2386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894910765771775e-05,
+      "loss": 0.0334,
+      "step": 2387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892157167601487e-05,
+      "loss": 0.0851,
+      "step": 2388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6889402573637902e-05,
+      "loss": 0.0624,
+      "step": 2389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688664698427901e-05,
+      "loss": 0.011,
+      "step": 2390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6883890399922952e-05,
+      "loss": 0.0107,
+      "step": 2391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881132820968003e-05,
+      "loss": 0.0182,
+      "step": 2392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6878374247812577e-05,
+      "loss": 0.0755,
+      "step": 2393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6875614680855247e-05,
+      "loss": 0.0859,
+      "step": 2394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6872854120494716e-05,
+      "loss": 0.0161,
+      "step": 2395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6870092567129837e-05,
+      "loss": 0.0056,
+      "step": 2396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867330021159604e-05,
+      "loss": 0.0049,
+      "step": 2397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864566482983155e-05,
+      "loss": 0.0282,
+      "step": 2398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6861801952999775e-05,
+      "loss": 0.0355,
+      "step": 2399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859036431608885e-05,
+      "loss": 0.0396,
+      "step": 2400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6856269919210056e-05,
+      "loss": 0.0038,
+      "step": 2401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853502416203e-05,
+      "loss": 0.1286,
+      "step": 2402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685073392298757e-05,
+      "loss": 0.0151,
+      "step": 2403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6847964439963763e-05,
+      "loss": 0.0579,
+      "step": 2404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6845193967531722e-05,
+      "loss": 0.0389,
+      "step": 2405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6842422506091727e-05,
+      "loss": 0.0373,
+      "step": 2406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683965005604421e-05,
+      "loss": 0.012,
+      "step": 2407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836876617789734e-05,
+      "loss": 0.0574,
+      "step": 2408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683410219172902e-05,
+      "loss": 0.0812,
+      "step": 2409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6831326778262912e-05,
+      "loss": 0.0157,
+      "step": 2410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828550377792415e-05,
+      "loss": 0.04,
+      "step": 2411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682577299071867e-05,
+      "loss": 0.0278,
+      "step": 2412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822994617442954e-05,
+      "loss": 0.0589,
+      "step": 2413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820215258366695e-05,
+      "loss": 0.0067,
+      "step": 2414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817434913891462e-05,
+      "loss": 0.0063,
+      "step": 2415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814653584418966e-05,
+      "loss": 0.043,
+      "step": 2416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681187127035105e-05,
+      "loss": 0.0127,
+      "step": 2417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680908797208972e-05,
+      "loss": 0.0183,
+      "step": 2418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680630369003711e-05,
+      "loss": 0.0268,
+      "step": 2419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803518424595494e-05,
+      "loss": 0.0528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800732176167296e-05,
+      "loss": 0.0412,
+      "step": 2421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679794494515508e-05,
+      "loss": 0.0074,
+      "step": 2422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6795156731961554e-05,
+      "loss": 0.0313,
+      "step": 2423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6792367536989553e-05,
+      "loss": 0.0282,
+      "step": 2424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6789577360642078e-05,
+      "loss": 0.1235,
+      "step": 2425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786786203322255e-05,
+      "loss": 0.025,
+      "step": 2426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6783994065433357e-05,
+      "loss": 0.0052,
+      "step": 2427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6781200947378797e-05,
+      "loss": 0.1062,
+      "step": 2428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677840684956213e-05,
+      "loss": 0.0131,
+      "step": 2429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677561177238706e-05,
+      "loss": 0.0429,
+      "step": 2430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6772815716257414e-05,
+      "loss": 0.0039,
+      "step": 2431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770018681577177e-05,
+      "loss": 0.0669,
+      "step": 2432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676722066875047e-05,
+      "loss": 0.0037,
+      "step": 2433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764421678181563e-05,
+      "loss": 0.0018,
+      "step": 2434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676162171027485e-05,
+      "loss": 0.004,
+      "step": 2435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675882076543488e-05,
+      "loss": 0.0474,
+      "step": 2436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756018844066343e-05,
+      "loss": 0.0346,
+      "step": 2437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675321594657406e-05,
+      "loss": 0.0078,
+      "step": 2438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750412073363002e-05,
+      "loss": 0.0012,
+      "step": 2439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674760722483828e-05,
+      "loss": 0.0021,
+      "step": 2440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6744801401405138e-05,
+      "loss": 0.0237,
+      "step": 2441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674199460346898e-05,
+      "loss": 0.0037,
+      "step": 2442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673918683143533e-05,
+      "loss": 0.0708,
+      "step": 2443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673637808570986e-05,
+      "loss": 0.0027,
+      "step": 2444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6733568366698382e-05,
+      "loss": 0.0059,
+      "step": 2445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730757674806858e-05,
+      "loss": 0.0689,
+      "step": 2446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6727946010441374e-05,
+      "loss": 0.1024,
+      "step": 2447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725133374008167e-05,
+      "loss": 0.0104,
+      "step": 2448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722319765913618e-05,
+      "loss": 0.0852,
+      "step": 2449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719505186564236e-05,
+      "loss": 0.0024,
+      "step": 2450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6716689636366687e-05,
+      "loss": 0.0465,
+      "step": 2451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713873115727756e-05,
+      "loss": 0.0039,
+      "step": 2452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711055625054387e-05,
+      "loss": 0.0582,
+      "step": 2453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6708237164753653e-05,
+      "loss": 0.0013,
+      "step": 2454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670541773523277e-05,
+      "loss": 0.0054,
+      "step": 2455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702597336899105e-05,
+      "loss": 0.0034,
+      "step": 2456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699775970160144e-05,
+      "loss": 0.0959,
+      "step": 2457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6696953635423532e-05,
+      "loss": 0.0417,
+      "step": 2458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6694130333097037e-05,
+      "loss": 0.0126,
+      "step": 2459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.0271,
+      "step": 2460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688480827306224e-05,
+      "loss": 0.0694,
+      "step": 2461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685654624658155e-05,
+      "loss": 0.0715,
+      "step": 2462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6682827456052716e-05,
+      "loss": 0.0366,
+      "step": 2463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679999321898375e-05,
+      "loss": 0.0112,
+      "step": 2464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667717022260375e-05,
+      "loss": 0.0212,
+      "step": 2465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66743401585776e-05,
+      "loss": 0.0029,
+      "step": 2466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667150913022881e-05,
+      "loss": 0.0078,
+      "step": 2467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666867713796642e-05,
+      "loss": 0.0063,
+      "step": 2468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66658441821996e-05,
+      "loss": 0.0467,
+      "step": 2469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666301026333766e-05,
+      "loss": 0.0578,
+      "step": 2470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666017538179005e-05,
+      "loss": 0.0741,
+      "step": 2471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657339537966363e-05,
+      "loss": 0.0548,
+      "step": 2472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6654502732276324e-05,
+      "loss": 0.0883,
+      "step": 2473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651664965129804e-05,
+      "loss": 0.0497,
+      "step": 2474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664882623693681e-05,
+      "loss": 0.0381,
+      "step": 2475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664598654810748e-05,
+      "loss": 0.0181,
+      "step": 2476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643145899052103e-05,
+      "loss": 0.0023,
+      "step": 2477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664030429018111e-05,
+      "loss": 0.007,
+      "step": 2478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637461721905047e-05,
+      "loss": 0.0778,
+      "step": 2479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6634618194634623e-05,
+      "loss": 0.1665,
+      "step": 2480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663177370878068e-05,
+      "loss": 0.0666,
+      "step": 2481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6628928264754187e-05,
+      "loss": 0.028,
+      "step": 2482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6626081862966265e-05,
+      "loss": 0.0781,
+      "step": 2483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623234503828167e-05,
+      "loss": 0.0044,
+      "step": 2484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662038618775129e-05,
+      "loss": 0.0501,
+      "step": 2485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6617536915147156e-05,
+      "loss": 0.0305,
+      "step": 2486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6614686686427437e-05,
+      "loss": 0.0206,
+      "step": 2487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661183550200394e-05,
+      "loss": 0.0694,
+      "step": 2488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6608983362288612e-05,
+      "loss": 0.1123,
+      "step": 2489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606130267693537e-05,
+      "loss": 0.0154,
+      "step": 2490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6603276218630932e-05,
+      "loss": 0.0587,
+      "step": 2491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660042121551316e-05,
+      "loss": 0.043,
+      "step": 2492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659756525875272e-05,
+      "loss": 0.0206,
+      "step": 2493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6594708348762236e-05,
+      "loss": 0.049,
+      "step": 2494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6591850485954487e-05,
+      "loss": 0.0902,
+      "step": 2495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588991670742384e-05,
+      "loss": 0.0009,
+      "step": 2496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6586131903538974e-05,
+      "loss": 0.0306,
+      "step": 2497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583271184757444e-05,
+      "loss": 0.0371,
+      "step": 2498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580409514811114e-05,
+      "loss": 0.0214,
+      "step": 2499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657754689411344e-05,
+      "loss": 0.0265,
+      "step": 2500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657468332307803e-05,
+      "loss": 0.0288,
+      "step": 2501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6571818802118607e-05,
+      "loss": 0.0125,
+      "step": 2502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656895333164905e-05,
+      "loss": 0.0281,
+      "step": 2503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656608691208337e-05,
+      "loss": 0.0096,
+      "step": 2504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6563219543835708e-05,
+      "loss": 0.0548,
+      "step": 2505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656035122732035e-05,
+      "loss": 0.0092,
+      "step": 2506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6557481962951713e-05,
+      "loss": 0.0615,
+      "step": 2507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6554611751144358e-05,
+      "loss": 0.0382,
+      "step": 2508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6551740592312978e-05,
+      "loss": 0.0038,
+      "step": 2509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654886848687241e-05,
+      "loss": 0.0333,
+      "step": 2510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654599543523761e-05,
+      "loss": 0.1195,
+      "step": 2511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6543121437823692e-05,
+      "loss": 0.0097,
+      "step": 2512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654024649504589e-05,
+      "loss": 0.0416,
+      "step": 2513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6537370607319587e-05,
+      "loss": 0.0217,
+      "step": 2514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65344937750603e-05,
+      "loss": 0.0744,
+      "step": 2515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653161599868367e-05,
+      "loss": 0.01,
+      "step": 2516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528737278605492e-05,
+      "loss": 0.0176,
+      "step": 2517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6525857615241686e-05,
+      "loss": 0.0047,
+      "step": 2518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522977009008315e-05,
+      "loss": 0.0217,
+      "step": 2519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6520095460321574e-05,
+      "loss": 0.0135,
+      "step": 2520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517212969597793e-05,
+      "loss": 0.0043,
+      "step": 2521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651432953725344e-05,
+      "loss": 0.0019,
+      "step": 2522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651144516370512e-05,
+      "loss": 0.029,
+      "step": 2523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650855984936958e-05,
+      "loss": 0.0137,
+      "step": 2524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6505673594663685e-05,
+      "loss": 0.1044,
+      "step": 2525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6502786400004454e-05,
+      "loss": 0.0023,
+      "step": 2526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499898265809034e-05,
+      "loss": 0.1187,
+      "step": 2527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497009192494705e-05,
+      "loss": 0.0065,
+      "step": 2528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6494119180478896e-05,
+      "loss": 0.0304,
+      "step": 2529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649122823017915e-05,
+      "loss": 0.0071,
+      "step": 2530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6488336342013166e-05,
+      "loss": 0.0799,
+      "step": 2531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485443516398767e-05,
+      "loss": 0.101,
+      "step": 2532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482549753753912e-05,
+      "loss": 0.0326,
+      "step": 2533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6479655054496705e-05,
+      "loss": 0.0881,
+      "step": 2534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647675941904537e-05,
+      "loss": 0.0191,
+      "step": 2535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647386284781828e-05,
+      "loss": 0.032,
+      "step": 2536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6470965341233934e-05,
+      "loss": 0.0008,
+      "step": 2537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646806689971097e-05,
+      "loss": 0.0865,
+      "step": 2538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6465167523668168e-05,
+      "loss": 0.0078,
+      "step": 2539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646226721352443e-05,
+      "loss": 0.0011,
+      "step": 2540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459365969698796e-05,
+      "loss": 0.038,
+      "step": 2541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456463792610446e-05,
+      "loss": 0.0789,
+      "step": 2542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453560682678698e-05,
+      "loss": 0.0067,
+      "step": 2543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6450656640322994e-05,
+      "loss": 0.0299,
+      "step": 2544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447751665962915e-05,
+      "loss": 0.0094,
+      "step": 2545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644484576001818e-05,
+      "loss": 0.0197,
+      "step": 2546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6441938922908644e-05,
+      "loss": 0.012,
+      "step": 2547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439031155054288e-05,
+      "loss": 0.0068,
+      "step": 2548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6436122456875233e-05,
+      "loss": 0.0043,
+      "step": 2549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6433212828791734e-05,
+      "loss": 0.0395,
+      "step": 2550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6430302271224178e-05,
+      "loss": 0.1494,
+      "step": 2551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427390784593094e-05,
+      "loss": 0.0305,
+      "step": 2552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642447836931914e-05,
+      "loss": 0.0017,
+      "step": 2553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421565025823104e-05,
+      "loss": 0.0368,
+      "step": 2554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641865075452591e-05,
+      "loss": 0.0282,
+      "step": 2555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641573555584862e-05,
+      "loss": 0.0588,
+      "step": 2556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6412819430212434e-05,
+      "loss": 0.0106,
+      "step": 2557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640990237803867e-05,
+      "loss": 0.0069,
+      "step": 2558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6406984399748794e-05,
+      "loss": 0.0866,
+      "step": 2559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64040654957644e-05,
+      "loss": 0.1891,
+      "step": 2560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6401145666507225e-05,
+      "loss": 0.0246,
+      "step": 2561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639822491239912e-05,
+      "loss": 0.0129,
+      "step": 2562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395303233862094e-05,
+      "loss": 0.0751,
+      "step": 2563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639238063131827e-05,
+      "loss": 0.0015,
+      "step": 2564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6389457105189913e-05,
+      "loss": 0.0028,
+      "step": 2565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386532655899418e-05,
+      "loss": 0.0009,
+      "step": 2566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638360728386932e-05,
+      "loss": 0.0049,
+      "step": 2567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638068098952228e-05,
+      "loss": 0.0234,
+      "step": 2568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6377753773281096e-05,
+      "loss": 0.0096,
+      "step": 2569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6374825635568695e-05,
+      "loss": 0.0137,
+      "step": 2570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6371896576808147e-05,
+      "loss": 0.0012,
+      "step": 2571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6368966597422647e-05,
+      "loss": 0.0209,
+      "step": 2572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366035697835518e-05,
+      "loss": 0.0017,
+      "step": 2573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636310387847023e-05,
+      "loss": 0.0171,
+      "step": 2574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6360171139750377e-05,
+      "loss": 0.0014,
+      "step": 2575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357237482099682e-05,
+      "loss": 0.0037,
+      "step": 2576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635430290594201e-05,
+      "loss": 0.0078,
+      "step": 2577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6351367411701357e-05,
+      "loss": 0.0627,
+      "step": 2578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348430999801844e-05,
+      "loss": 0.0597,
+      "step": 2579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6345493670667737e-05,
+      "loss": 0.1158,
+      "step": 2580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634255542472342e-05,
+      "loss": 0.0057,
+      "step": 2581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633961626239342e-05,
+      "loss": 0.0017,
+      "step": 2582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6336676184102394e-05,
+      "loss": 0.0658,
+      "step": 2583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633373519027513e-05,
+      "loss": 0.0008,
+      "step": 2584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6330793281336556e-05,
+      "loss": 0.0762,
+      "step": 2585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632785045771171e-05,
+      "loss": 0.0014,
+      "step": 2586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324906719825786e-05,
+      "loss": 0.0009,
+      "step": 2587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6321962068104102e-05,
+      "loss": 0.0709,
+      "step": 2588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631901650297211e-05,
+      "loss": 0.0058,
+      "step": 2589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6316070024855385e-05,
+      "loss": 0.1283,
+      "step": 2590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631312263417964e-05,
+      "loss": 0.0625,
+      "step": 2591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6310174331370728e-05,
+      "loss": 0.0579,
+      "step": 2592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6307225116854623e-05,
+      "loss": 0.0241,
+      "step": 2593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304274991057427e-05,
+      "loss": 0.0041,
+      "step": 2594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6301323954405387e-05,
+      "loss": 0.0008,
+      "step": 2595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6298372007324873e-05,
+      "loss": 0.08,
+      "step": 2596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629541915024239e-05,
+      "loss": 0.0801,
+      "step": 2597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629246538358457e-05,
+      "loss": 0.0883,
+      "step": 2598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.628951070777818e-05,
+      "loss": 0.0469,
+      "step": 2599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6286555123250125e-05,
+      "loss": 0.0036,
+      "step": 2600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628359863042742e-05,
+      "loss": 0.1803,
+      "step": 2601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628064122973724e-05,
+      "loss": 0.0392,
+      "step": 2602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627768292160687e-05,
+      "loss": 0.0018,
+      "step": 2603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274723706463728e-05,
+      "loss": 0.0315,
+      "step": 2604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6271763584735373e-05,
+      "loss": 0.0032,
+      "step": 2605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6268802556849486e-05,
+      "loss": 0.0217,
+      "step": 2606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626584062323389e-05,
+      "loss": 0.0055,
+      "step": 2607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626287778431652e-05,
+      "loss": 0.0363,
+      "step": 2608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259914040525463e-05,
+      "loss": 0.0696,
+      "step": 2609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256949392288924e-05,
+      "loss": 0.0467,
+      "step": 2610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6253983840035243e-05,
+      "loss": 0.0224,
+      "step": 2611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251017384192885e-05,
+      "loss": 0.0201,
+      "step": 2612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624805002519045e-05,
+      "loss": 0.0414,
+      "step": 2613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624508176345667e-05,
+      "loss": 0.091,
+      "step": 2614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6242112599420407e-05,
+      "loss": 0.011,
+      "step": 2615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623914253351065e-05,
+      "loss": 0.0548,
+      "step": 2616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236171566156523e-05,
+      "loss": 0.0746,
+      "step": 2617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6233199697787276e-05,
+      "loss": 0.0337,
+      "step": 2618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6230226928832288e-05,
+      "loss": 0.0443,
+      "step": 2619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6227253259721075e-05,
+      "loss": 0.0149,
+      "step": 2620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6224278690883277e-05,
+      "loss": 0.0743,
+      "step": 2621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221303222748667e-05,
+      "loss": 0.0013,
+      "step": 2622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218326855747146e-05,
+      "loss": 0.0761,
+      "step": 2623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215349590308744e-05,
+      "loss": 0.0524,
+      "step": 2624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6212371426863627e-05,
+      "loss": 0.0169,
+      "step": 2625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620939236584208e-05,
+      "loss": 0.0377,
+      "step": 2626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620641240767453e-05,
+      "loss": 0.0257,
+      "step": 2627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6203431552791526e-05,
+      "loss": 0.0846,
+      "step": 2628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200449801623747e-05,
+      "loss": 0.0401,
+      "step": 2629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6197467154602004e-05,
+      "loss": 0.072,
+      "step": 2630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194483612157232e-05,
+      "loss": 0.0579,
+      "step": 2631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61914991747205e-05,
+      "loss": 0.0486,
+      "step": 2632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6188513842723015e-05,
+      "loss": 0.013,
+      "step": 2633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185527616596096e-05,
+      "loss": 0.0142,
+      "step": 2634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61825404967712e-05,
+      "loss": 0.1039,
+      "step": 2635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617955248367991e-05,
+      "loss": 0.1768,
+      "step": 2636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617656357775395e-05,
+      "loss": 0.0838,
+      "step": 2637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617357377942515e-05,
+      "loss": 0.0259,
+      "step": 2638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6170583089125492e-05,
+      "loss": 0.0041,
+      "step": 2639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616759150728707e-05,
+      "loss": 0.0236,
+      "step": 2640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6164599034342122e-05,
+      "loss": 0.1184,
+      "step": 2641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161605670723004e-05,
+      "loss": 0.0121,
+      "step": 2642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61586114168622e-05,
+      "loss": 0.0246,
+      "step": 2643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6155616273192332e-05,
+      "loss": 0.0327,
+      "step": 2644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152620240146136e-05,
+      "loss": 0.0165,
+      "step": 2645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6149623318156493e-05,
+      "loss": 0.0035,
+      "step": 2646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61466255076564e-05,
+      "loss": 0.0172,
+      "step": 2647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614362680907899e-05,
+      "loss": 0.0258,
+      "step": 2648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614062722285752e-05,
+      "loss": 0.041,
+      "step": 2649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137626749425377e-05,
+      "loss": 0.0919,
+      "step": 2650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6134625389216076e-05,
+      "loss": 0.0057,
+      "step": 2651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131623142663255e-05,
+      "loss": 0.0234,
+      "step": 2652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6128620010200694e-05,
+      "loss": 0.039,
+      "step": 2653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6125615992262284e-05,
+      "loss": 0.0014,
+      "step": 2654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612261108928205e-05,
+      "loss": 0.0371,
+      "step": 2655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6119605301694156e-05,
+      "loss": 0.0595,
+      "step": 2656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611659862993288e-05,
+      "loss": 0.006,
+      "step": 2657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611359107443263e-05,
+      "loss": 0.0512,
+      "step": 2658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6110582635627946e-05,
+      "loss": 0.006,
+      "step": 2659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6107573313953495e-05,
+      "loss": 0.0085,
+      "step": 2660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6104563109844063e-05,
+      "loss": 0.0031,
+      "step": 2661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101552023734582e-05,
+      "loss": 0.087,
+      "step": 2662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609854005606009e-05,
+      "loss": 0.0223,
+      "step": 2663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609552720725577e-05,
+      "loss": 0.0436,
+      "step": 2664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609251347775692e-05,
+      "loss": 0.1574,
+      "step": 2665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608949886799897e-05,
+      "loss": 0.0118,
+      "step": 2666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608648337841748e-05,
+      "loss": 0.0169,
+      "step": 2667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608346700944813e-05,
+      "loss": 0.0456,
+      "step": 2668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6080449761526737e-05,
+      "loss": 0.063,
+      "step": 2669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607743163508923e-05,
+      "loss": 0.0918,
+      "step": 2670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074412630571685e-05,
+      "loss": 0.0753,
+      "step": 2671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071392748410287e-05,
+      "loss": 0.0819,
+      "step": 2672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606837198904136e-05,
+      "loss": 0.0729,
+      "step": 2673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6065350352901346e-05,
+      "loss": 0.0015,
+      "step": 2674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606232784042682e-05,
+      "loss": 0.0189,
+      "step": 2675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059304452054475e-05,
+      "loss": 0.0108,
+      "step": 2676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6056280188221142e-05,
+      "loss": 0.0148,
+      "step": 2677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605325504936377e-05,
+      "loss": 0.0403,
+      "step": 2678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605022903591944e-05,
+      "loss": 0.0154,
+      "step": 2679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604720214832536e-05,
+      "loss": 0.0128,
+      "step": 2680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604417438701885e-05,
+      "loss": 0.0048,
+      "step": 2681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6041145752437376e-05,
+      "loss": 0.0012,
+      "step": 2682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603811624501852e-05,
+      "loss": 0.0791,
+      "step": 2683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6035085865199992e-05,
+      "loss": 0.0009,
+      "step": 2684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6032054613419627e-05,
+      "loss": 0.0202,
+      "step": 2685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6029022490115383e-05,
+      "loss": 0.188,
+      "step": 2686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025989495725353e-05,
+      "loss": 0.0154,
+      "step": 2687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6022955630687752e-05,
+      "loss": 0.06,
+      "step": 2688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601992089544091e-05,
+      "loss": 0.0011,
+      "step": 2689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60168852904233e-05,
+      "loss": 0.0054,
+      "step": 2690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013848816073507e-05,
+      "loss": 0.0301,
+      "step": 2691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6010811472830253e-05,
+      "loss": 0.004,
+      "step": 2692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6007773261132376e-05,
+      "loss": 0.0076,
+      "step": 2693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600473418141884e-05,
+      "loss": 0.0497,
+      "step": 2694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600169423412875e-05,
+      "loss": 0.0283,
+      "step": 2695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998653419701306e-05,
+      "loss": 0.0454,
+      "step": 2696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5995611738575867e-05,
+      "loss": 0.0057,
+      "step": 2697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5992569191191892e-05,
+      "loss": 0.1514,
+      "step": 2698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598952577798898e-05,
+      "loss": 0.0035,
+      "step": 2699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986481499406846e-05,
+      "loss": 0.0158,
+      "step": 2700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5983436355885333e-05,
+      "loss": 0.0204,
+      "step": 2701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5980390347864413e-05,
+      "loss": 0.0334,
+      "step": 2702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597734347578418e-05,
+      "loss": 0.007,
+      "step": 2703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5974295740084853e-05,
+      "loss": 0.0032,
+      "step": 2704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5971247141206768e-05,
+      "loss": 0.0473,
+      "step": 2705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5968197679590406e-05,
+      "loss": 0.0607,
+      "step": 2706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5965147355676344e-05,
+      "loss": 0.0049,
+      "step": 2707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596209616990531e-05,
+      "loss": 0.0573,
+      "step": 2708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595904412271814e-05,
+      "loss": 0.0416,
+      "step": 2709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5955991214555807e-05,
+      "loss": 0.1351,
+      "step": 2710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5952937445859396e-05,
+      "loss": 0.0036,
+      "step": 2711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594988281707012e-05,
+      "loss": 0.03,
+      "step": 2712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5946827328629327e-05,
+      "loss": 0.0174,
+      "step": 2713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594377098097847e-05,
+      "loss": 0.0095,
+      "step": 2714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940713774559145e-05,
+      "loss": 0.0853,
+      "step": 2715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593765570981306e-05,
+      "loss": 0.0379,
+      "step": 2716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5934596787182047e-05,
+      "loss": 0.0224,
+      "step": 2717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593153700710807e-05,
+      "loss": 0.01,
+      "step": 2718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592847637003321e-05,
+      "loss": 0.0492,
+      "step": 2719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5925414876399678e-05,
+      "loss": 0.0229,
+      "step": 2720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5922352526649803e-05,
+      "loss": 0.0473,
+      "step": 2721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5919289321226036e-05,
+      "loss": 0.0481,
+      "step": 2722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916225260570964e-05,
+      "loss": 0.0577,
+      "step": 2723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591316034512728e-05,
+      "loss": 0.0427,
+      "step": 2724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910094575337814e-05,
+      "loss": 0.1065,
+      "step": 2725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590702795164551e-05,
+      "loss": 0.0054,
+      "step": 2726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590396047449345e-05,
+      "loss": 0.034,
+      "step": 2727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5900892144324823e-05,
+      "loss": 0.0439,
+      "step": 2728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5897822961582944e-05,
+      "loss": 0.0075,
+      "step": 2729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894752926711264e-05,
+      "loss": 0.0145,
+      "step": 2730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589168204015334e-05,
+      "loss": 0.0144,
+      "step": 2731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888610302352866e-05,
+      "loss": 0.0423,
+      "step": 2732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885537713753648e-05,
+      "loss": 0.0504,
+      "step": 2733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5882464274799623e-05,
+      "loss": 0.066,
+      "step": 2734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5879389985934854e-05,
+      "loss": 0.0214,
+      "step": 2735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5876314847603506e-05,
+      "loss": 0.0692,
+      "step": 2736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587323886024989e-05,
+      "loss": 0.0019,
+      "step": 2737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5870162024318433e-05,
+      "loss": 0.0158,
+      "step": 2738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586708434025368e-05,
+      "loss": 0.0454,
+      "step": 2739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5864005808500303e-05,
+      "loss": 0.0107,
+      "step": 2740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5860926429503093e-05,
+      "loss": 0.0036,
+      "step": 2741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5857846203706967e-05,
+      "loss": 0.0193,
+      "step": 2742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854765131556956e-05,
+      "loss": 0.0068,
+      "step": 2743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585168321349823e-05,
+      "loss": 0.0668,
+      "step": 2744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584860044997606e-05,
+      "loss": 0.0766,
+      "step": 2745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584551684143586e-05,
+      "loss": 0.0045,
+      "step": 2746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584243238832315e-05,
+      "loss": 0.0301,
+      "step": 2747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839347091083582e-05,
+      "loss": 0.1841,
+      "step": 2748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836260950162922e-05,
+      "loss": 0.0042,
+      "step": 2749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583317396600707e-05,
+      "loss": 0.0192,
+      "step": 2750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830086139062025e-05,
+      "loss": 0.019,
+      "step": 2751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582699746977394e-05,
+      "loss": 0.0592,
+      "step": 2752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823907958589062e-05,
+      "loss": 0.0016,
+      "step": 2753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582081760595377e-05,
+      "loss": 0.0065,
+      "step": 2754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5817726412314575e-05,
+      "loss": 0.0066,
+      "step": 2755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814634378118085e-05,
+      "loss": 0.0117,
+      "step": 2756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581154150381105e-05,
+      "loss": 0.0525,
+      "step": 2757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580844778984034e-05,
+      "loss": 0.0599,
+      "step": 2758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580535323665293e-05,
+      "loss": 0.0037,
+      "step": 2759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802257844695938e-05,
+      "loss": 0.0176,
+      "step": 2760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579916161441658e-05,
+      "loss": 0.1024,
+      "step": 2761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796064546262228e-05,
+      "loss": 0.0017,
+      "step": 2762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579296664068033e-05,
+      "loss": 0.0034,
+      "step": 2763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578986789811849e-05,
+      "loss": 0.0466,
+      "step": 2764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5786768319024414e-05,
+      "loss": 0.1064,
+      "step": 2765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5783667903845943e-05,
+      "loss": 0.0652,
+      "step": 2766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5780566653031028e-05,
+      "loss": 0.0829,
+      "step": 2767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5777464567027742e-05,
+      "loss": 0.0613,
+      "step": 2768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5774361646284285e-05,
+      "loss": 0.0049,
+      "step": 2769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5771257891248973e-05,
+      "loss": 0.047,
+      "step": 2770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576815330237024e-05,
+      "loss": 0.1463,
+      "step": 2771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5765047880096642e-05,
+      "loss": 0.0641,
+      "step": 2772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576194162487686e-05,
+      "loss": 0.0475,
+      "step": 2773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5758834537159695e-05,
+      "loss": 0.0021,
+      "step": 2774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5755726617394065e-05,
+      "loss": 0.0618,
+      "step": 2775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5752617866029005e-05,
+      "loss": 0.0202,
+      "step": 2776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574950828351367e-05,
+      "loss": 0.0973,
+      "step": 2777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574639787029735e-05,
+      "loss": 0.0532,
+      "step": 2778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5743286626829437e-05,
+      "loss": 0.246,
+      "step": 2779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5740174553559454e-05,
+      "loss": 0.0539,
+      "step": 2780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737061650937033e-05,
+      "loss": 0.086,
+      "step": 2781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573394791941194e-05,
+      "loss": 0.0162,
+      "step": 2782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573083335943405e-05,
+      "loss": 0.0206,
+      "step": 2783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727717971453358e-05,
+      "loss": 0.0505,
+      "step": 2784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572460175591999e-05,
+      "loss": 0.0285,
+      "step": 2785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5721484713284173e-05,
+      "loss": 0.0154,
+      "step": 2786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5718366843996275e-05,
+      "loss": 0.0639,
+      "step": 2787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5715248148506766e-05,
+      "loss": 0.0555,
+      "step": 2788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5712128627266242e-05,
+      "loss": 0.0044,
+      "step": 2789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5709008280725415e-05,
+      "loss": 0.0061,
+      "step": 2790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5705887109335124e-05,
+      "loss": 0.0313,
+      "step": 2791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702765113546324e-05,
+      "loss": 0.0536,
+      "step": 2792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569964229381008e-05,
+      "loss": 0.0493,
+      "step": 2793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569651865057759e-05,
+      "loss": 0.0609,
+      "step": 2794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693394184300164e-05,
+      "loss": 0.0398,
+      "step": 2795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569026889542923e-05,
+      "loss": 0.0749,
+      "step": 2796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5687142784416334e-05,
+      "loss": 0.0882,
+      "step": 2797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568401585171315e-05,
+      "loss": 0.0501,
+      "step": 2798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568088809777145e-05,
+      "loss": 0.0366,
+      "step": 2799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5677759523043155e-05,
+      "loss": 0.0503,
+      "step": 2800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5674630127980278e-05,
+      "loss": 0.0412,
+      "step": 2801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671499913034962e-05,
+      "loss": 0.1599,
+      "step": 2802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566836887865947e-05,
+      "loss": 0.0285,
+      "step": 2803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5665237025306178e-05,
+      "loss": 0.14,
+      "step": 2804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5662104353427582e-05,
+      "loss": 0.0399,
+      "step": 2805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658970863476298e-05,
+      "loss": 0.0282,
+      "step": 2806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5655836555905062e-05,
+      "loss": 0.0568,
+      "step": 2807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565270143116672e-05,
+      "loss": 0.0205,
+      "step": 2808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5649565489714244e-05,
+      "loss": 0.026,
+      "step": 2809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5646428732000726e-05,
+      "loss": 0.094,
+      "step": 2810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564329115847936e-05,
+      "loss": 0.0565,
+      "step": 2811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640152769603482e-05,
+      "loss": 0.0343,
+      "step": 2812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563701356582652e-05,
+      "loss": 0.0344,
+      "step": 2813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5633873547602046e-05,
+      "loss": 0.0316,
+      "step": 2814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5630732715383726e-05,
+      "loss": 0.0341,
+      "step": 2815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562759106962536e-05,
+      "loss": 0.0124,
+      "step": 2816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562444861078086e-05,
+      "loss": 0.033,
+      "step": 2817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5621305339304247e-05,
+      "loss": 0.0689,
+      "step": 2818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5618161255649677e-05,
+      "loss": 0.058,
+      "step": 2819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561501636027141e-05,
+      "loss": 0.0222,
+      "step": 2820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5611870653623826e-05,
+      "loss": 0.0871,
+      "step": 2821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560872413616142e-05,
+      "loss": 0.0169,
+      "step": 2822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5605576808338818e-05,
+      "loss": 0.1002,
+      "step": 2823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5602428670610748e-05,
+      "loss": 0.0087,
+      "step": 2824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5599279723432054e-05,
+      "loss": 0.1295,
+      "step": 2825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596129967257704e-05,
+      "loss": 0.046,
+      "step": 2826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5592979402542786e-05,
+      "loss": 0.0109,
+      "step": 2827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55898280297425e-05,
+      "loss": 0.0088,
+      "step": 2828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5586675849312153e-05,
+      "loss": 0.009,
+      "step": 2829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558352286170719e-05,
+      "loss": 0.0298,
+      "step": 2830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558036906738316e-05,
+      "loss": 0.0079,
+      "step": 2831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557721446679573e-05,
+      "loss": 0.0286,
+      "step": 2832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5574059060400674e-05,
+      "loss": 0.0257,
+      "step": 2833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5570902848653903e-05,
+      "loss": 0.0302,
+      "step": 2834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5567745832011424e-05,
+      "loss": 0.0152,
+      "step": 2835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5564588010929375e-05,
+      "loss": 0.0489,
+      "step": 2836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561429385864005e-05,
+      "loss": 0.0397,
+      "step": 2837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555826995727168e-05,
+      "loss": 0.006,
+      "step": 2838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5555109725608874e-05,
+      "loss": 0.0308,
+      "step": 2839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5551948691332192e-05,
+      "loss": 0.0115,
+      "step": 2840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554878685489834e-05,
+      "loss": 0.1284,
+      "step": 2841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5545624216764145e-05,
+      "loss": 0.0697,
+      "step": 2842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542460777386564e-05,
+      "loss": 0.031,
+      "step": 2843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5539296537222645e-05,
+      "loss": 0.0054,
+      "step": 2844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553613149672957e-05,
+      "loss": 0.0409,
+      "step": 2845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5532965656364627e-05,
+      "loss": 0.0751,
+      "step": 2846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5529799016585226e-05,
+      "loss": 0.0032,
+      "step": 2847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552663157784889e-05,
+      "loss": 0.0115,
+      "step": 2848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552346334061326e-05,
+      "loss": 0.0941,
+      "step": 2849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552029430533608e-05,
+      "loss": 0.0503,
+      "step": 2850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551712447247523e-05,
+      "loss": 0.0062,
+      "step": 2851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5513953842488682e-05,
+      "loss": 0.1636,
+      "step": 2852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5510782415834547e-05,
+      "loss": 0.0248,
+      "step": 2853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5507610192971034e-05,
+      "loss": 0.0176,
+      "step": 2854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5504437174356472e-05,
+      "loss": 0.0029,
+      "step": 2855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5501263360449303e-05,
+      "loss": 0.0317,
+      "step": 2856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5498088751708094e-05,
+      "loss": 0.0054,
+      "step": 2857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5494913348591512e-05,
+      "loss": 0.009,
+      "step": 2858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5491737151558352e-05,
+      "loss": 0.0955,
+      "step": 2859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548856016106751e-05,
+      "loss": 0.0303,
+      "step": 2860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548538237757801e-05,
+      "loss": 0.0639,
+      "step": 2861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5482203801548986e-05,
+      "loss": 0.0086,
+      "step": 2862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547902443343968e-05,
+      "loss": 0.0835,
+      "step": 2863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475844273709453e-05,
+      "loss": 0.0235,
+      "step": 2864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472663322817786e-05,
+      "loss": 0.0669,
+      "step": 2865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469481581224274e-05,
+      "loss": 0.0046,
+      "step": 2866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466299049388606e-05,
+      "loss": 0.0636,
+      "step": 2867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463115727770616e-05,
+      "loss": 0.0521,
+      "step": 2868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5459931616830228e-05,
+      "loss": 0.0301,
+      "step": 2869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456746717027494e-05,
+      "loss": 0.0056,
+      "step": 2870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5453561028822573e-05,
+      "loss": 0.0149,
+      "step": 2871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5450374552675743e-05,
+      "loss": 0.0119,
+      "step": 2872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5447187289047386e-05,
+      "loss": 0.0146,
+      "step": 2873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544399923839801e-05,
+      "loss": 0.1175,
+      "step": 2874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5440810401188228e-05,
+      "loss": 0.0063,
+      "step": 2875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5437620777878776e-05,
+      "loss": 0.0671,
+      "step": 2876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543443036893049e-05,
+      "loss": 0.0213,
+      "step": 2877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543123917480433e-05,
+      "loss": 0.0283,
+      "step": 2878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5428047195961368e-05,
+      "loss": 0.1838,
+      "step": 2879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542485443286278e-05,
+      "loss": 0.0093,
+      "step": 2880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5421660885969875e-05,
+      "loss": 0.003,
+      "step": 2881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5418466555744058e-05,
+      "loss": 0.0786,
+      "step": 2882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541527144264685e-05,
+      "loss": 0.0038,
+      "step": 2883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541207554713989e-05,
+      "loss": 0.0674,
+      "step": 2884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408878869684926e-05,
+      "loss": 0.0754,
+      "step": 2885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5405681410743825e-05,
+      "loss": 0.0781,
+      "step": 2886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5402483170778563e-05,
+      "loss": 0.1065,
+      "step": 2887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5399284150251217e-05,
+      "loss": 0.0691,
+      "step": 2888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396084349624e-05,
+      "loss": 0.0024,
+      "step": 2889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5392883769359223e-05,
+      "loss": 0.0293,
+      "step": 2890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389682409919308e-05,
+      "loss": 0.0034,
+      "step": 2891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53864802717668e-05,
+      "loss": 0.0519,
+      "step": 2892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5383277355364348e-05,
+      "loss": 0.0141,
+      "step": 2893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5380073661174716e-05,
+      "loss": 0.0539,
+      "step": 2894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5376869189660784e-05,
+      "loss": 0.0048,
+      "step": 2895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5373663941285536e-05,
+      "loss": 0.0032,
+      "step": 2896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370457916512072e-05,
+      "loss": 0.0553,
+      "step": 2897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5367251115803607e-05,
+      "loss": 0.0079,
+      "step": 2898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5364043539623472e-05,
+      "loss": 0.0016,
+      "step": 2899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53608351884351e-05,
+      "loss": 0.0118,
+      "step": 2900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357626062702035e-05,
+      "loss": 0.0034,
+      "step": 2901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5354416162887943e-05,
+      "loss": 0.0036,
+      "step": 2902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5351205489456596e-05,
+      "loss": 0.0354,
+      "step": 2903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5347994042871885e-05,
+      "loss": 0.0129,
+      "step": 2904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53447818235978e-05,
+      "loss": 0.0383,
+      "step": 2905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341568832098447e-05,
+      "loss": 0.0485,
+      "step": 2906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533835506883805e-05,
+      "loss": 0.0577,
+      "step": 2907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5335140534280944e-05,
+      "loss": 0.0014,
+      "step": 2908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533192522889156e-05,
+      "loss": 0.014,
+      "step": 2909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328709153134466e-05,
+      "loss": 0.0688,
+      "step": 2910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5325492307474314e-05,
+      "loss": 0.0868,
+      "step": 2911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532227469237589e-05,
+      "loss": 0.0014,
+      "step": 2912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5319056308304082e-05,
+      "loss": 0.0082,
+      "step": 2913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315837155723885e-05,
+      "loss": 0.0667,
+      "step": 2914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5312617235100408e-05,
+      "loss": 0.0196,
+      "step": 2915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530939654689887e-05,
+      "loss": 0.0114,
+      "step": 2916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5306175091584608e-05,
+      "loss": 0.0519,
+      "step": 2917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5302952869623066e-05,
+      "loss": 0.0534,
+      "step": 2918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529972988147979e-05,
+      "loss": 0.0463,
+      "step": 2919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529650612762045e-05,
+      "loss": 0.0031,
+      "step": 2920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529328160851082e-05,
+      "loss": 0.0854,
+      "step": 2921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5290056324616786e-05,
+      "loss": 0.0266,
+      "step": 2922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528683027640434e-05,
+      "loss": 0.0869,
+      "step": 2923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528360346433959e-05,
+      "loss": 0.1075,
+      "step": 2924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5280375888888754e-05,
+      "loss": 0.0028,
+      "step": 2925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5277147550518156e-05,
+      "loss": 0.0719,
+      "step": 2926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273918449694236e-05,
+      "loss": 0.0009,
+      "step": 2927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527068858688354e-05,
+      "loss": 0.0242,
+      "step": 2928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5267457962552725e-05,
+      "loss": 0.0057,
+      "step": 2929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526422657716856e-05,
+      "loss": 0.0082,
+      "step": 2930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5260994431197916e-05,
+      "loss": 0.0145,
+      "step": 2931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257761525107787e-05,
+      "loss": 0.0185,
+      "step": 2932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5254527859365271e-05,
+      "loss": 0.0987,
+      "step": 2933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5251293434437573e-05,
+      "loss": 0.0135,
+      "step": 2934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248058250792008e-05,
+      "loss": 0.0075,
+      "step": 2935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5244822308896001e-05,
+      "loss": 0.0163,
+      "step": 2936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524158560921709e-05,
+      "loss": 0.0237,
+      "step": 2937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523834815222292e-05,
+      "loss": 0.0798,
+      "step": 2938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5235109938381243e-05,
+      "loss": 0.0762,
+      "step": 2939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5231870968159928e-05,
+      "loss": 0.0873,
+      "step": 2940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522863124202694e-05,
+      "loss": 0.0068,
+      "step": 2941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5225390760450372e-05,
+      "loss": 0.0066,
+      "step": 2942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222149523898408e-05,
+      "loss": 0.0133,
+      "step": 2943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521890753283935e-05,
+      "loss": 0.0024,
+      "step": 2944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5215664787741609e-05,
+      "loss": 0.007,
+      "step": 2945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212421289073703e-05,
+      "loss": 0.0511,
+      "step": 2946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520917703730426e-05,
+      "loss": 0.0712,
+      "step": 2947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5205932032902018e-05,
+      "loss": 0.0068,
+      "step": 2948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202686276335816e-05,
+      "loss": 0.0507,
+      "step": 2949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5199439768074613e-05,
+      "loss": 0.0086,
+      "step": 2950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196192508587469e-05,
+      "loss": 0.1089,
+      "step": 2951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5192944498343556e-05,
+      "loss": 0.0139,
+      "step": 2952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5189695737812153e-05,
+      "loss": 0.0227,
+      "step": 2953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5186446227462646e-05,
+      "loss": 0.0015,
+      "step": 2954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183195967764536e-05,
+      "loss": 0.07,
+      "step": 2955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5179944959187423e-05,
+      "loss": 0.0297,
+      "step": 2956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5176693202201023e-05,
+      "loss": 0.0777,
+      "step": 2957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173440697275151e-05,
+      "loss": 0.0228,
+      "step": 2958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170187444879738e-05,
+      "loss": 0.0809,
+      "step": 2959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166933445484821e-05,
+      "loss": 0.0112,
+      "step": 2960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163678699560549e-05,
+      "loss": 0.0041,
+      "step": 2961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5160423207577171e-05,
+      "loss": 0.0821,
+      "step": 2962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5157166970005045e-05,
+      "loss": 0.0058,
+      "step": 2963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5153909987314641e-05,
+      "loss": 0.1561,
+      "step": 2964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5150652259976535e-05,
+      "loss": 0.0078,
+      "step": 2965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147393788461411e-05,
+      "loss": 0.0997,
+      "step": 2966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5144134573240056e-05,
+      "loss": 0.0016,
+      "step": 2967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5140874614783374e-05,
+      "loss": 0.0964,
+      "step": 2968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513761391356237e-05,
+      "loss": 0.1227,
+      "step": 2969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5134352470048153e-05,
+      "loss": 0.0027,
+      "step": 2970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131090284711942e-05,
+      "loss": 0.0154,
+      "step": 2971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5127827358025069e-05,
+      "loss": 0.007,
+      "step": 2972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124563690458966e-05,
+      "loss": 0.003,
+      "step": 2973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5121299282485177e-05,
+      "loss": 0.0354,
+      "step": 2974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118034134575347e-05,
+      "loss": 0.053,
+      "step": 2975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114768247201235e-05,
+      "loss": 0.0261,
+      "step": 2976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5111501620834701e-05,
+      "loss": 0.1133,
+      "step": 2977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108234255947712e-05,
+      "loss": 0.007,
+      "step": 2978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104966153012349e-05,
+      "loss": 0.0022,
+      "step": 2979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510169731250079e-05,
+      "loss": 0.0521,
+      "step": 2980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098427734885324e-05,
+      "loss": 0.0789,
+      "step": 2981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095157420638349e-05,
+      "loss": 0.0039,
+      "step": 2982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091886370232369e-05,
+      "loss": 0.0567,
+      "step": 2983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5088614584139985e-05,
+      "loss": 0.0712,
+      "step": 2984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085342062833915e-05,
+      "loss": 0.1185,
+      "step": 2985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508206880678698e-05,
+      "loss": 0.0267,
+      "step": 2986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078794816472111e-05,
+      "loss": 0.0353,
+      "step": 2987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075520092362332e-05,
+      "loss": 0.0206,
+      "step": 2988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5072244634930788e-05,
+      "loss": 0.0028,
+      "step": 2989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068968444650725e-05,
+      "loss": 0.0018,
+      "step": 2990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506569152199549e-05,
+      "loss": 0.0036,
+      "step": 2991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506241386743854e-05,
+      "loss": 0.0084,
+      "step": 2992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059135481453444e-05,
+      "loss": 0.0026,
+      "step": 2993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055856364513858e-05,
+      "loss": 0.0774,
+      "step": 2994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052576517093567e-05,
+      "loss": 0.0084,
+      "step": 2995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5049295939666445e-05,
+      "loss": 0.0108,
+      "step": 2996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5046014632706479e-05,
+      "loss": 0.0434,
+      "step": 2997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5042732596687755e-05,
+      "loss": 0.0412,
+      "step": 2998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503944983208447e-05,
+      "loss": 0.0857,
+      "step": 2999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036166339370929e-05,
+      "loss": 0.0041,
+      "step": 3000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503288211902153e-05,
+      "loss": 0.0304,
+      "step": 3001
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502959717151079e-05,
+      "loss": 0.0937,
+      "step": 3002
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5026311497313327e-05,
+      "loss": 0.0442,
+      "step": 3003
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023025096903854e-05,
+      "loss": 0.0778,
+      "step": 3004
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019737970757206e-05,
+      "loss": 0.0717,
+      "step": 3005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5016450119348309e-05,
+      "loss": 0.0183,
+      "step": 3006
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013161543152195e-05,
+      "loss": 0.0039,
+      "step": 3007
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009872242644013e-05,
+      "loss": 0.0414,
+      "step": 3008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5006582218299004e-05,
+      "loss": 0.0506,
+      "step": 3009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003291470592513e-05,
+      "loss": 0.008,
+      "step": 3010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.0013,
+      "step": 3011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996707806997024e-05,
+      "loss": 0.0054,
+      "step": 3012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993414892059245e-05,
+      "loss": 0.0375,
+      "step": 3013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4990121255662431e-05,
+      "loss": 0.0058,
+      "step": 3014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986826898282453e-05,
+      "loss": 0.0562,
+      "step": 3015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4983531820395287e-05,
+      "loss": 0.038,
+      "step": 3016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980236022477015e-05,
+      "loss": 0.0135,
+      "step": 3017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4976939505003819e-05,
+      "loss": 0.0589,
+      "step": 3018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4973642268451985e-05,
+      "loss": 0.0132,
+      "step": 3019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970344313297904e-05,
+      "loss": 0.0109,
+      "step": 3020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967045640018074e-05,
+      "loss": 0.0469,
+      "step": 3021
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4963746249089098e-05,
+      "loss": 0.0329,
+      "step": 3022
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.496044614098767e-05,
+      "loss": 0.0141,
+      "step": 3023
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957145316190604e-05,
+      "loss": 0.0011,
+      "step": 3024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4953843775174807e-05,
+      "loss": 0.0348,
+      "step": 3025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4950541518417292e-05,
+      "loss": 0.0748,
+      "step": 3026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494723854639518e-05,
+      "loss": 0.0192,
+      "step": 3027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4943934859585685e-05,
+      "loss": 0.0393,
+      "step": 3028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940630458466135e-05,
+      "loss": 0.0062,
+      "step": 3029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4937325343513953e-05,
+      "loss": 0.1655,
+      "step": 3030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934019515206675e-05,
+      "loss": 0.0035,
+      "step": 3031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493071297402193e-05,
+      "loss": 0.0159,
+      "step": 3032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927405720437454e-05,
+      "loss": 0.0309,
+      "step": 3033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4924097754931084e-05,
+      "loss": 0.002,
+      "step": 3034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4920789077980766e-05,
+      "loss": 0.0116,
+      "step": 3035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917479690064545e-05,
+      "loss": 0.0478,
+      "step": 3036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4914169591660561e-05,
+      "loss": 0.0025,
+      "step": 3037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910858783247074e-05,
+      "loss": 0.0955,
+      "step": 3038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907547265302426e-05,
+      "loss": 0.0019,
+      "step": 3039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904235038305084e-05,
+      "loss": 0.0358,
+      "step": 3040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900922102733596e-05,
+      "loss": 0.0081,
+      "step": 3041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4897608459066624e-05,
+      "loss": 0.0997,
+      "step": 3042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489429410778293e-05,
+      "loss": 0.0961,
+      "step": 3043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890979049361383e-05,
+      "loss": 0.0666,
+      "step": 3044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887663284280947e-05,
+      "loss": 0.0139,
+      "step": 3045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4884346813020685e-05,
+      "loss": 0.0155,
+      "step": 3046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4881029636059774e-05,
+      "loss": 0.0183,
+      "step": 3047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877711753877483e-05,
+      "loss": 0.0045,
+      "step": 3048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874393166953194e-05,
+      "loss": 0.0352,
+      "step": 3049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4871073875766375e-05,
+      "loss": 0.0768,
+      "step": 3050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4867753880796607e-05,
+      "loss": 0.0035,
+      "step": 3051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4864433182523572e-05,
+      "loss": 0.0258,
+      "step": 3052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486111178142705e-05,
+      "loss": 0.0629,
+      "step": 3053
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485778967798692e-05,
+      "loss": 0.024,
+      "step": 3054
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485446687268317e-05,
+      "loss": 0.0362,
+      "step": 3055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851143365995886e-05,
+      "loss": 0.0583,
+      "step": 3056
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847819158405258e-05,
+      "loss": 0.0938,
+      "step": 3057
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484449425039157e-05,
+      "loss": 0.0338,
+      "step": 3058
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4841168642435214e-05,
+      "loss": 0.0385,
+      "step": 3059
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837842335016676e-05,
+      "loss": 0.0813,
+      "step": 3060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4834515328616555e-05,
+      "loss": 0.0267,
+      "step": 3061
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831187623715542e-05,
+      "loss": 0.0343,
+      "step": 3062
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4827859220794425e-05,
+      "loss": 0.0788,
+      "step": 3063
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824530120334107e-05,
+      "loss": 0.0042,
+      "step": 3064
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821200322815577e-05,
+      "loss": 0.0649,
+      "step": 3065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817869828719933e-05,
+      "loss": 0.0415,
+      "step": 3066
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4814538638528375e-05,
+      "loss": 0.0016,
+      "step": 3067
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811206752722196e-05,
+      "loss": 0.0495,
+      "step": 3068
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807874171782795e-05,
+      "loss": 0.0641,
+      "step": 3069
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804540896191673e-05,
+      "loss": 0.0257,
+      "step": 3070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4801206926430425e-05,
+      "loss": 0.0244,
+      "step": 3071
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479787226298075e-05,
+      "loss": 0.0153,
+      "step": 3072
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4794536906324451e-05,
+      "loss": 0.0062,
+      "step": 3073
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4791200856943422e-05,
+      "loss": 0.0033,
+      "step": 3074
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787864115319668e-05,
+      "loss": 0.0238,
+      "step": 3075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4784526681935282e-05,
+      "loss": 0.0056,
+      "step": 3076
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4781188557272468e-05,
+      "loss": 0.0054,
+      "step": 3077
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777849741813524e-05,
+      "loss": 0.0332,
+      "step": 3078
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4774510236040853e-05,
+      "loss": 0.0782,
+      "step": 3079
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771170040436946e-05,
+      "loss": 0.0164,
+      "step": 3080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4767829155484407e-05,
+      "loss": 0.0023,
+      "step": 3081
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764487581665931e-05,
+      "loss": 0.0142,
+      "step": 3082
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761145319464319e-05,
+      "loss": 0.0874,
+      "step": 3083
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757802369362466e-05,
+      "loss": 0.0638,
+      "step": 3084
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4754458731843366e-05,
+      "loss": 0.0469,
+      "step": 3085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751114407390122e-05,
+      "loss": 0.026,
+      "step": 3086
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747769396485917e-05,
+      "loss": 0.0129,
+      "step": 3087
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474442369961406e-05,
+      "loss": 0.0048,
+      "step": 3088
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741077317257933e-05,
+      "loss": 0.0463,
+      "step": 3089
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4737730249901032e-05,
+      "loss": 0.2188,
+      "step": 3090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4734382498026946e-05,
+      "loss": 0.0533,
+      "step": 3091
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4731034062119373e-05,
+      "loss": 0.0066,
+      "step": 3092
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4727684942662095e-05,
+      "loss": 0.0226,
+      "step": 3093
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4724335140139e-05,
+      "loss": 0.0581,
+      "step": 3094
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4720984655034076e-05,
+      "loss": 0.0321,
+      "step": 3095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4717633487831407e-05,
+      "loss": 0.1289,
+      "step": 3096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471428163901518e-05,
+      "loss": 0.0216,
+      "step": 3097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4710929109069674e-05,
+      "loss": 0.064,
+      "step": 3098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4707575898479267e-05,
+      "loss": 0.0391,
+      "step": 3099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4704222007728446e-05,
+      "loss": 0.0024,
+      "step": 3100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470086743730178e-05,
+      "loss": 0.0409,
+      "step": 3101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697512187683952e-05,
+      "loss": 0.0131,
+      "step": 3102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694156259359727e-05,
+      "loss": 0.0674,
+      "step": 3103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469079965281398e-05,
+      "loss": 0.2116,
+      "step": 3104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4687442368531683e-05,
+      "loss": 0.0183,
+      "step": 3105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684084406997903e-05,
+      "loss": 0.0307,
+      "step": 3106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680725768697804e-05,
+      "loss": 0.0438,
+      "step": 3107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4677366454116647e-05,
+      "loss": 0.0365,
+      "step": 3108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4674006463739793e-05,
+      "loss": 0.0196,
+      "step": 3109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467064579805271e-05,
+      "loss": 0.0168,
+      "step": 3110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466728445754094e-05,
+      "loss": 0.0973,
+      "step": 3111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4663922442690144e-05,
+      "loss": 0.0371,
+      "step": 3112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660559753986072e-05,
+      "loss": 0.0387,
+      "step": 3113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657196391914572e-05,
+      "loss": 0.0708,
+      "step": 3114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653832356961593e-05,
+      "loss": 0.0941,
+      "step": 3115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650467649613168e-05,
+      "loss": 0.0393,
+      "step": 3116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4647102270355449e-05,
+      "loss": 0.0505,
+      "step": 3117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643736219674665e-05,
+      "loss": 0.2319,
+      "step": 3118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4640369498057155e-05,
+      "loss": 0.1687,
+      "step": 3119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637002105989344e-05,
+      "loss": 0.2768,
+      "step": 3120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4633634043957767e-05,
+      "loss": 0.0263,
+      "step": 3121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630265312449042e-05,
+      "loss": 0.0531,
+      "step": 3122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626895911949896e-05,
+      "loss": 0.0221,
+      "step": 3123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623525842947141e-05,
+      "loss": 0.036,
+      "step": 3124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4620155105927699e-05,
+      "loss": 0.0887,
+      "step": 3125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4616783701378573e-05,
+      "loss": 0.0195,
+      "step": 3126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461341162978688e-05,
+      "loss": 0.0562,
+      "step": 3127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4610038891639818e-05,
+      "loss": 0.0321,
+      "step": 3128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4606665487424685e-05,
+      "loss": 0.1498,
+      "step": 3129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460329141762888e-05,
+      "loss": 0.0351,
+      "step": 3130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4599916682739898e-05,
+      "loss": 0.008,
+      "step": 3131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4596541283245326e-05,
+      "loss": 0.0362,
+      "step": 3132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459316521963285e-05,
+      "loss": 0.007,
+      "step": 3133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589788492390247e-05,
+      "loss": 0.0824,
+      "step": 3134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45864111020054e-05,
+      "loss": 0.0537,
+      "step": 3135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583033048966273e-05,
+      "loss": 0.0068,
+      "step": 3136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4579654333760942e-05,
+      "loss": 0.0071,
+      "step": 3137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4576274956877566e-05,
+      "loss": 0.0078,
+      "step": 3138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572894918804406e-05,
+      "loss": 0.0482,
+      "step": 3139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4569514220029817e-05,
+      "loss": 0.0921,
+      "step": 3140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456613286104225e-05,
+      "loss": 0.0688,
+      "step": 3141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562750842330251e-05,
+      "loss": 0.0156,
+      "step": 3142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455936816438246e-05,
+      "loss": 0.1285,
+      "step": 3143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4555984827687613e-05,
+      "loss": 0.0412,
+      "step": 3144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4552600832734542e-05,
+      "loss": 0.0566,
+      "step": 3145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549216180012175e-05,
+      "loss": 0.0079,
+      "step": 3146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545830870009532e-05,
+      "loss": 0.0448,
+      "step": 3147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542444903215728e-05,
+      "loss": 0.0179,
+      "step": 3148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4539058280119982e-05,
+      "loss": 0.031,
+      "step": 3149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4535671001211595e-05,
+      "loss": 0.0038,
+      "step": 3150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4532283066979967e-05,
+      "loss": 0.0349,
+      "step": 3151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528894477914592e-05,
+      "loss": 0.0248,
+      "step": 3152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525505234505066e-05,
+      "loss": 0.1355,
+      "step": 3153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522115337241074e-05,
+      "loss": 0.0279,
+      "step": 3154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518724786612394e-05,
+      "loss": 0.1877,
+      "step": 3155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515333583108896e-05,
+      "loss": 0.0058,
+      "step": 3156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4511941727220551e-05,
+      "loss": 0.0597,
+      "step": 3157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4508549219437423e-05,
+      "loss": 0.0272,
+      "step": 3158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505156060249668e-05,
+      "loss": 0.0327,
+      "step": 3159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4501762250147535e-05,
+      "loss": 0.0044,
+      "step": 3160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4498367789621369e-05,
+      "loss": 0.0178,
+      "step": 3161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4494972679161611e-05,
+      "loss": 0.0207,
+      "step": 3162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4491576919258794e-05,
+      "loss": 0.034,
+      "step": 3163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448818051040354e-05,
+      "loss": 0.0215,
+      "step": 3164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448478345308657e-05,
+      "loss": 0.1322,
+      "step": 3165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481385747798705e-05,
+      "loss": 0.0603,
+      "step": 3166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4477987395030844e-05,
+      "loss": 0.0095,
+      "step": 3167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4474588395273994e-05,
+      "loss": 0.0455,
+      "step": 3168
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471188749019246e-05,
+      "loss": 0.0352,
+      "step": 3169
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446778845675779e-05,
+      "loss": 0.0518,
+      "step": 3170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464387518980905e-05,
+      "loss": 0.0345,
+      "step": 3171
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446098593617997e-05,
+      "loss": 0.015,
+      "step": 3172
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4457583708846448e-05,
+      "loss": 0.0279,
+      "step": 3173
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44541808374719e-05,
+      "loss": 0.0031,
+      "step": 3174
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4450777322547984e-05,
+      "loss": 0.0272,
+      "step": 3175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444737316456645e-05,
+      "loss": 0.0964,
+      "step": 3176
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4443968364019124e-05,
+      "loss": 0.0473,
+      "step": 3177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4440562921397954e-05,
+      "loss": 0.0073,
+      "step": 3178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4437156837194955e-05,
+      "loss": 0.0913,
+      "step": 3179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433750111902252e-05,
+      "loss": 0.0025,
+      "step": 3180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430342746012049e-05,
+      "loss": 0.0091,
+      "step": 3181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4426934740016655e-05,
+      "loss": 0.0015,
+      "step": 3182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423526094408467e-05,
+      "loss": 0.0231,
+      "step": 3183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4420116809679966e-05,
+      "loss": 0.0179,
+      "step": 3184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4416706886323741e-05,
+      "loss": 0.0052,
+      "step": 3185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441329632483246e-05,
+      "loss": 0.1091,
+      "step": 3186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409885125698886e-05,
+      "loss": 0.0393,
+      "step": 3187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440647328941588e-05,
+      "loss": 0.0187,
+      "step": 3188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4403060816476393e-05,
+      "loss": 0.0306,
+      "step": 3189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4399647707373462e-05,
+      "loss": 0.0203,
+      "step": 3190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4396233962600223e-05,
+      "loss": 0.003,
+      "step": 3191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392819582649898e-05,
+      "loss": 0.0358,
+      "step": 3192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4389404568015804e-05,
+      "loss": 0.1058,
+      "step": 3193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4385988919191353e-05,
+      "loss": 0.0057,
+      "step": 3194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4382572636670042e-05,
+      "loss": 0.0121,
+      "step": 3195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379155720945464e-05,
+      "loss": 0.0078,
+      "step": 3196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375738172511305e-05,
+      "loss": 0.0838,
+      "step": 3197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4372319991861334e-05,
+      "loss": 0.0027,
+      "step": 3198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368901179489419e-05,
+      "loss": 0.0018,
+      "step": 3199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4365481735889515e-05,
+      "loss": 0.0032,
+      "step": 3200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4362061661555675e-05,
+      "loss": 0.0299,
+      "step": 3201
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4358640956982038e-05,
+      "loss": 0.0728,
+      "step": 3202
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435521962266283e-05,
+      "loss": 0.0424,
+      "step": 3203
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4351797659092376e-05,
+      "loss": 0.0773,
+      "step": 3204
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434837506676509e-05,
+      "loss": 0.0034,
+      "step": 3205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434495184617547e-05,
+      "loss": 0.0006,
+      "step": 3206
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4341527997818116e-05,
+      "loss": 0.0176,
+      "step": 3207
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433810352218771e-05,
+      "loss": 0.0054,
+      "step": 3208
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4334678419779024e-05,
+      "loss": 0.0984,
+      "step": 3209
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4331252691086928e-05,
+      "loss": 0.008,
+      "step": 3210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4327826336606383e-05,
+      "loss": 0.1326,
+      "step": 3211
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324399356832426e-05,
+      "loss": 0.1546,
+      "step": 3212
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43209717522602e-05,
+      "loss": 0.0019,
+      "step": 3213
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4317543523384928e-05,
+      "loss": 0.0163,
+      "step": 3214
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314114670701937e-05,
+      "loss": 0.0465,
+      "step": 3215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310685194706626e-05,
+      "loss": 0.0256,
+      "step": 3216
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4307255095894497e-05,
+      "loss": 0.0074,
+      "step": 3217
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4303824374761134e-05,
+      "loss": 0.0134,
+      "step": 3218
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430039303180222e-05,
+      "loss": 0.0038,
+      "step": 3219
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4296961067513519e-05,
+      "loss": 0.0344,
+      "step": 3220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429352848239089e-05,
+      "loss": 0.0218,
+      "step": 3221
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290095276930278e-05,
+      "loss": 0.0542,
+      "step": 3222
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4286661451627722e-05,
+      "loss": 0.0034,
+      "step": 3223
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4283227006979348e-05,
+      "loss": 0.0869,
+      "step": 3224
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427979194348137e-05,
+      "loss": 0.0058,
+      "step": 3225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276356261630096e-05,
+      "loss": 0.0587,
+      "step": 3226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272919961921917e-05,
+      "loss": 0.0408,
+      "step": 3227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269483044853318e-05,
+      "loss": 0.0668,
+      "step": 3228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4266045510920873e-05,
+      "loss": 0.0537,
+      "step": 3229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262607360621245e-05,
+      "loss": 0.0165,
+      "step": 3230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4259168594451185e-05,
+      "loss": 0.0494,
+      "step": 3231
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4255729212907529e-05,
+      "loss": 0.0479,
+      "step": 3232
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4252289216487212e-05,
+      "loss": 0.0171,
+      "step": 3233
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424884860568725e-05,
+      "loss": 0.0289,
+      "step": 3234
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4245407381004744e-05,
+      "loss": 0.1027,
+      "step": 3235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4241965542936894e-05,
+      "loss": 0.0532,
+      "step": 3236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4238523091980987e-05,
+      "loss": 0.1335,
+      "step": 3237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235080028634392e-05,
+      "loss": 0.0079,
+      "step": 3238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423163635339457e-05,
+      "loss": 0.0139,
+      "step": 3239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4228192066759068e-05,
+      "loss": 0.0125,
+      "step": 3240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224747169225527e-05,
+      "loss": 0.0231,
+      "step": 3241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4221301661291675e-05,
+      "loss": 0.0806,
+      "step": 3242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4217855543455323e-05,
+      "loss": 0.0661,
+      "step": 3243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214408816214373e-05,
+      "loss": 0.0083,
+      "step": 3244
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421096148006682e-05,
+      "loss": 0.0518,
+      "step": 3245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4207513535510736e-05,
+      "loss": 0.0792,
+      "step": 3246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420406498304429e-05,
+      "loss": 0.006,
+      "step": 3247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4200615823165738e-05,
+      "loss": 0.0954,
+      "step": 3248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4197166056373414e-05,
+      "loss": 0.0041,
+      "step": 3249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419371568316576e-05,
+      "loss": 0.004,
+      "step": 3250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4190264704041284e-05,
+      "loss": 0.0582,
+      "step": 3251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186813119498595e-05,
+      "loss": 0.0047,
+      "step": 3252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418336093003638e-05,
+      "loss": 0.0131,
+      "step": 3253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179908136153424e-05,
+      "loss": 0.0041,
+      "step": 3254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4176454738348593e-05,
+      "loss": 0.0338,
+      "step": 3255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173000737120839e-05,
+      "loss": 0.0088,
+      "step": 3256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4169546132969204e-05,
+      "loss": 0.0753,
+      "step": 3257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166090926392815e-05,
+      "loss": 0.0656,
+      "step": 3258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4162635117890892e-05,
+      "loss": 0.0844,
+      "step": 3259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4159178707962737e-05,
+      "loss": 0.0333,
+      "step": 3260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415572169710773e-05,
+      "loss": 0.052,
+      "step": 3261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415226408582536e-05,
+      "loss": 0.0815,
+      "step": 3262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4148805874615182e-05,
+      "loss": 0.0015,
+      "step": 3263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414534706397685e-05,
+      "loss": 0.1171,
+      "step": 3264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141887654410096e-05,
+      "loss": 0.0414,
+      "step": 3265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4138427646414746e-05,
+      "loss": 0.0684,
+      "step": 3266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4134967040490707e-05,
+      "loss": 0.0038,
+      "step": 3267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131505837137973e-05,
+      "loss": 0.0092,
+      "step": 3268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4128044036856629e-05,
+      "loss": 0.0077,
+      "step": 3269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4124581640146844e-05,
+      "loss": 0.0584,
+      "step": 3270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121118647508872e-05,
+      "loss": 0.0409,
+      "step": 3271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4117655059443052e-05,
+      "loss": 0.0017,
+      "step": 3272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4114190876449809e-05,
+      "loss": 0.0551,
+      "step": 3273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4110726099029659e-05,
+      "loss": 0.0483,
+      "step": 3274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107260727683195e-05,
+      "loss": 0.0029,
+      "step": 3275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4103794762911106e-05,
+      "loss": 0.0598,
+      "step": 3276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100328205214161e-05,
+      "loss": 0.0008,
+      "step": 3277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096861055093215e-05,
+      "loss": 0.0423,
+      "step": 3278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4093393313049209e-05,
+      "loss": 0.002,
+      "step": 3279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4089924979583169e-05,
+      "loss": 0.0908,
+      "step": 3280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4086456055196208e-05,
+      "loss": 0.0292,
+      "step": 3281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4082986540389525e-05,
+      "loss": 0.0247,
+      "step": 3282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40795164356644e-05,
+      "loss": 0.0156,
+      "step": 3283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40760457415222e-05,
+      "loss": 0.0085,
+      "step": 3284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407257445846438e-05,
+      "loss": 0.0059,
+      "step": 3285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406910258699248e-05,
+      "loss": 0.0047,
+      "step": 3286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4065630127608122e-05,
+      "loss": 0.076,
+      "step": 3287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4062157080813013e-05,
+      "loss": 0.0348,
+      "step": 3288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4058683447108945e-05,
+      "loss": 0.0106,
+      "step": 3289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4055209226997801e-05,
+      "loss": 0.0098,
+      "step": 3290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051734420981537e-05,
+      "loss": 0.0349,
+      "step": 3291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4048259029562204e-05,
+      "loss": 0.0095,
+      "step": 3292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044783053241935e-05,
+      "loss": 0.0175,
+      "step": 3293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4041306492522945e-05,
+      "loss": 0.0031,
+      "step": 3294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4037829347907535e-05,
+      "loss": 0.0133,
+      "step": 3295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034351619898088e-05,
+      "loss": 0.0056,
+      "step": 3296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4030873308997073e-05,
+      "loss": 0.026,
+      "step": 3297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027394415707048e-05,
+      "loss": 0.0051,
+      "step": 3298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023914940530648e-05,
+      "loss": 0.0642,
+      "step": 3299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4020434883970596e-05,
+      "loss": 0.0167,
+      "step": 3300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016954246529697e-05,
+      "loss": 0.0043,
+      "step": 3301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401347302871084e-05,
+      "loss": 0.0461,
+      "step": 3302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4009991231017001e-05,
+      "loss": 0.0045,
+      "step": 3303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4006508853951234e-05,
+      "loss": 0.0026,
+      "step": 3304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003025898016682e-05,
+      "loss": 0.0948,
+      "step": 3305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999542363716571e-05,
+      "loss": 0.0236,
+      "step": 3306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3996058251554208e-05,
+      "loss": 0.0528,
+      "step": 3307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992573562032988e-05,
+      "loss": 0.0538,
+      "step": 3308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398908829565638e-05,
+      "loss": 0.0051,
+      "step": 3309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3985602452927944e-05,
+      "loss": 0.1129,
+      "step": 3310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982116034351325e-05,
+      "loss": 0.001,
+      "step": 3311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397862904043025e-05,
+      "loss": 0.0654,
+      "step": 3312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3975141471668521e-05,
+      "loss": 0.001,
+      "step": 3313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3971653328570033e-05,
+      "loss": 0.0402,
+      "step": 3314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3968164611638759e-05,
+      "loss": 0.0277,
+      "step": 3315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3964675321378756e-05,
+      "loss": 0.0021,
+      "step": 3316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3961185458294166e-05,
+      "loss": 0.0002,
+      "step": 3317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957695022889209e-05,
+      "loss": 0.0548,
+      "step": 3318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954204015668192e-05,
+      "loss": 0.0516,
+      "step": 3319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3950712437135501e-05,
+      "loss": 0.022,
+      "step": 3320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3947220287795611e-05,
+      "loss": 0.0301,
+      "step": 3321
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394372756815307e-05,
+      "loss": 0.0182,
+      "step": 3322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940234278712517e-05,
+      "loss": 0.0121,
+      "step": 3323
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936740419978665e-05,
+      "loss": 0.0036,
+      "step": 3324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3933245992456323e-05,
+      "loss": 0.0484,
+      "step": 3325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3929750996650363e-05,
+      "loss": 0.0994,
+      "step": 3326
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3926255433065756e-05,
+      "loss": 0.0413,
+      "step": 3327
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3922759302207548e-05,
+      "loss": 0.0076,
+      "step": 3328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919262604580864e-05,
+      "loss": 0.0346,
+      "step": 3329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3915765340690916e-05,
+      "loss": 0.0121,
+      "step": 3330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3912267511042994e-05,
+      "loss": 0.039,
+      "step": 3331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3908769116142476e-05,
+      "loss": 0.0126,
+      "step": 3332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905270156494812e-05,
+      "loss": 0.0283,
+      "step": 3333
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3901770632605546e-05,
+      "loss": 0.0835,
+      "step": 3334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3898270544980292e-05,
+      "loss": 0.012,
+      "step": 3335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894769894124748e-05,
+      "loss": 0.0937,
+      "step": 3336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891268680544699e-05,
+      "loss": 0.0077,
+      "step": 3337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887766904746008e-05,
+      "loss": 0.0612,
+      "step": 3338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884264567234616e-05,
+      "loss": 0.0018,
+      "step": 3339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388076166851655e-05,
+      "loss": 0.0373,
+      "step": 3340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3877258209097913e-05,
+      "loss": 0.0117,
+      "step": 3341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873754189484897e-05,
+      "loss": 0.0108,
+      "step": 3342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387024961018377e-05,
+      "loss": 0.0114,
+      "step": 3343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3866744471700877e-05,
+      "loss": 0.0533,
+      "step": 3344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3863238774542648e-05,
+      "loss": 0.0259,
+      "step": 3345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3859732519215596e-05,
+      "loss": 0.0254,
+      "step": 3346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3856225706226311e-05,
+      "loss": 0.0646,
+      "step": 3347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3852718336081465e-05,
+      "loss": 0.0528,
+      "step": 3348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3849210409287813e-05,
+      "loss": 0.0018,
+      "step": 3349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3845701926352182e-05,
+      "loss": 0.0829,
+      "step": 3350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3842192887781487e-05,
+      "loss": 0.0016,
+      "step": 3351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838683294082725e-05,
+      "loss": 0.0423,
+      "step": 3352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835173145762961e-05,
+      "loss": 0.0614,
+      "step": 3353
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831662443329355e-05,
+      "loss": 0.0778,
+      "step": 3354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3828151187289142e-05,
+      "loss": 0.1112,
+      "step": 3355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824639378149634e-05,
+      "loss": 0.0014,
+      "step": 3356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382112701641822e-05,
+      "loss": 0.0509,
+      "step": 3357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.381761410260238e-05,
+      "loss": 0.0014,
+      "step": 3358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3814100637209663e-05,
+      "loss": 0.0004,
+      "step": 3359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810586620747702e-05,
+      "loss": 0.0019,
+      "step": 3360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380707205372421e-05,
+      "loss": 0.0062,
+      "step": 3361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380355693664698e-05,
+      "loss": 0.0284,
+      "step": 3362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3800041270023885e-05,
+      "loss": 0.0355,
+      "step": 3363
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3796525054362872e-05,
+      "loss": 0.0048,
+      "step": 3364
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793008290171975e-05,
+      "loss": 0.0786,
+      "step": 3365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3789490977959297e-05,
+      "loss": 0.1296,
+      "step": 3366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3785973118233034e-05,
+      "loss": 0.055,
+      "step": 3367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378245471150145e-05,
+      "loss": 0.0223,
+      "step": 3368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3778935758272893e-05,
+      "loss": 0.0276,
+      "step": 3369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377541625905579e-05,
+      "loss": 0.0028,
+      "step": 3370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771896214358642e-05,
+      "loss": 0.0017,
+      "step": 3371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3768375624690033e-05,
+      "loss": 0.0412,
+      "step": 3372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376485449055863e-05,
+      "loss": 0.0083,
+      "step": 3373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761332812473167e-05,
+      "loss": 0.0471,
+      "step": 3374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375781059094247e-05,
+      "loss": 0.1583,
+      "step": 3375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3754287826475428e-05,
+      "loss": 0.0287,
+      "step": 3376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375076451958103e-05,
+      "loss": 0.0494,
+      "step": 3377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374724067076832e-05,
+      "loss": 0.003,
+      "step": 3378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3743716280546435e-05,
+      "loss": 0.1332,
+      "step": 3379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740191349424587e-05,
+      "loss": 0.0348,
+      "step": 3380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736665877912063e-05,
+      "loss": 0.0298,
+      "step": 3381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3733139866518235e-05,
+      "loss": 0.0027,
+      "step": 3382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729613315752541e-05,
+      "loss": 0.0253,
+      "step": 3383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726086226124516e-05,
+      "loss": 0.0513,
+      "step": 3384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3722558598143747e-05,
+      "loss": 0.0065,
+      "step": 3385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719030432319925e-05,
+      "loss": 0.1265,
+      "step": 3386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37155017291628e-05,
+      "loss": 0.0025,
+      "step": 3387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3711972489182208e-05,
+      "loss": 0.1568,
+      "step": 3388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708442712888062e-05,
+      "loss": 0.0674,
+      "step": 3389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704912400790354e-05,
+      "loss": 0.0248,
+      "step": 3390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701381553399147e-05,
+      "loss": 0.0644,
+      "step": 3391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3697850171224584e-05,
+      "loss": 0.0046,
+      "step": 3392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3694318254776888e-05,
+      "loss": 0.0282,
+      "step": 3393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369078580456636e-05,
+      "loss": 0.0011,
+      "step": 3394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3687252821103374e-05,
+      "loss": 0.0046,
+      "step": 3395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3683719304898382e-05,
+      "loss": 0.0261,
+      "step": 3396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3680185256461917e-05,
+      "loss": 0.0043,
+      "step": 3397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676650676304582e-05,
+      "loss": 0.0354,
+      "step": 3398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673115564937061e-05,
+      "loss": 0.113,
+      "step": 3399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3669579922870116e-05,
+      "loss": 0.0728,
+      "step": 3400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3666043750614584e-05,
+      "loss": 0.004,
+      "step": 3401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3662507048681376e-05,
+      "loss": 0.0048,
+      "step": 3402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3658969817581484e-05,
+      "loss": 0.1258,
+      "step": 3403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655432057825978e-05,
+      "loss": 0.0046,
+      "step": 3404
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651893769925994e-05,
+      "loss": 0.1048,
+      "step": 3405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3648354954392758e-05,
+      "loss": 0.0042,
+      "step": 3406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3644815611737558e-05,
+      "loss": 0.0899,
+      "step": 3407
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3641275742471774e-05,
+      "loss": 0.0086,
+      "step": 3408
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3637735347106847e-05,
+      "loss": 0.0302,
+      "step": 3409
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634194426154301e-05,
+      "loss": 0.0379,
+      "step": 3410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363065298012574e-05,
+      "loss": 0.0116,
+      "step": 3411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627111009532838e-05,
+      "loss": 0.0049,
+      "step": 3412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362356851488735e-05,
+      "loss": 0.0489,
+      "step": 3413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3620025496701091e-05,
+      "loss": 0.0014,
+      "step": 3414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616481955485974e-05,
+      "loss": 0.0193,
+      "step": 3415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3612937891753974e-05,
+      "loss": 0.0017,
+      "step": 3416
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609393306017149e-05,
+      "loss": 0.0028,
+      "step": 3417
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3605848198787624e-05,
+      "loss": 0.0987,
+      "step": 3418
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602302570577601e-05,
+      "loss": 0.0106,
+      "step": 3419
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598756421899365e-05,
+      "loss": 0.0053,
+      "step": 3420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3595209753265266e-05,
+      "loss": 0.0916,
+      "step": 3421
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3591662565187742e-05,
+      "loss": 0.0095,
+      "step": 3422
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588114858179289e-05,
+      "loss": 0.0355,
+      "step": 3423
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3584566632752494e-05,
+      "loss": 0.0188,
+      "step": 3424
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358101788942001e-05,
+      "loss": 0.0278,
+      "step": 3425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3577468628694566e-05,
+      "loss": 0.0112,
+      "step": 3426
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3573918851088963e-05,
+      "loss": 0.0064,
+      "step": 3427
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3570368557116087e-05,
+      "loss": 0.0063,
+      "step": 3428
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566817747288887e-05,
+      "loss": 0.0357,
+      "step": 3429
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3563266422120399e-05,
+      "loss": 0.0076,
+      "step": 3430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3559714582123713e-05,
+      "loss": 0.0042,
+      "step": 3431
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355616222781202e-05,
+      "loss": 0.1723,
+      "step": 3432
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552609359698561e-05,
+      "loss": 0.101,
+      "step": 3433
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3549055978296666e-05,
+      "loss": 0.054,
+      "step": 3434
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3545502084119738e-05,
+      "loss": 0.0071,
+      "step": 3435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541947677681246e-05,
+      "loss": 0.0189,
+      "step": 3436
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3538392759494739e-05,
+      "loss": 0.0011,
+      "step": 3437
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3534837330073843e-05,
+      "loss": 0.038,
+      "step": 3438
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353128138993225e-05,
+      "loss": 0.0009,
+      "step": 3439
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3527724939583733e-05,
+      "loss": 0.0752,
+      "step": 3440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3524167979542129e-05,
+      "loss": 0.0039,
+      "step": 3441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520610510321358e-05,
+      "loss": 0.0307,
+      "step": 3442
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517052532435417e-05,
+      "loss": 0.0875,
+      "step": 3443
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3513494046398363e-05,
+      "loss": 0.0194,
+      "step": 3444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3509935052724335e-05,
+      "loss": 0.038,
+      "step": 3445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506375551927546e-05,
+      "loss": 0.0746,
+      "step": 3446
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3502815544522276e-05,
+      "loss": 0.093,
+      "step": 3447
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499255031022887e-05,
+      "loss": 0.0871,
+      "step": 3448
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3495694011943804e-05,
+      "loss": 0.0543,
+      "step": 3449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492132487799534e-05,
+      "loss": 0.0634,
+      "step": 3450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3488570459104656e-05,
+      "loss": 0.0965,
+      "step": 3451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485007926373815e-05,
+      "loss": 0.0466,
+      "step": 3452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3481444890121735e-05,
+      "loss": 0.1364,
+      "step": 3453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3477881350863208e-05,
+      "loss": 0.0181,
+      "step": 3454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3474317309113106e-05,
+      "loss": 0.0855,
+      "step": 3455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3470752765386368e-05,
+      "loss": 0.0107,
+      "step": 3456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467187720198005e-05,
+      "loss": 0.0967,
+      "step": 3457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463622174063105e-05,
+      "loss": 0.0644,
+      "step": 3458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346005612749682e-05,
+      "loss": 0.031,
+      "step": 3459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3456489581014389e-05,
+      "loss": 0.0483,
+      "step": 3460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3452922535131108e-05,
+      "loss": 0.0022,
+      "step": 3461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344935499036235e-05,
+      "loss": 0.0238,
+      "step": 3462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3445786947223563e-05,
+      "loss": 0.0606,
+      "step": 3463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442218406230265e-05,
+      "loss": 0.1018,
+      "step": 3464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438649367898048e-05,
+      "loss": 0.0349,
+      "step": 3465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435079832742577e-05,
+      "loss": 0.0448,
+      "step": 3466
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3431509801279578e-05,
+      "loss": 0.0075,
+      "step": 3467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427939274024861e-05,
+      "loss": 0.0691,
+      "step": 3468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3424368251494305e-05,
+      "loss": 0.0376,
+      "step": 3469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3420796734203855e-05,
+      "loss": 0.0397,
+      "step": 3470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417224722669534e-05,
+      "loss": 0.019,
+      "step": 3471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3413652217407432e-05,
+      "loss": 0.0205,
+      "step": 3472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3410079218933714e-05,
+      "loss": 0.0493,
+      "step": 3473
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3406505727764611e-05,
+      "loss": 0.0066,
+      "step": 3474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3402931744416432e-05,
+      "loss": 0.0331,
+      "step": 3475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339935726940555e-05,
+      "loss": 0.0811,
+      "step": 3476
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3395782303248419e-05,
+      "loss": 0.0047,
+      "step": 3477
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339220684646155e-05,
+      "loss": 0.0225,
+      "step": 3478
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388630899561536e-05,
+      "loss": 0.0379,
+      "step": 3479
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338505446306504e-05,
+      "loss": 0.0067,
+      "step": 3480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338147753748879e-05,
+      "loss": 0.0258,
+      "step": 3481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377900123349586e-05,
+      "loss": 0.0042,
+      "step": 3482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374322221164302e-05,
+      "loss": 0.0031,
+      "step": 3483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370743831449882e-05,
+      "loss": 0.018,
+      "step": 3484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3367164954723337e-05,
+      "loss": 0.041,
+      "step": 3485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3363585591501751e-05,
+      "loss": 0.0152,
+      "step": 3486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3360005742302283e-05,
+      "loss": 0.0134,
+      "step": 3487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3356425407642151e-05,
+      "loss": 0.0202,
+      "step": 3488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335284458803865e-05,
+      "loss": 0.0449,
+      "step": 3489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349263284009144e-05,
+      "loss": 0.0198,
+      "step": 3490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3345681496071068e-05,
+      "loss": 0.0107,
+      "step": 3491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342099224741932e-05,
+      "loss": 0.0326,
+      "step": 3492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3338516470539303e-05,
+      "loss": 0.0034,
+      "step": 3493
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3334933233980824e-05,
+      "loss": 0.0782,
+      "step": 3494
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3331349515584213e-05,
+      "loss": 0.0028,
+      "step": 3495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3327765315867253e-05,
+      "loss": 0.0152,
+      "step": 3496
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324180635347795e-05,
+      "loss": 0.0048,
+      "step": 3497
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320595474543758e-05,
+      "loss": 0.0127,
+      "step": 3498
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3317009833973138e-05,
+      "loss": 0.0083,
+      "step": 3499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3313423714153996e-05,
+      "loss": 0.0032,
+      "step": 3500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330983711560446e-05,
+      "loss": 0.0735,
+      "step": 3501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330625003884273e-05,
+      "loss": 0.0015,
+      "step": 3502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3302662484387075e-05,
+      "loss": 0.0047,
+      "step": 3503
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3299074452755829e-05,
+      "loss": 0.0284,
+      "step": 3504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3295485944467405e-05,
+      "loss": 0.0046,
+      "step": 3505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291896960040271e-05,
+      "loss": 0.0055,
+      "step": 3506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288307499992976e-05,
+      "loss": 0.0698,
+      "step": 3507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3284717564844133e-05,
+      "loss": 0.0022,
+      "step": 3508
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281127155112417e-05,
+      "loss": 0.0177,
+      "step": 3509
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3277536271316587e-05,
+      "loss": 0.0015,
+      "step": 3510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3273944913975452e-05,
+      "loss": 0.0125,
+      "step": 3511
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270353083607904e-05,
+      "loss": 0.0112,
+      "step": 3512
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3266760780732897e-05,
+      "loss": 0.0266,
+      "step": 3513
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3263168005869458e-05,
+      "loss": 0.0444,
+      "step": 3514
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259574759536675e-05,
+      "loss": 0.0018,
+      "step": 3515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325598104225371e-05,
+      "loss": 0.065,
+      "step": 3516
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3252386854539786e-05,
+      "loss": 0.0107,
+      "step": 3517
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248792196914204e-05,
+      "loss": 0.0148,
+      "step": 3518
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3245197069896324e-05,
+      "loss": 0.0103,
+      "step": 3519
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3241601474005576e-05,
+      "loss": 0.0451,
+      "step": 3520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3238005409761468e-05,
+      "loss": 0.0187,
+      "step": 3521
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323440887768356e-05,
+      "loss": 0.0419,
+      "step": 3522
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3230811878291487e-05,
+      "loss": 0.0181,
+      "step": 3523
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3227214412104952e-05,
+      "loss": 0.0052,
+      "step": 3524
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3223616479643721e-05,
+      "loss": 0.0201,
+      "step": 3525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220018081427637e-05,
+      "loss": 0.0042,
+      "step": 3526
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3216419217976602e-05,
+      "loss": 0.0998,
+      "step": 3527
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3212819889810585e-05,
+      "loss": 0.0663,
+      "step": 3528
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209220097449626e-05,
+      "loss": 0.0048,
+      "step": 3529
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320561984141383e-05,
+      "loss": 0.044,
+      "step": 3530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202019122223373e-05,
+      "loss": 0.0097,
+      "step": 3531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3198417940398487e-05,
+      "loss": 0.0162,
+      "step": 3532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194816296459483e-05,
+      "loss": 0.045,
+      "step": 3533
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3191214190926734e-05,
+      "loss": 0.0211,
+      "step": 3534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3187611624320681e-05,
+      "loss": 0.0464,
+      "step": 3535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318400859716183e-05,
+      "loss": 0.0066,
+      "step": 3536
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3180405109970749e-05,
+      "loss": 0.0044,
+      "step": 3537
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3176801163268082e-05,
+      "loss": 0.0012,
+      "step": 3538
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173196757574535e-05,
+      "loss": 0.0658,
+      "step": 3539
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316959189341088e-05,
+      "loss": 0.0051,
+      "step": 3540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165986571297953e-05,
+      "loss": 0.0351,
+      "step": 3541
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316238079175666e-05,
+      "loss": 0.1286,
+      "step": 3542
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3158774555307972e-05,
+      "loss": 0.0605,
+      "step": 3543
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3155167862472922e-05,
+      "loss": 0.0038,
+      "step": 3544
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151560713772618e-05,
+      "loss": 0.0111,
+      "step": 3545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3147953109728224e-05,
+      "loss": 0.057,
+      "step": 3546
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144345050860976e-05,
+      "loss": 0.0658,
+      "step": 3547
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140736537692176e-05,
+      "loss": 0.1216,
+      "step": 3548
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3137127570743189e-05,
+      "loss": 0.2063,
+      "step": 3549
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313351815053544e-05,
+      "loss": 0.0481,
+      "step": 3550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3129908277590433e-05,
+      "loss": 0.092,
+      "step": 3551
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3126297952429723e-05,
+      "loss": 0.0247,
+      "step": 3552
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3122687175574946e-05,
+      "loss": 0.0155,
+      "step": 3553
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3119075947547788e-05,
+      "loss": 0.0043,
+      "step": 3554
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311546426887001e-05,
+      "loss": 0.0331,
+      "step": 3555
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111852140063435e-05,
+      "loss": 0.0029,
+      "step": 3556
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3108239561649945e-05,
+      "loss": 0.0033,
+      "step": 3557
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104626534151503e-05,
+      "loss": 0.1087,
+      "step": 3558
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3101013058090117e-05,
+      "loss": 0.0383,
+      "step": 3559
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3097399133987876e-05,
+      "loss": 0.0929,
+      "step": 3560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093784762366923e-05,
+      "loss": 0.003,
+      "step": 3561
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.0031,
+      "step": 3562
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3086554678657809e-05,
+      "loss": 0.0262,
+      "step": 3563
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3082938967614256e-05,
+      "loss": 0.099,
+      "step": 3564
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307932281114123e-05,
+      "loss": 0.0297,
+      "step": 3565
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3075706209761201e-05,
+      "loss": 0.0445,
+      "step": 3566
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.30720891639967e-05,
+      "loss": 0.09,
+      "step": 3567
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3068471674370327e-05,
+      "loss": 0.0398,
+      "step": 3568
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3064853741404741e-05,
+      "loss": 0.1867,
+      "step": 3569
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3061235365622678e-05,
+      "loss": 0.0047,
+      "step": 3570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305761654754692e-05,
+      "loss": 0.0229,
+      "step": 3571
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3053997287700322e-05,
+      "loss": 0.0031,
+      "step": 3572
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3050377586605803e-05,
+      "loss": 0.0298,
+      "step": 3573
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046757444786346e-05,
+      "loss": 0.0467,
+      "step": 3574
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3043136862765e-05,
+      "loss": 0.0603,
+      "step": 3575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3039515841064867e-05,
+      "loss": 0.0062,
+      "step": 3576
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3035894380209126e-05,
+      "loss": 0.0088,
+      "step": 3577
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032272480721009e-05,
+      "loss": 0.0025,
+      "step": 3578
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028650143123817e-05,
+      "loss": 0.0087,
+      "step": 3579
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3025027367940916e-05,
+      "loss": 0.0628,
+      "step": 3580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021404155695728e-05,
+      "loss": 0.0167,
+      "step": 3581
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3017780506911744e-05,
+      "loss": 0.0384,
+      "step": 3582
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3014156422112514e-05,
+      "loss": 0.0021,
+      "step": 3583
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301053190182166e-05,
+      "loss": 0.054,
+      "step": 3584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3006906946562855e-05,
+      "loss": 0.025,
+      "step": 3585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3003281556859837e-05,
+      "loss": 0.0621,
+      "step": 3586
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999655733236414e-05,
+      "loss": 0.0029,
+      "step": 3587
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2996029476216455e-05,
+      "loss": 0.0152,
+      "step": 3588
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992402786323885e-05,
+      "loss": 0.0271,
+      "step": 3589
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988775664082697e-05,
+      "loss": 0.0158,
+      "step": 3590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2985148110016947e-05,
+      "loss": 0.0372,
+      "step": 3591
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2981520124650748e-05,
+      "loss": 0.0443,
+      "step": 3592
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977891708508285e-05,
+      "loss": 0.0399,
+      "step": 3593
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974262862113791e-05,
+      "loss": 0.0079,
+      "step": 3594
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2970633585991575e-05,
+      "loss": 0.0565,
+      "step": 3595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2967003880666001e-05,
+      "loss": 0.0622,
+      "step": 3596
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2963373746661496e-05,
+      "loss": 0.1619,
+      "step": 3597
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2959743184502553e-05,
+      "loss": 0.0041,
+      "step": 3598
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2956112194713715e-05,
+      "loss": 0.0046,
+      "step": 3599
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29524807778196e-05,
+      "loss": 0.0161,
+      "step": 3600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2948848934344885e-05,
+      "loss": 0.0519,
+      "step": 3601
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2945216664814307e-05,
+      "loss": 0.0067,
+      "step": 3602
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941583969752656e-05,
+      "loss": 0.0565,
+      "step": 3603
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2937950849684794e-05,
+      "loss": 0.0597,
+      "step": 3604
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934317305135651e-05,
+      "loss": 0.0093,
+      "step": 3605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2930683336630199e-05,
+      "loss": 0.0394,
+      "step": 3606
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927048944693486e-05,
+      "loss": 0.0085,
+      "step": 3607
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292341412985061e-05,
+      "loss": 0.0368,
+      "step": 3608
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919778892626746e-05,
+      "loss": 0.0141,
+      "step": 3609
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2916143233547117e-05,
+      "loss": 0.0257,
+      "step": 3610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2912507153137009e-05,
+      "loss": 0.0487,
+      "step": 3611
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2908870651921771e-05,
+      "loss": 0.0409,
+      "step": 3612
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2905233730426814e-05,
+      "loss": 0.016,
+      "step": 3613
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2901596389177606e-05,
+      "loss": 0.0043,
+      "step": 3614
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897958628699679e-05,
+      "loss": 0.0351,
+      "step": 3615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2894320449518624e-05,
+      "loss": 0.1148,
+      "step": 3616
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2890681852160092e-05,
+      "loss": 0.0686,
+      "step": 3617
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2887042837149798e-05,
+      "loss": 0.007,
+      "step": 3618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288340340501351e-05,
+      "loss": 0.0207,
+      "step": 3619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879763556277062e-05,
+      "loss": 0.0324,
+      "step": 3620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2876123291466351e-05,
+      "loss": 0.0503,
+      "step": 3621
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2872482611107324e-05,
+      "loss": 0.0527,
+      "step": 3622
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868841515726e-05,
+      "loss": 0.0494,
+      "step": 3623
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2865200005848446e-05,
+      "loss": 0.0712,
+      "step": 3624
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2861558082000803e-05,
+      "loss": 0.1518,
+      "step": 3625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2857915744709255e-05,
+      "loss": 0.0064,
+      "step": 3626
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2854272994500059e-05,
+      "loss": 0.0119,
+      "step": 3627
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2850629831899531e-05,
+      "loss": 0.0194,
+      "step": 3628
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846986257434034e-05,
+      "loss": 0.0074,
+      "step": 3629
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2843342271630005e-05,
+      "loss": 0.0185,
+      "step": 3630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283969787501393e-05,
+      "loss": 0.0182,
+      "step": 3631
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2836053068112366e-05,
+      "loss": 0.0876,
+      "step": 3632
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2832407851451922e-05,
+      "loss": 0.0094,
+      "step": 3633
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828762225559259e-05,
+      "loss": 0.0552,
+      "step": 3634
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825116190961109e-05,
+      "loss": 0.0155,
+      "step": 3635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282146974818426e-05,
+      "loss": 0.0022,
+      "step": 3636
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817822897755557e-05,
+      "loss": 0.0053,
+      "step": 3637
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28141756402019e-05,
+      "loss": 0.0198,
+      "step": 3638
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2810527976050262e-05,
+      "loss": 0.0044,
+      "step": 3639
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2806879905827656e-05,
+      "loss": 0.0639,
+      "step": 3640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280323143006117e-05,
+      "loss": 0.1488,
+      "step": 3641
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799582549277936e-05,
+      "loss": 0.0055,
+      "step": 3642
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795933264005156e-05,
+      "loss": 0.1428,
+      "step": 3643
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2792283574770086e-05,
+      "loss": 0.0078,
+      "step": 3644
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788633482100044e-05,
+      "loss": 0.0027,
+      "step": 3645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2784982986522399e-05,
+      "loss": 0.0042,
+      "step": 3646
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781332088564579e-05,
+      "loss": 0.0482,
+      "step": 3647
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277768078875408e-05,
+      "loss": 0.1576,
+      "step": 3648
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774029087618448e-05,
+      "loss": 0.0173,
+      "step": 3649
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2770376985685287e-05,
+      "loss": 0.1372,
+      "step": 3650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2766724483482257e-05,
+      "loss": 0.0763,
+      "step": 3651
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763071581537087e-05,
+      "loss": 0.0711,
+      "step": 3652
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2759418280377551e-05,
+      "loss": 0.119,
+      "step": 3653
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2755764580531488e-05,
+      "loss": 0.0138,
+      "step": 3654
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275211048252679e-05,
+      "loss": 0.0473,
+      "step": 3655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748455986891406e-05,
+      "loss": 0.0018,
+      "step": 3656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274480109415335e-05,
+      "loss": 0.0543,
+      "step": 3657
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2741145804840688e-05,
+      "loss": 0.0124,
+      "step": 3658
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2737490119481541e-05,
+      "loss": 0.0638,
+      "step": 3659
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2733834038604095e-05,
+      "loss": 0.137,
+      "step": 3660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730177562736583e-05,
+      "loss": 0.008,
+      "step": 3661
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27265206924073e-05,
+      "loss": 0.0067,
+      "step": 3662
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2722863428144608e-05,
+      "loss": 0.0249,
+      "step": 3663
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27192057704769e-05,
+      "loss": 0.0617,
+      "step": 3664
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715547719932656e-05,
+      "loss": 0.006,
+      "step": 3665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711889277040392e-05,
+      "loss": 0.0749,
+      "step": 3666
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2708230442328693e-05,
+      "loss": 0.041,
+      "step": 3667
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270457121632619e-05,
+      "loss": 0.0058,
+      "step": 3668
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2700911599561574e-05,
+      "loss": 0.023,
+      "step": 3669
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697251592563598e-05,
+      "loss": 0.0208,
+      "step": 3670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2693591195861065e-05,
+      "loss": 0.0259,
+      "step": 3671
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689930409982843e-05,
+      "loss": 0.0805,
+      "step": 3672
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2686269235457842e-05,
+      "loss": 0.0779,
+      "step": 3673
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2682607672815042e-05,
+      "loss": 0.0052,
+      "step": 3674
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2678945722583466e-05,
+      "loss": 0.0021,
+      "step": 3675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2675283385292212e-05,
+      "loss": 0.0124,
+      "step": 3676
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267162066147041e-05,
+      "loss": 0.1037,
+      "step": 3677
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667957551647263e-05,
+      "loss": 0.0302,
+      "step": 3678
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2664294056352025e-05,
+      "loss": 0.0321,
+      "step": 3679
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2660630176114009e-05,
+      "loss": 0.0046,
+      "step": 3680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2656965911462573e-05,
+      "loss": 0.0056,
+      "step": 3681
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265330126292714e-05,
+      "loss": 0.0272,
+      "step": 3682
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2649636231037188e-05,
+      "loss": 0.0164,
+      "step": 3683
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645970816322248e-05,
+      "loss": 0.0123,
+      "step": 3684
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2642305019311909e-05,
+      "loss": 0.0025,
+      "step": 3685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2638638840535808e-05,
+      "loss": 0.0387,
+      "step": 3686
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2634972280523643e-05,
+      "loss": 0.0056,
+      "step": 3687
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263130533980517e-05,
+      "loss": 0.0843,
+      "step": 3688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627638018910195e-05,
+      "loss": 0.0408,
+      "step": 3689
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262397031836858e-05,
+      "loss": 0.0096,
+      "step": 3690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620302238710238e-05,
+      "loss": 0.1274,
+      "step": 3691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2616633780465145e-05,
+      "loss": 0.0736,
+      "step": 3692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261296494416333e-05,
+      "loss": 0.0283,
+      "step": 3693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260929573033487e-05,
+      "loss": 0.0329,
+      "step": 3694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605626139509899e-05,
+      "loss": 0.0704,
+      "step": 3695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260195617221861e-05,
+      "loss": 0.0543,
+      "step": 3696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2598285828991249e-05,
+      "loss": 0.0316,
+      "step": 3697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259461511035811e-05,
+      "loss": 0.0479,
+      "step": 3698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2590944016849554e-05,
+      "loss": 0.0737,
+      "step": 3699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2587272548995979e-05,
+      "loss": 0.0486,
+      "step": 3700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2583600707327854e-05,
+      "loss": 0.0076,
+      "step": 3701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257992849237569e-05,
+      "loss": 0.0094,
+      "step": 3702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576255904670058e-05,
+      "loss": 0.0075,
+      "step": 3703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257258294474158e-05,
+      "loss": 0.0171,
+      "step": 3704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2568909613120934e-05,
+      "loss": 0.0016,
+      "step": 3705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565235910338852e-05,
+      "loss": 0.0039,
+      "step": 3706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561561836926115e-05,
+      "loss": 0.0281,
+      "step": 3707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2557887393413566e-05,
+      "loss": 0.0189,
+      "step": 3708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2554212580332088e-05,
+      "loss": 0.0494,
+      "step": 3709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255053739821263e-05,
+      "loss": 0.0269,
+      "step": 3710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546861847586194e-05,
+      "loss": 0.0828,
+      "step": 3711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2543185928983825e-05,
+      "loss": 0.0832,
+      "step": 3712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2539509642936629e-05,
+      "loss": 0.0046,
+      "step": 3713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2535832989975766e-05,
+      "loss": 0.0683,
+      "step": 3714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2532155970632448e-05,
+      "loss": 0.0026,
+      "step": 3715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528478585437935e-05,
+      "loss": 0.0298,
+      "step": 3716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2524800834923541e-05,
+      "loss": 0.0558,
+      "step": 3717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521122719620638e-05,
+      "loss": 0.0594,
+      "step": 3718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2517444240060647e-05,
+      "loss": 0.0081,
+      "step": 3719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513765396775047e-05,
+      "loss": 0.0475,
+      "step": 3720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510086190295358e-05,
+      "loss": 0.1023,
+      "step": 3721
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2506406621153164e-05,
+      "loss": 0.0386,
+      "step": 3722
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502726689880093e-05,
+      "loss": 0.0302,
+      "step": 3723
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499046397007835e-05,
+      "loss": 0.0328,
+      "step": 3724
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2495365743068122e-05,
+      "loss": 0.0089,
+      "step": 3725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2491684728592744e-05,
+      "loss": 0.0346,
+      "step": 3726
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2488003354113538e-05,
+      "loss": 0.0021,
+      "step": 3727
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2484321620162405e-05,
+      "loss": 0.0509,
+      "step": 3728
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480639527271283e-05,
+      "loss": 0.1205,
+      "step": 3729
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476957075972167e-05,
+      "loss": 0.0406,
+      "step": 3730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2473274266797111e-05,
+      "loss": 0.0014,
+      "step": 3731
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2469591100278212e-05,
+      "loss": 0.0294,
+      "step": 3732
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465907576947623e-05,
+      "loss": 0.0042,
+      "step": 3733
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2462223697337543e-05,
+      "loss": 0.0159,
+      "step": 3734
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245853946198023e-05,
+      "loss": 0.0079,
+      "step": 3735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2454854871407993e-05,
+      "loss": 0.0115,
+      "step": 3736
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2451169926153184e-05,
+      "loss": 0.033,
+      "step": 3737
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2447484626748216e-05,
+      "loss": 0.0542,
+      "step": 3738
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2443798973725543e-05,
+      "loss": 0.025,
+      "step": 3739
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440112967617682e-05,
+      "loss": 0.078,
+      "step": 3740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2436426608957191e-05,
+      "loss": 0.0673,
+      "step": 3741
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432739898276684e-05,
+      "loss": 0.0082,
+      "step": 3742
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429052836108826e-05,
+      "loss": 0.0202,
+      "step": 3743
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2425365422986328e-05,
+      "loss": 0.0335,
+      "step": 3744
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242167765944196e-05,
+      "loss": 0.0079,
+      "step": 3745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417989546008533e-05,
+      "loss": 0.0113,
+      "step": 3746
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414301083218919e-05,
+      "loss": 0.0715,
+      "step": 3747
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410612271606026e-05,
+      "loss": 0.0087,
+      "step": 3748
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2406923111702828e-05,
+      "loss": 0.0139,
+      "step": 3749
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2403233604042344e-05,
+      "loss": 0.0096,
+      "step": 3750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239954374915764e-05,
+      "loss": 0.0551,
+      "step": 3751
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395853547581829e-05,
+      "loss": 0.0543,
+      "step": 3752
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392162999848083e-05,
+      "loss": 0.0538,
+      "step": 3753
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238847210648962e-05,
+      "loss": 0.008,
+      "step": 3754
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2384780868039709e-05,
+      "loss": 0.0901,
+      "step": 3755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2381089285031666e-05,
+      "loss": 0.0157,
+      "step": 3756
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2377397357998861e-05,
+      "loss": 0.0505,
+      "step": 3757
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2373705087474707e-05,
+      "loss": 0.0184,
+      "step": 3758
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2370012473992676e-05,
+      "loss": 0.0294,
+      "step": 3759
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2366319518086285e-05,
+      "loss": 0.0851,
+      "step": 3760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2362626220289094e-05,
+      "loss": 0.0266,
+      "step": 3761
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235893258113472e-05,
+      "loss": 0.0039,
+      "step": 3762
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2355238601156834e-05,
+      "loss": 0.0795,
+      "step": 3763
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2351544280889143e-05,
+      "loss": 0.0471,
+      "step": 3764
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234784962086541e-05,
+      "loss": 0.0027,
+      "step": 3765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234415462161945e-05,
+      "loss": 0.005,
+      "step": 3766
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2340459283685126e-05,
+      "loss": 0.0372,
+      "step": 3767
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2336763607596348e-05,
+      "loss": 0.0338,
+      "step": 3768
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233306759388707e-05,
+      "loss": 0.0025,
+      "step": 3769
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329371243091303e-05,
+      "loss": 0.0532,
+      "step": 3770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2325674555743106e-05,
+      "loss": 0.0051,
+      "step": 3771
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232197753237658e-05,
+      "loss": 0.0484,
+      "step": 3772
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231828017352588e-05,
+      "loss": 0.0003,
+      "step": 3773
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2314582479725207e-05,
+      "loss": 0.0065,
+      "step": 3774
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2310884451508812e-05,
+      "loss": 0.0068,
+      "step": 3775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307186089410999e-05,
+      "loss": 0.0064,
+      "step": 3776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2303487393966111e-05,
+      "loss": 0.0468,
+      "step": 3777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2299788365708541e-05,
+      "loss": 0.0152,
+      "step": 3778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2296089005172733e-05,
+      "loss": 0.0076,
+      "step": 3779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229238931289318e-05,
+      "loss": 0.0808,
+      "step": 3780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2288689289404424e-05,
+      "loss": 0.0455,
+      "step": 3781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2284988935241046e-05,
+      "loss": 0.0107,
+      "step": 3782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2281288250937684e-05,
+      "loss": 0.0259,
+      "step": 3783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2277587237029025e-05,
+      "loss": 0.0652,
+      "step": 3784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2273885894049791e-05,
+      "loss": 0.0927,
+      "step": 3785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2270184222534767e-05,
+      "loss": 0.0177,
+      "step": 3786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266482223018772e-05,
+      "loss": 0.0036,
+      "step": 3787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262779896036681e-05,
+      "loss": 0.0431,
+      "step": 3788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2259077242123416e-05,
+      "loss": 0.0061,
+      "step": 3789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2255374261813944e-05,
+      "loss": 0.0009,
+      "step": 3790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2251670955643277e-05,
+      "loss": 0.0118,
+      "step": 3791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2247967324146476e-05,
+      "loss": 0.0481,
+      "step": 3792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2244263367858653e-05,
+      "loss": 0.0403,
+      "step": 3793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224055908731496e-05,
+      "loss": 0.0653,
+      "step": 3794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2236854483050605e-05,
+      "loss": 0.0832,
+      "step": 3795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233149555600827e-05,
+      "loss": 0.1041,
+      "step": 3796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2229444305500929e-05,
+      "loss": 0.0149,
+      "step": 3797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225738733286253e-05,
+      "loss": 0.1086,
+      "step": 3798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222032839492186e-05,
+      "loss": 0.04,
+      "step": 3799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221832662465416e-05,
+      "loss": 0.1018,
+      "step": 3800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2214620089307662e-05,
+      "loss": 0.0734,
+      "step": 3801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221091323398822e-05,
+      "loss": 0.0203,
+      "step": 3802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2207206059231407e-05,
+      "loss": 0.0586,
+      "step": 3803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220349856557284e-05,
+      "loss": 0.0568,
+      "step": 3804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2199790753548192e-05,
+      "loss": 0.0566,
+      "step": 3805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2196082623693169e-05,
+      "loss": 0.0085,
+      "step": 3806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219237417654353e-05,
+      "loss": 0.0577,
+      "step": 3807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2188665412635085e-05,
+      "loss": 0.0196,
+      "step": 3808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184956332503679e-05,
+      "loss": 0.0727,
+      "step": 3809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2181246936685209e-05,
+      "loss": 0.0063,
+      "step": 3810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177537225715615e-05,
+      "loss": 0.0533,
+      "step": 3811
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173827200130888e-05,
+      "loss": 0.0078,
+      "step": 3812
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170116860467057e-05,
+      "loss": 0.0067,
+      "step": 3813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2166406207260198e-05,
+      "loss": 0.007,
+      "step": 3814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2162695241046433e-05,
+      "loss": 0.0698,
+      "step": 3815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2158983962361938e-05,
+      "loss": 0.0072,
+      "step": 3816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2155272371742922e-05,
+      "loss": 0.0131,
+      "step": 3817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2151560469725639e-05,
+      "loss": 0.0075,
+      "step": 3818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147848256846397e-05,
+      "loss": 0.0129,
+      "step": 3819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2144135733641545e-05,
+      "loss": 0.1498,
+      "step": 3820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2140422900647473e-05,
+      "loss": 0.0276,
+      "step": 3821
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213670975840062e-05,
+      "loss": 0.0327,
+      "step": 3822
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213299630743747e-05,
+      "loss": 0.0608,
+      "step": 3823
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2129282548294547e-05,
+      "loss": 0.0867,
+      "step": 3824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2125568481508427e-05,
+      "loss": 0.0249,
+      "step": 3825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2121854107615726e-05,
+      "loss": 0.0784,
+      "step": 3826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21181394271531e-05,
+      "loss": 0.0038,
+      "step": 3827
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2114424440657255e-05,
+      "loss": 0.0035,
+      "step": 3828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211070914866494e-05,
+      "loss": 0.0977,
+      "step": 3829
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106993551712956e-05,
+      "loss": 0.0017,
+      "step": 3830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2103277650338128e-05,
+      "loss": 0.0472,
+      "step": 3831
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099561445077343e-05,
+      "loss": 0.043,
+      "step": 3832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209584493646753e-05,
+      "loss": 0.0027,
+      "step": 3833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092128125045652e-05,
+      "loss": 0.0371,
+      "step": 3834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2088411011348721e-05,
+      "loss": 0.0367,
+      "step": 3835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20846935959138e-05,
+      "loss": 0.0034,
+      "step": 3836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208097587927798e-05,
+      "loss": 0.0414,
+      "step": 3837
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077257861978416e-05,
+      "loss": 0.1977,
+      "step": 3838
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073539544552283e-05,
+      "loss": 0.002,
+      "step": 3839
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2069820927536821e-05,
+      "loss": 0.0016,
+      "step": 3840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2066102011469295e-05,
+      "loss": 0.0206,
+      "step": 3841
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2062382796887029e-05,
+      "loss": 0.0468,
+      "step": 3842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2058663284327379e-05,
+      "loss": 0.0433,
+      "step": 3843
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2054943474327749e-05,
+      "loss": 0.0297,
+      "step": 3844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051223367425581e-05,
+      "loss": 0.0057,
+      "step": 3845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047502964158373e-05,
+      "loss": 0.0492,
+      "step": 3846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204378226506365e-05,
+      "loss": 0.0041,
+      "step": 3847
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040061270678989e-05,
+      "loss": 0.0853,
+      "step": 3848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036339981542001e-05,
+      "loss": 0.0464,
+      "step": 3849
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032618398190354e-05,
+      "loss": 0.1533,
+      "step": 3850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2028896521161746e-05,
+      "loss": 0.0293,
+      "step": 3851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025174350993923e-05,
+      "loss": 0.0616,
+      "step": 3852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2021451888224669e-05,
+      "loss": 0.0345,
+      "step": 3853
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2017729133391815e-05,
+      "loss": 0.0188,
+      "step": 3854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2014006087033234e-05,
+      "loss": 0.1084,
+      "step": 3855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010282749686839e-05,
+      "loss": 0.1917,
+      "step": 3856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2006559121890585e-05,
+      "loss": 0.1218,
+      "step": 3857
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002835204182466e-05,
+      "loss": 0.0414,
+      "step": 3858
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1999110997100525e-05,
+      "loss": 0.1107,
+      "step": 3859
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1995386501182846e-05,
+      "loss": 0.0633,
+      "step": 3860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199166171696755e-05,
+      "loss": 0.0461,
+      "step": 3861
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987936644992797e-05,
+      "loss": 0.0295,
+      "step": 3862
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1984211285796799e-05,
+      "loss": 0.1245,
+      "step": 3863
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1980485639917798e-05,
+      "loss": 0.0179,
+      "step": 3864
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1976759707894092e-05,
+      "loss": 0.0378,
+      "step": 3865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973033490264e-05,
+      "loss": 0.0315,
+      "step": 3866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1969306987565905e-05,
+      "loss": 0.0302,
+      "step": 3867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1965580200338211e-05,
+      "loss": 0.0308,
+      "step": 3868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1961853129119381e-05,
+      "loss": 0.0591,
+      "step": 3869
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1958125774447899e-05,
+      "loss": 0.0464,
+      "step": 3870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954398136862308e-05,
+      "loss": 0.1342,
+      "step": 3871
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1950670216901184e-05,
+      "loss": 0.0383,
+      "step": 3872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946942015103145e-05,
+      "loss": 0.0168,
+      "step": 3873
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194321353200685e-05,
+      "loss": 0.0617,
+      "step": 3874
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1939484768150993e-05,
+      "loss": 0.0259,
+      "step": 3875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935755724074318e-05,
+      "loss": 0.0365,
+      "step": 3876
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1932026400315606e-05,
+      "loss": 0.0581,
+      "step": 3877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1928296797413677e-05,
+      "loss": 0.1397,
+      "step": 3878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1924566915907388e-05,
+      "loss": 0.0012,
+      "step": 3879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920836756335644e-05,
+      "loss": 0.0035,
+      "step": 3880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917106319237386e-05,
+      "loss": 0.0275,
+      "step": 3881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913375605151593e-05,
+      "loss": 0.0207,
+      "step": 3882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190964461461729e-05,
+      "loss": 0.0193,
+      "step": 3883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905913348173535e-05,
+      "loss": 0.0346,
+      "step": 3884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902181806359427e-05,
+      "loss": 0.0325,
+      "step": 3885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1898449989714116e-05,
+      "loss": 0.0152,
+      "step": 3886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1894717898776773e-05,
+      "loss": 0.0038,
+      "step": 3887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1890985534086626e-05,
+      "loss": 0.0229,
+      "step": 3888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887252896182928e-05,
+      "loss": 0.0195,
+      "step": 3889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188351998560498e-05,
+      "loss": 0.028,
+      "step": 3890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1879786802892128e-05,
+      "loss": 0.0068,
+      "step": 3891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187605334858374e-05,
+      "loss": 0.0056,
+      "step": 3892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1872319623219238e-05,
+      "loss": 0.0734,
+      "step": 3893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1868585627338077e-05,
+      "loss": 0.0018,
+      "step": 3894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1864851361479757e-05,
+      "loss": 0.0631,
+      "step": 3895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861116826183807e-05,
+      "loss": 0.0349,
+      "step": 3896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857382021989802e-05,
+      "loss": 0.0375,
+      "step": 3897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1853646949437356e-05,
+      "loss": 0.0129,
+      "step": 3898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849911609066118e-05,
+      "loss": 0.0929,
+      "step": 3899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1846176001415781e-05,
+      "loss": 0.0315,
+      "step": 3900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184244012702607e-05,
+      "loss": 0.0103,
+      "step": 3901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838703986436754e-05,
+      "loss": 0.0384,
+      "step": 3902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1834967580187635e-05,
+      "loss": 0.0088,
+      "step": 3903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831230908818563e-05,
+      "loss": 0.0567,
+      "step": 3904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827493972869415e-05,
+      "loss": 0.0109,
+      "step": 3905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182375677288011e-05,
+      "loss": 0.0861,
+      "step": 3906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1820019309390611e-05,
+      "loss": 0.0051,
+      "step": 3907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816281582940916e-05,
+      "loss": 0.032,
+      "step": 3908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1812543594071055e-05,
+      "loss": 0.0037,
+      "step": 3909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1808805343321102e-05,
+      "loss": 0.0605,
+      "step": 3910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805066831231165e-05,
+      "loss": 0.0024,
+      "step": 3911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1801328058341393e-05,
+      "loss": 0.0011,
+      "step": 3912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797589025191978e-05,
+      "loss": 0.0101,
+      "step": 3913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1793849732323132e-05,
+      "loss": 0.0518,
+      "step": 3914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1790110180275127e-05,
+      "loss": 0.0261,
+      "step": 3915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786370369588257e-05,
+      "loss": 0.0412,
+      "step": 3916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1782630300802853e-05,
+      "loss": 0.03,
+      "step": 3917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778889974459296e-05,
+      "loss": 0.0384,
+      "step": 3918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177514939109799e-05,
+      "loss": 0.0061,
+      "step": 3919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1771408551259382e-05,
+      "loss": 0.006,
+      "step": 3920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1767667455483964e-05,
+      "loss": 0.0022,
+      "step": 3921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176392610431225e-05,
+      "loss": 0.0316,
+      "step": 3922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1760184498284801e-05,
+      "loss": 0.0061,
+      "step": 3923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1756442637942212e-05,
+      "loss": 0.0135,
+      "step": 3924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.175270052382511e-05,
+      "loss": 0.0106,
+      "step": 3925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1748958156474176e-05,
+      "loss": 0.0033,
+      "step": 3926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17452155364301e-05,
+      "loss": 0.0715,
+      "step": 3927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741472664233633e-05,
+      "loss": 0.0082,
+      "step": 3928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1737729540425554e-05,
+      "loss": 0.1447,
+      "step": 3929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733986165546667e-05,
+      "loss": 0.0347,
+      "step": 3930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730242540137835e-05,
+      "loss": 0.0033,
+      "step": 3931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1726498664739939e-05,
+      "loss": 0.0202,
+      "step": 3932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722754539893896e-05,
+      "loss": 0.0049,
+      "step": 3933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1719010166140676e-05,
+      "loss": 0.1172,
+      "step": 3934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715265544021273e-05,
+      "loss": 0.0075,
+      "step": 3935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1711520674076711e-05,
+      "loss": 0.0562,
+      "step": 3936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707775556848058e-05,
+      "loss": 0.0397,
+      "step": 3937
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170403019287642e-05,
+      "loss": 0.0062,
+      "step": 3938
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700284582702933e-05,
+      "loss": 0.033,
+      "step": 3939
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1696538726868768e-05,
+      "loss": 0.0016,
+      "step": 3940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169279262591514e-05,
+      "loss": 0.0408,
+      "step": 3941
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168904628038329e-05,
+      "loss": 0.1793,
+      "step": 3942
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685299690814499e-05,
+      "loss": 0.0067,
+      "step": 3943
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168155285775008e-05,
+      "loss": 0.0101,
+      "step": 3944
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1677805781731383e-05,
+      "loss": 0.1201,
+      "step": 3945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1674058463299798e-05,
+      "loss": 0.0498,
+      "step": 3946
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670310902996739e-05,
+      "loss": 0.0043,
+      "step": 3947
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166656310136367e-05,
+      "loss": 0.0533,
+      "step": 3948
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1662815058942072e-05,
+      "loss": 0.0796,
+      "step": 3949
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659066776273475e-05,
+      "loss": 0.0171,
+      "step": 3950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1655318253899436e-05,
+      "loss": 0.0723,
+      "step": 3951
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651569492361554e-05,
+      "loss": 0.0052,
+      "step": 3952
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1647820492201454e-05,
+      "loss": 0.0414,
+      "step": 3953
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16440712539608e-05,
+      "loss": 0.0243,
+      "step": 3954
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164032177818129e-05,
+      "loss": 0.0345,
+      "step": 3955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1636572065404657e-05,
+      "loss": 0.0028,
+      "step": 3956
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1632822116172666e-05,
+      "loss": 0.0036,
+      "step": 3957
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1629071931027119e-05,
+      "loss": 0.0127,
+      "step": 3958
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1625321510509848e-05,
+      "loss": 0.1345,
+      "step": 3959
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1621570855162722e-05,
+      "loss": 0.0415,
+      "step": 3960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161781996552765e-05,
+      "loss": 0.0262,
+      "step": 3961
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161406884214656e-05,
+      "loss": 0.043,
+      "step": 3962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1610317485561425e-05,
+      "loss": 0.042,
+      "step": 3963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160656589631425e-05,
+      "loss": 0.0152,
+      "step": 3964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1602814074947071e-05,
+      "loss": 0.1373,
+      "step": 3965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1599062022001964e-05,
+      "loss": 0.0349,
+      "step": 3966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1595309738021023e-05,
+      "loss": 0.0163,
+      "step": 3967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1591557223546394e-05,
+      "loss": 0.0324,
+      "step": 3968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1587804479120247e-05,
+      "loss": 0.0382,
+      "step": 3969
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584051505284786e-05,
+      "loss": 0.0567,
+      "step": 3970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1580298302582249e-05,
+      "loss": 0.0186,
+      "step": 3971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1576544871554904e-05,
+      "loss": 0.0859,
+      "step": 3972
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1572791212745056e-05,
+      "loss": 0.0325,
+      "step": 3973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569037326695043e-05,
+      "loss": 0.0024,
+      "step": 3974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1565283213947232e-05,
+      "loss": 0.0097,
+      "step": 3975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1561528875044026e-05,
+      "loss": 0.008,
+      "step": 3976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1557774310527859e-05,
+      "loss": 0.0016,
+      "step": 3977
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1554019520941202e-05,
+      "loss": 0.0659,
+      "step": 3978
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550264506826552e-05,
+      "loss": 0.0295,
+      "step": 3979
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546509268726437e-05,
+      "loss": 0.0026,
+      "step": 3980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1542753807183429e-05,
+      "loss": 0.0042,
+      "step": 3981
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1538998122740121e-05,
+      "loss": 0.0434,
+      "step": 3982
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1535242215939145e-05,
+      "loss": 0.2794,
+      "step": 3983
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1531486087323159e-05,
+      "loss": 0.0834,
+      "step": 3984
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527729737434859e-05,
+      "loss": 0.0565,
+      "step": 3985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523973166816967e-05,
+      "loss": 0.0849,
+      "step": 3986
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152021637601224e-05,
+      "loss": 0.0072,
+      "step": 3987
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1516459365563476e-05,
+      "loss": 0.0278,
+      "step": 3988
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1512702136013482e-05,
+      "loss": 0.0015,
+      "step": 3989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508944687905119e-05,
+      "loss": 0.1026,
+      "step": 3990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150518702178127e-05,
+      "loss": 0.011,
+      "step": 3991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501429138184854e-05,
+      "loss": 0.0119,
+      "step": 3992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497671037658811e-05,
+      "loss": 0.022,
+      "step": 3993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1493912720746118e-05,
+      "loss": 0.012,
+      "step": 3994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149015418798979e-05,
+      "loss": 0.0062,
+      "step": 3995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486395439932869e-05,
+      "loss": 0.035,
+      "step": 3996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148263647711842e-05,
+      "loss": 0.0537,
+      "step": 3997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147887730008955e-05,
+      "loss": 0.023,
+      "step": 3998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1475117909389396e-05,
+      "loss": 0.0072,
+      "step": 3999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471358305561115e-05,
+      "loss": 0.0191,
+      "step": 4000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1467598489147911e-05,
+      "loss": 0.0219,
+      "step": 4001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1463838460693001e-05,
+      "loss": 0.0096,
+      "step": 4002
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1460078220739648e-05,
+      "loss": 0.0009,
+      "step": 4003
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1456317769831138e-05,
+      "loss": 0.003,
+      "step": 4004
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1452557108510789e-05,
+      "loss": 0.0788,
+      "step": 4005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1448796237321949e-05,
+      "loss": 0.0189,
+      "step": 4006
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445035156807995e-05,
+      "loss": 0.0604,
+      "step": 4007
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441273867512337e-05,
+      "loss": 0.0162,
+      "step": 4008
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1437512369978414e-05,
+      "loss": 0.006,
+      "step": 4009
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1433750664749697e-05,
+      "loss": 0.0249,
+      "step": 4010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429988752369681e-05,
+      "loss": 0.0568,
+      "step": 4011
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1426226633381899e-05,
+      "loss": 0.1117,
+      "step": 4012
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1422464308329904e-05,
+      "loss": 0.0837,
+      "step": 4013
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1418701777757293e-05,
+      "loss": 0.0108,
+      "step": 4014
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1414939042207677e-05,
+      "loss": 0.0585,
+      "step": 4015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411176102224706e-05,
+      "loss": 0.0169,
+      "step": 4016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140741295835206e-05,
+      "loss": 0.0042,
+      "step": 4017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1403649611133444e-05,
+      "loss": 0.023,
+      "step": 4018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399886061112598e-05,
+      "loss": 0.0109,
+      "step": 4019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1396122308833277e-05,
+      "loss": 0.0766,
+      "step": 4020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392358354839286e-05,
+      "loss": 0.0079,
+      "step": 4021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1388594199674443e-05,
+      "loss": 0.0091,
+      "step": 4022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1384829843882607e-05,
+      "loss": 0.0069,
+      "step": 4023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381065288007656e-05,
+      "loss": 0.0606,
+      "step": 4024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1377300532593501e-05,
+      "loss": 0.0495,
+      "step": 4025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1373535578184083e-05,
+      "loss": 0.0259,
+      "step": 4026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1369770425323371e-05,
+      "loss": 0.0271,
+      "step": 4027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136600507455536e-05,
+      "loss": 0.0056,
+      "step": 4028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1362239526424075e-05,
+      "loss": 0.0002,
+      "step": 4029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135847378147357e-05,
+      "loss": 0.0888,
+      "step": 4030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354707840247936e-05,
+      "loss": 0.0194,
+      "step": 4031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1350941703291274e-05,
+      "loss": 0.1497,
+      "step": 4032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134717537114773e-05,
+      "loss": 0.0106,
+      "step": 4033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343408844361465e-05,
+      "loss": 0.0305,
+      "step": 4034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1339642123476679e-05,
+      "loss": 0.0898,
+      "step": 4035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1335875209037598e-05,
+      "loss": 0.0342,
+      "step": 4036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1332108101588468e-05,
+      "loss": 0.1643,
+      "step": 4037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1328340801673571e-05,
+      "loss": 0.0569,
+      "step": 4038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324573309837216e-05,
+      "loss": 0.0344,
+      "step": 4039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320805626623738e-05,
+      "loss": 0.0077,
+      "step": 4040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13170377525775e-05,
+      "loss": 0.0717,
+      "step": 4041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1313269688242887e-05,
+      "loss": 0.0124,
+      "step": 4042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1309501434164323e-05,
+      "loss": 0.0202,
+      "step": 4043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1305732990886252e-05,
+      "loss": 0.0104,
+      "step": 4044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301964358953142e-05,
+      "loss": 0.0171,
+      "step": 4045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1298195538909498e-05,
+      "loss": 0.0613,
+      "step": 4046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1294426531299847e-05,
+      "loss": 0.0058,
+      "step": 4047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1290657336668738e-05,
+      "loss": 0.0207,
+      "step": 4048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128688795556076e-05,
+      "loss": 0.1139,
+      "step": 4049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1283118388520513e-05,
+      "loss": 0.024,
+      "step": 4050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1279348636092634e-05,
+      "loss": 0.0622,
+      "step": 4051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275578698821793e-05,
+      "loss": 0.0067,
+      "step": 4052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271808577252667e-05,
+      "loss": 0.0026,
+      "step": 4053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126803827192998e-05,
+      "loss": 0.0655,
+      "step": 4054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1264267783398463e-05,
+      "loss": 0.0063,
+      "step": 4055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260497112202895e-05,
+      "loss": 0.0164,
+      "step": 4056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1256726258888064e-05,
+      "loss": 0.04,
+      "step": 4057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252955223998795e-05,
+      "loss": 0.0015,
+      "step": 4058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1249184008079932e-05,
+      "loss": 0.0394,
+      "step": 4059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1245412611676347e-05,
+      "loss": 0.0362,
+      "step": 4060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1241641035332944e-05,
+      "loss": 0.119,
+      "step": 4061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1237869279594647e-05,
+      "loss": 0.0102,
+      "step": 4062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12340973450064e-05,
+      "loss": 0.1004,
+      "step": 4063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1230325232113189e-05,
+      "loss": 0.0026,
+      "step": 4064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1226552941460011e-05,
+      "loss": 0.0291,
+      "step": 4065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222780473591902e-05,
+      "loss": 0.0077,
+      "step": 4066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219007829053905e-05,
+      "loss": 0.0318,
+      "step": 4067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121523500839111e-05,
+      "loss": 0.0351,
+      "step": 4068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1211462012148613e-05,
+      "loss": 0.007,
+      "step": 4069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1207688840871552e-05,
+      "loss": 0.0041,
+      "step": 4070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120391549510508e-05,
+      "loss": 0.0134,
+      "step": 4071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1200141975394373e-05,
+      "loss": 0.0369,
+      "step": 4072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1196368282284647e-05,
+      "loss": 0.0249,
+      "step": 4073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192594416321122e-05,
+      "loss": 0.0496,
+      "step": 4074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188820378049066e-05,
+      "loss": 0.0197,
+      "step": 4075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1185046168013752e-05,
+      "loss": 0.0033,
+      "step": 4076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1181271786760486e-05,
+      "loss": 0.0818,
+      "step": 4077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1177497234834599e-05,
+      "loss": 0.0042,
+      "step": 4078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173722512781447e-05,
+      "loss": 0.0912,
+      "step": 4079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116994762114641e-05,
+      "loss": 0.0532,
+      "step": 4080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1166172560474894e-05,
+      "loss": 0.0079,
+      "step": 4081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162397331312323e-05,
+      "loss": 0.0906,
+      "step": 4082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1158621934204153e-05,
+      "loss": 0.0152,
+      "step": 4083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1154846369695864e-05,
+      "loss": 0.0342,
+      "step": 4084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115107063833295e-05,
+      "loss": 0.0058,
+      "step": 4085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147294740660943e-05,
+      "loss": 0.0713,
+      "step": 4086
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114351867722539e-05,
+      "loss": 0.0085,
+      "step": 4087
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139742448571867e-05,
+      "loss": 0.0059,
+      "step": 4088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113596605524597e-05,
+      "loss": 0.0479,
+      "step": 4089
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1132189497793316e-05,
+      "loss": 0.0178,
+      "step": 4090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1128412776759556e-05,
+      "loss": 0.1047,
+      "step": 4091
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124635892690353e-05,
+      "loss": 0.0158,
+      "step": 4092
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1120858846131408e-05,
+      "loss": 0.0164,
+      "step": 4093
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1117081637628426e-05,
+      "loss": 0.094,
+      "step": 4094
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113304267727157e-05,
+      "loss": 0.0345,
+      "step": 4095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110952673697335e-05,
+      "loss": 0.0925,
+      "step": 4096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1105749045912804e-05,
+      "loss": 0.0914,
+      "step": 4097
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1101971195091317e-05,
+      "loss": 0.0276,
+      "step": 4098
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1098193185054726e-05,
+      "loss": 0.1724,
+      "step": 4099
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1094415016348886e-05,
+      "loss": 0.0296,
+      "step": 4100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1090636689519673e-05,
+      "loss": 0.0518,
+      "step": 4101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086858205112991e-05,
+      "loss": 0.005,
+      "step": 4102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108307956367476e-05,
+      "loss": 0.0074,
+      "step": 4103
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1079300765750924e-05,
+      "loss": 0.0186,
+      "step": 4104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1075521811887455e-05,
+      "loss": 0.0334,
+      "step": 4105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071742702630346e-05,
+      "loss": 0.0806,
+      "step": 4106
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1067963438525606e-05,
+      "loss": 0.0603,
+      "step": 4107
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1064184020119273e-05,
+      "loss": 0.0234,
+      "step": 4108
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1060404447957407e-05,
+      "loss": 0.1009,
+      "step": 4109
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1056624722586086e-05,
+      "loss": 0.0058,
+      "step": 4110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1052844844551416e-05,
+      "loss": 0.0186,
+      "step": 4111
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1049064814399514e-05,
+      "loss": 0.044,
+      "step": 4112
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 0.0582,
+      "step": 4113
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1041504299928644e-05,
+      "loss": 0.1382,
+      "step": 4114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1037723816702033e-05,
+      "loss": 0.0077,
+      "step": 4115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1033943183542914e-05,
+      "loss": 0.0884,
+      "step": 4116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1030162400997518e-05,
+      "loss": 0.0201,
+      "step": 4117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10263814696121e-05,
+      "loss": 0.0059,
+      "step": 4118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1022600389932942e-05,
+      "loss": 0.0531,
+      "step": 4119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018819162506338e-05,
+      "loss": 0.0211,
+      "step": 4120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101503778787861e-05,
+      "loss": 0.1366,
+      "step": 4121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011256266596096e-05,
+      "loss": 0.0681,
+      "step": 4122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007474599205164e-05,
+      "loss": 0.0392,
+      "step": 4123
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003692786252191e-05,
+      "loss": 0.0025,
+      "step": 4124
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999910828283582e-05,
+      "loss": 0.0171,
+      "step": 4125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0996128725845764e-05,
+      "loss": 0.0377,
+      "step": 4126
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0992346479485183e-05,
+      "loss": 0.0527,
+      "step": 4127
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0988564089748308e-05,
+      "loss": 0.0332,
+      "step": 4128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984781557181624e-05,
+      "loss": 0.028,
+      "step": 4129
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0980998882331637e-05,
+      "loss": 0.0017,
+      "step": 4130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097721606574488e-05,
+      "loss": 0.0031,
+      "step": 4131
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0973433107967901e-05,
+      "loss": 0.003,
+      "step": 4132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969650009547265e-05,
+      "loss": 0.0336,
+      "step": 4133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096586677102957e-05,
+      "loss": 0.0532,
+      "step": 4134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096208339296142e-05,
+      "loss": 0.0222,
+      "step": 4135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958299875889451e-05,
+      "loss": 0.0147,
+      "step": 4136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0954516220360309e-05,
+      "loss": 0.0773,
+      "step": 4137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950732426920664e-05,
+      "loss": 0.0016,
+      "step": 4138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0946948496117209e-05,
+      "loss": 0.0197,
+      "step": 4139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0943164428496653e-05,
+      "loss": 0.1272,
+      "step": 4140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0939380224605727e-05,
+      "loss": 0.0424,
+      "step": 4141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093559588499118e-05,
+      "loss": 0.0036,
+      "step": 4142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093181141019978e-05,
+      "loss": 0.0354,
+      "step": 4143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092802680077832e-05,
+      "loss": 0.0028,
+      "step": 4144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0924242057273603e-05,
+      "loss": 0.0071,
+      "step": 4145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092045718023246e-05,
+      "loss": 0.0106,
+      "step": 4146
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0916672170201737e-05,
+      "loss": 0.041,
+      "step": 4147
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0912887027728299e-05,
+      "loss": 0.0092,
+      "step": 4148
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0909101753359033e-05,
+      "loss": 0.0124,
+      "step": 4149
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0905316347640844e-05,
+      "loss": 0.0604,
+      "step": 4150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0901530811120655e-05,
+      "loss": 0.0501,
+      "step": 4151
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897745144345406e-05,
+      "loss": 0.003,
+      "step": 4152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0893959347862058e-05,
+      "loss": 0.0109,
+      "step": 4153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890173422217597e-05,
+      "loss": 0.0118,
+      "step": 4154
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886387367959012e-05,
+      "loss": 0.0486,
+      "step": 4155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0882601185633326e-05,
+      "loss": 0.0391,
+      "step": 4156
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0878814875787576e-05,
+      "loss": 0.1775,
+      "step": 4157
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0875028438968811e-05,
+      "loss": 0.1334,
+      "step": 4158
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0871241875724107e-05,
+      "loss": 0.0041,
+      "step": 4159
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0867455186600551e-05,
+      "loss": 0.0077,
+      "step": 4160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0863668372145255e-05,
+      "loss": 0.0114,
+      "step": 4161
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0859881432905346e-05,
+      "loss": 0.0006,
+      "step": 4162
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0856094369427968e-05,
+      "loss": 0.1109,
+      "step": 4163
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852307182260284e-05,
+      "loss": 0.0294,
+      "step": 4164
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0848519871949472e-05,
+      "loss": 0.0054,
+      "step": 4165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0844732439042736e-05,
+      "loss": 0.1609,
+      "step": 4166
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0840944884087286e-05,
+      "loss": 0.0036,
+      "step": 4167
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0837157207630362e-05,
+      "loss": 0.0284,
+      "step": 4168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083336941021921e-05,
+      "loss": 0.0449,
+      "step": 4169
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0829581492401103e-05,
+      "loss": 0.0617,
+      "step": 4170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825793454723325e-05,
+      "loss": 0.0241,
+      "step": 4171
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0822005297733183e-05,
+      "loss": 0.0011,
+      "step": 4172
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081821702197799e-05,
+      "loss": 0.0314,
+      "step": 4173
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0814428628005092e-05,
+      "loss": 0.0278,
+      "step": 4174
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081064011636184e-05,
+      "loss": 0.0068,
+      "step": 4175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080685148759561e-05,
+      "loss": 0.1127,
+      "step": 4176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0803062742253785e-05,
+      "loss": 0.0059,
+      "step": 4177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0799273880883777e-05,
+      "loss": 0.0248,
+      "step": 4178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0795484904033003e-05,
+      "loss": 0.0175,
+      "step": 4179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791695812248906e-05,
+      "loss": 0.0727,
+      "step": 4180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0787906606078942e-05,
+      "loss": 0.0043,
+      "step": 4181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078411728607058e-05,
+      "loss": 0.0396,
+      "step": 4182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780327852771312e-05,
+      "loss": 0.0336,
+      "step": 4183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776538306728645e-05,
+      "loss": 0.0012,
+      "step": 4184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07727486484901e-05,
+      "loss": 0.0915,
+      "step": 4185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076895887860321e-05,
+      "loss": 0.0421,
+      "step": 4186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765168997615531e-05,
+      "loss": 0.0093,
+      "step": 4187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761379006074633e-05,
+      "loss": 0.0523,
+      "step": 4188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0757588904528106e-05,
+      "loss": 0.0441,
+      "step": 4189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753798693523545e-05,
+      "loss": 0.0017,
+      "step": 4190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0750008373608571e-05,
+      "loss": 0.0083,
+      "step": 4191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746217945330821e-05,
+      "loss": 0.0036,
+      "step": 4192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0742427409237939e-05,
+      "loss": 0.0453,
+      "step": 4193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073863676587759e-05,
+      "loss": 0.0075,
+      "step": 4194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0734846015797456e-05,
+      "loss": 0.1083,
+      "step": 4195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073105515954523e-05,
+      "loss": 0.0827,
+      "step": 4196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0727264197668626e-05,
+      "loss": 0.0516,
+      "step": 4197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723473130715367e-05,
+      "loss": 0.0094,
+      "step": 4198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07196819592332e-05,
+      "loss": 0.0065,
+      "step": 4199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0715890683769872e-05,
+      "loss": 0.0148,
+      "step": 4200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0712099304873161e-05,
+      "loss": 0.0024,
+      "step": 4201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070830782309085e-05,
+      "loss": 0.0113,
+      "step": 4202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0704516238970748e-05,
+      "loss": 0.0062,
+      "step": 4203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0700724553060662e-05,
+      "loss": 0.1402,
+      "step": 4204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696932765908426e-05,
+      "loss": 0.0561,
+      "step": 4205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0693140878061886e-05,
+      "loss": 0.0015,
+      "step": 4206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689348890068903e-05,
+      "loss": 0.017,
+      "step": 4207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685556802477347e-05,
+      "loss": 0.0793,
+      "step": 4208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0681764615835106e-05,
+      "loss": 0.137,
+      "step": 4209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0677972330690089e-05,
+      "loss": 0.0483,
+      "step": 4210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0674179947590213e-05,
+      "loss": 0.0035,
+      "step": 4211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0670387467083405e-05,
+      "loss": 0.0413,
+      "step": 4212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0666594889717615e-05,
+      "loss": 0.0082,
+      "step": 4213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0662802216040798e-05,
+      "loss": 0.0474,
+      "step": 4214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659009446600929e-05,
+      "loss": 0.0583,
+      "step": 4215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0655216581945997e-05,
+      "loss": 0.0017,
+      "step": 4216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0651423622624001e-05,
+      "loss": 0.0011,
+      "step": 4217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0647630569182955e-05,
+      "loss": 0.0265,
+      "step": 4218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0643837422170893e-05,
+      "loss": 0.0136,
+      "step": 4219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0640044182135853e-05,
+      "loss": 0.0176,
+      "step": 4220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0636250849625886e-05,
+      "loss": 0.0075,
+      "step": 4221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632457425189067e-05,
+      "loss": 0.0828,
+      "step": 4222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0628663909373474e-05,
+      "loss": 0.0055,
+      "step": 4223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0624870302727205e-05,
+      "loss": 0.0536,
+      "step": 4224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0621076605798365e-05,
+      "loss": 0.103,
+      "step": 4225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0617282819135077e-05,
+      "loss": 0.0941,
+      "step": 4226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061348894328548e-05,
+      "loss": 0.0028,
+      "step": 4227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060969497879771e-05,
+      "loss": 0.0455,
+      "step": 4228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060590092621994e-05,
+      "loss": 0.0073,
+      "step": 4229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060210678610033e-05,
+      "loss": 0.0788,
+      "step": 4230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0598312558987072e-05,
+      "loss": 0.0234,
+      "step": 4231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0594518245428365e-05,
+      "loss": 0.0707,
+      "step": 4232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0590723845972418e-05,
+      "loss": 0.0388,
+      "step": 4233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586929361167455e-05,
+      "loss": 0.0772,
+      "step": 4234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0583134791561707e-05,
+      "loss": 0.0011,
+      "step": 4235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0579340137703422e-05,
+      "loss": 0.0105,
+      "step": 4236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0575545400140863e-05,
+      "loss": 0.0573,
+      "step": 4237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571750579422302e-05,
+      "loss": 0.0577,
+      "step": 4238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0567955676096018e-05,
+      "loss": 0.0239,
+      "step": 4239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056416069071031e-05,
+      "loss": 0.0013,
+      "step": 4240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560365623813485e-05,
+      "loss": 0.0041,
+      "step": 4241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0556570475953862e-05,
+      "loss": 0.0231,
+      "step": 4242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552775247679769e-05,
+      "loss": 0.0357,
+      "step": 4243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0548979939539554e-05,
+      "loss": 0.0116,
+      "step": 4244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0545184552081567e-05,
+      "loss": 0.0662,
+      "step": 4245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0541389085854177e-05,
+      "loss": 0.0229,
+      "step": 4246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053759354140576e-05,
+      "loss": 0.0011,
+      "step": 4247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05337979192847e-05,
+      "loss": 0.0188,
+      "step": 4248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530002220039396e-05,
+      "loss": 0.025,
+      "step": 4249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0526206444218266e-05,
+      "loss": 0.0034,
+      "step": 4250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0522410592369727e-05,
+      "loss": 0.0015,
+      "step": 4251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0518614665042212e-05,
+      "loss": 0.0192,
+      "step": 4252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0514818662784166e-05,
+      "loss": 0.0087,
+      "step": 4253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0511022586144042e-05,
+      "loss": 0.0852,
+      "step": 4254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0507226435670307e-05,
+      "loss": 0.0635,
+      "step": 4255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050343021191143e-05,
+      "loss": 0.0387,
+      "step": 4256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499633915415903e-05,
+      "loss": 0.0077,
+      "step": 4257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0495837546732224e-05,
+      "loss": 0.0378,
+      "step": 4258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04920411064089e-05,
+      "loss": 0.0056,
+      "step": 4259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0488244594994444e-05,
+      "loss": 0.0038,
+      "step": 4260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048444801303739e-05,
+      "loss": 0.0068,
+      "step": 4261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048065136108627e-05,
+      "loss": 0.0018,
+      "step": 4262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476854639689635e-05,
+      "loss": 0.0247,
+      "step": 4263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047305784939605e-05,
+      "loss": 0.0531,
+      "step": 4264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046926099075407e-05,
+      "loss": 0.0153,
+      "step": 4265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046546406431228e-05,
+      "loss": 0.0049,
+      "step": 4266
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046166707061927e-05,
+      "loss": 0.0657,
+      "step": 4267
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457870010223637e-05,
+      "loss": 0.0029,
+      "step": 4268
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0454072883673985e-05,
+      "loss": 0.1076,
+      "step": 4269
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0450275691518932e-05,
+      "loss": 0.0054,
+      "step": 4270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0446478434307103e-05,
+      "loss": 0.004,
+      "step": 4271
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442681112587137e-05,
+      "loss": 0.0016,
+      "step": 4272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0438883726907678e-05,
+      "loss": 0.0059,
+      "step": 4273
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0435086277817379e-05,
+      "loss": 0.0683,
+      "step": 4274
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0431288765864904e-05,
+      "loss": 0.0355,
+      "step": 4275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427491191598926e-05,
+      "loss": 0.0639,
+      "step": 4276
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423693555568126e-05,
+      "loss": 0.097,
+      "step": 4277
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0419895858321194e-05,
+      "loss": 0.0126,
+      "step": 4278
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0416098100406828e-05,
+      "loss": 0.0907,
+      "step": 4279
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0412300282373737e-05,
+      "loss": 0.0255,
+      "step": 4280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408502404770641e-05,
+      "loss": 0.001,
+      "step": 4281
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0404704468146263e-05,
+      "loss": 0.032,
+      "step": 4282
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0400906473049334e-05,
+      "loss": 0.0067,
+      "step": 4283
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397108420028598e-05,
+      "loss": 0.0017,
+      "step": 4284
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0393310309632809e-05,
+      "loss": 0.0007,
+      "step": 4285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0389512142410723e-05,
+      "loss": 0.0027,
+      "step": 4286
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0385713918911104e-05,
+      "loss": 0.0021,
+      "step": 4287
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381915639682733e-05,
+      "loss": 0.0439,
+      "step": 4288
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0378117305274394e-05,
+      "loss": 0.0012,
+      "step": 4289
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037431891623487e-05,
+      "loss": 0.0437,
+      "step": 4290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037052047311297e-05,
+      "loss": 0.0053,
+      "step": 4291
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366721976457494e-05,
+      "loss": 0.0058,
+      "step": 4292
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036292342681726e-05,
+      "loss": 0.0827,
+      "step": 4293
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359124824741091e-05,
+      "loss": 0.0401,
+      "step": 4294
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355326170777812e-05,
+      "loss": 0.0081,
+      "step": 4295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.035152746547627e-05,
+      "loss": 0.0438,
+      "step": 4296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0347728709385298e-05,
+      "loss": 0.0432,
+      "step": 4297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0343929903053755e-05,
+      "loss": 0.0563,
+      "step": 4298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0340131047030504e-05,
+      "loss": 0.0057,
+      "step": 4299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0336332141864404e-05,
+      "loss": 0.0046,
+      "step": 4300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0332533188104332e-05,
+      "loss": 0.0406,
+      "step": 4301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032873418629917e-05,
+      "loss": 0.024,
+      "step": 4302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0324935136997807e-05,
+      "loss": 0.0078,
+      "step": 4303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321136040749134e-05,
+      "loss": 0.0323,
+      "step": 4304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031733689810205e-05,
+      "loss": 0.0069,
+      "step": 4305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031353770960547e-05,
+      "loss": 0.0003,
+      "step": 4306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0309738475808305e-05,
+      "loss": 0.1,
+      "step": 4307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0305939197259478e-05,
+      "loss": 0.0412,
+      "step": 4308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302139874507912e-05,
+      "loss": 0.0253,
+      "step": 4309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029834050810255e-05,
+      "loss": 0.0688,
+      "step": 4310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294541098592323e-05,
+      "loss": 0.0029,
+      "step": 4311
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0290741646526183e-05,
+      "loss": 0.0072,
+      "step": 4312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0286942152453079e-05,
+      "loss": 0.0569,
+      "step": 4313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0283142616921974e-05,
+      "loss": 0.0026,
+      "step": 4314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027934304048183e-05,
+      "loss": 0.0579,
+      "step": 4315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0275543423681622e-05,
+      "loss": 0.0075,
+      "step": 4316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0271743767070321e-05,
+      "loss": 0.0074,
+      "step": 4317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0267944071196911e-05,
+      "loss": 0.0015,
+      "step": 4318
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0264144336610383e-05,
+      "loss": 0.0654,
+      "step": 4319
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0260344563859726e-05,
+      "loss": 0.0345,
+      "step": 4320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0256544753493945e-05,
+      "loss": 0.1366,
+      "step": 4321
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0252744906062038e-05,
+      "loss": 0.0419,
+      "step": 4322
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0248945022113019e-05,
+      "loss": 0.0043,
+      "step": 4323
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0245145102195902e-05,
+      "loss": 0.0392,
+      "step": 4324
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0241345146859707e-05,
+      "loss": 0.002,
+      "step": 4325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0237545156653461e-05,
+      "loss": 0.0949,
+      "step": 4326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233745132126192e-05,
+      "loss": 0.0801,
+      "step": 4327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229945073826938e-05,
+      "loss": 0.075,
+      "step": 4328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022614498230474e-05,
+      "loss": 0.0016,
+      "step": 4329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022234485810864e-05,
+      "loss": 0.0094,
+      "step": 4330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0218544701787692e-05,
+      "loss": 0.0028,
+      "step": 4331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0214744513890945e-05,
+      "loss": 0.0039,
+      "step": 4332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021094429496746e-05,
+      "loss": 0.0054,
+      "step": 4333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0207144045566305e-05,
+      "loss": 0.045,
+      "step": 4334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0203343766236542e-05,
+      "loss": 0.0037,
+      "step": 4335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0199543457527247e-05,
+      "loss": 0.0054,
+      "step": 4336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0195743119987498e-05,
+      "loss": 0.0117,
+      "step": 4337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019194275416637e-05,
+      "loss": 0.0055,
+      "step": 4338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188142360612956e-05,
+      "loss": 0.0014,
+      "step": 4339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184341939876334e-05,
+      "loss": 0.0418,
+      "step": 4340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0180541492505605e-05,
+      "loss": 0.0035,
+      "step": 4341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0176741019049863e-05,
+      "loss": 0.0895,
+      "step": 4342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0172940520058209e-05,
+      "loss": 0.0357,
+      "step": 4343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016913999607975e-05,
+      "loss": 0.0571,
+      "step": 4344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0165339447663586e-05,
+      "loss": 0.0253,
+      "step": 4345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0161538875358833e-05,
+      "loss": 0.0408,
+      "step": 4346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015773827971461e-05,
+      "loss": 0.0631,
+      "step": 4347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0153937661280024e-05,
+      "loss": 0.1238,
+      "step": 4348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0150137020604203e-05,
+      "loss": 0.135,
+      "step": 4349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0146336358236273e-05,
+      "loss": 0.1463,
+      "step": 4350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014253567472536e-05,
+      "loss": 0.073,
+      "step": 4351
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0138734970620597e-05,
+      "loss": 0.0828,
+      "step": 4352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0134934246471109e-05,
+      "loss": 0.0126,
+      "step": 4353
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0131133502826041e-05,
+      "loss": 0.0323,
+      "step": 4354
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012733274023453e-05,
+      "loss": 0.0481,
+      "step": 4355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123531959245721e-05,
+      "loss": 0.0159,
+      "step": 4356
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0119731160408752e-05,
+      "loss": 0.038,
+      "step": 4357
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0115930344272777e-05,
+      "loss": 0.0083,
+      "step": 4358
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112129511386942e-05,
+      "loss": 0.1021,
+      "step": 4359
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0108328662300399e-05,
+      "loss": 0.0456,
+      "step": 4360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0104527797562308e-05,
+      "loss": 0.0062,
+      "step": 4361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0100726917721818e-05,
+      "loss": 0.0164,
+      "step": 4362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0096926023328095e-05,
+      "loss": 0.0118,
+      "step": 4363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0093125114930297e-05,
+      "loss": 0.0619,
+      "step": 4364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008932419307759e-05,
+      "loss": 0.0013,
+      "step": 4365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0085523258319131e-05,
+      "loss": 0.0303,
+      "step": 4366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0081722311204095e-05,
+      "loss": 0.0057,
+      "step": 4367
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077921352281648e-05,
+      "loss": 0.0649,
+      "step": 4368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074120382100962e-05,
+      "loss": 0.0066,
+      "step": 4369
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0070319401211209e-05,
+      "loss": 0.0044,
+      "step": 4370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006651841016156e-05,
+      "loss": 0.0909,
+      "step": 4371
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006271740950119e-05,
+      "loss": 0.0074,
+      "step": 4372
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005891639977928e-05,
+      "loss": 0.0868,
+      "step": 4373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055115381545006e-05,
+      "loss": 0.0273,
+      "step": 4374
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0051314355347546e-05,
+      "loss": 0.0051,
+      "step": 4375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0047513321736077e-05,
+      "loss": 0.0008,
+      "step": 4376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043712281259785e-05,
+      "loss": 0.0193,
+      "step": 4377
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039911234467856e-05,
+      "loss": 0.0311,
+      "step": 4378
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036110181909466e-05,
+      "loss": 0.0804,
+      "step": 4379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032309124133798e-05,
+      "loss": 0.0942,
+      "step": 4380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028508061690044e-05,
+      "loss": 0.0041,
+      "step": 4381
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024706995127385e-05,
+      "loss": 0.0144,
+      "step": 4382
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020905924995003e-05,
+      "loss": 0.012,
+      "step": 4383
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0017104851842094e-05,
+      "loss": 0.1254,
+      "step": 4384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0013303776217838e-05,
+      "loss": 0.1168,
+      "step": 4385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0009502698671427e-05,
+      "loss": 0.0469,
+      "step": 4386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0005701619752046e-05,
+      "loss": 0.0785,
+      "step": 4387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0001900540008883e-05,
+      "loss": 0.0086,
+      "step": 4388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.998099459991122e-06,
+      "loss": 0.1085,
+      "step": 4389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994298380247957e-06,
+      "loss": 0.0046,
+      "step": 4390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990497301328577e-06,
+      "loss": 0.043,
+      "step": 4391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986696223782164e-06,
+      "loss": 0.0068,
+      "step": 4392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982895148157908e-06,
+      "loss": 0.005,
+      "step": 4393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979094075004995e-06,
+      "loss": 0.0017,
+      "step": 4394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97529300487262e-06,
+      "loss": 0.0073,
+      "step": 4395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971491938309963e-06,
+      "loss": 0.077,
+      "step": 4396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.967690875866207e-06,
+      "loss": 0.001,
+      "step": 4397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963889818090539e-06,
+      "loss": 0.1076,
+      "step": 4398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960088765532147e-06,
+      "loss": 0.0077,
+      "step": 4399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956287718740217e-06,
+      "loss": 0.0197,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.952486678263925e-06,
+      "loss": 0.0256,
+      "step": 4401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948685644652459e-06,
+      "loss": 0.0064,
+      "step": 4402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.944884618454996e-06,
+      "loss": 0.0088,
+      "step": 4403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.941083600220722e-06,
+      "loss": 0.0052,
+      "step": 4404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937282590498812e-06,
+      "loss": 0.0718,
+      "step": 4405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.933481589838442e-06,
+      "loss": 0.0038,
+      "step": 4406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.929680598788793e-06,
+      "loss": 0.0007,
+      "step": 4407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925879617899041e-06,
+      "loss": 0.0022,
+      "step": 4408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922078647718352e-06,
+      "loss": 0.011,
+      "step": 4409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91827768879591e-06,
+      "loss": 0.0507,
+      "step": 4410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.914476741680872e-06,
+      "loss": 0.0625,
+      "step": 4411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.910675806922417e-06,
+      "loss": 0.0041,
+      "step": 4412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906874885069704e-06,
+      "loss": 0.0057,
+      "step": 4413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.903073976671907e-06,
+      "loss": 0.0058,
+      "step": 4414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.899273082278183e-06,
+      "loss": 0.0342,
+      "step": 4415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895472202437695e-06,
+      "loss": 0.0841,
+      "step": 4416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.891671337699603e-06,
+      "loss": 0.0016,
+      "step": 4417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88787048861306e-06,
+      "loss": 0.0004,
+      "step": 4418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.884069655727225e-06,
+      "loss": 0.0012,
+      "step": 4419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880268839591248e-06,
+      "loss": 0.0088,
+      "step": 4420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876468040754282e-06,
+      "loss": 0.0084,
+      "step": 4421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87266725976547e-06,
+      "loss": 0.002,
+      "step": 4422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.868866497173959e-06,
+      "loss": 0.0022,
+      "step": 4423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.865065753528895e-06,
+      "loss": 0.0067,
+      "step": 4424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86126502937941e-06,
+      "loss": 0.0173,
+      "step": 4425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857464325274643e-06,
+      "loss": 0.0073,
+      "step": 4426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85366364176373e-06,
+      "loss": 0.032,
+      "step": 4427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.849862979395799e-06,
+      "loss": 0.0036,
+      "step": 4428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84606233871998e-06,
+      "loss": 0.1202,
+      "step": 4429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842261720285396e-06,
+      "loss": 0.0371,
+      "step": 4430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83846112464117e-06,
+      "loss": 0.0693,
+      "step": 4431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.834660552336415e-06,
+      "loss": 0.055,
+      "step": 4432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830860003920253e-06,
+      "loss": 0.0064,
+      "step": 4433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82705947994179e-06,
+      "loss": 0.0094,
+      "step": 4434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.823258980950138e-06,
+      "loss": 0.0006,
+      "step": 4435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819458507494395e-06,
+      "loss": 0.0414,
+      "step": 4436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81565806012367e-06,
+      "loss": 0.0061,
+      "step": 4437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81185763938705e-06,
+      "loss": 0.0404,
+      "step": 4438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808057245833633e-06,
+      "loss": 0.0181,
+      "step": 4439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.804256880012507e-06,
+      "loss": 0.001,
+      "step": 4440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800456542472754e-06,
+      "loss": 0.0795,
+      "step": 4441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79665623376346e-06,
+      "loss": 0.0385,
+      "step": 4442
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792855954433697e-06,
+      "loss": 0.0015,
+      "step": 4443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789055705032544e-06,
+      "loss": 0.0019,
+      "step": 4444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78525548610906e-06,
+      "loss": 0.0047,
+      "step": 4445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.781455298212313e-06,
+      "loss": 0.0187,
+      "step": 4446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.777655141891361e-06,
+      "loss": 0.0014,
+      "step": 4447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.773855017695264e-06,
+      "loss": 0.0428,
+      "step": 4448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770054926173062e-06,
+      "loss": 0.0728,
+      "step": 4449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766254867873808e-06,
+      "loss": 0.0292,
+      "step": 4450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.762454843346544e-06,
+      "loss": 0.0402,
+      "step": 4451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758654853140296e-06,
+      "loss": 0.1769,
+      "step": 4452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.754854897804103e-06,
+      "loss": 0.0018,
+      "step": 4453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751054977886986e-06,
+      "loss": 0.0135,
+      "step": 4454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.747255093937966e-06,
+      "loss": 0.0662,
+      "step": 4455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.743455246506059e-06,
+      "loss": 0.0098,
+      "step": 4456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.739655436140278e-06,
+      "loss": 0.0047,
+      "step": 4457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73585566338962e-06,
+      "loss": 0.0011,
+      "step": 4458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73205592880309e-06,
+      "loss": 0.0012,
+      "step": 4459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72825623292968e-06,
+      "loss": 0.0233,
+      "step": 4460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724456576318383e-06,
+      "loss": 0.0332,
+      "step": 4461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720656959518172e-06,
+      "loss": 0.0694,
+      "step": 4462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716857383078027e-06,
+      "loss": 0.003,
+      "step": 4463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713057847546921e-06,
+      "loss": 0.0697,
+      "step": 4464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.709258353473822e-06,
+      "loss": 0.0541,
+      "step": 4465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70545890140768e-06,
+      "loss": 0.101,
+      "step": 4466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.701659491897455e-06,
+      "loss": 0.0452,
+      "step": 4467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69786012549209e-06,
+      "loss": 0.0064,
+      "step": 4468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.694060802740525e-06,
+      "loss": 0.0254,
+      "step": 4469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690261524191699e-06,
+      "loss": 0.0024,
+      "step": 4470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.686462290394533e-06,
+      "loss": 0.0331,
+      "step": 4471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682663101897952e-06,
+      "loss": 0.0266,
+      "step": 4472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678863959250869e-06,
+      "loss": 0.074,
+      "step": 4473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675064863002196e-06,
+      "loss": 0.0006,
+      "step": 4474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671265813700831e-06,
+      "loss": 0.0173,
+      "step": 4475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.667466811895668e-06,
+      "loss": 0.1432,
+      "step": 4476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663667858135596e-06,
+      "loss": 0.0187,
+      "step": 4477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6598689529695e-06,
+      "loss": 0.0131,
+      "step": 4478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656070096946246e-06,
+      "loss": 0.0497,
+      "step": 4479
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.652271290614707e-06,
+      "loss": 0.0089,
+      "step": 4480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.648472534523735e-06,
+      "loss": 0.0048,
+      "step": 4481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64467382922219e-06,
+      "loss": 0.1054,
+      "step": 4482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640875175258912e-06,
+      "loss": 0.0383,
+      "step": 4483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.637076573182742e-06,
+      "loss": 0.01,
+      "step": 4484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633278023542508e-06,
+      "loss": 0.0043,
+      "step": 4485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.629479526887032e-06,
+      "loss": 0.0842,
+      "step": 4486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625681083765131e-06,
+      "loss": 0.031,
+      "step": 4487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62188269472561e-06,
+      "loss": 0.0022,
+      "step": 4488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618084360317267e-06,
+      "loss": 0.008,
+      "step": 4489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.614286081088895e-06,
+      "loss": 0.041,
+      "step": 4490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61048785758928e-06,
+      "loss": 0.0609,
+      "step": 4491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.606689690367193e-06,
+      "loss": 0.0059,
+      "step": 4492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602891579971405e-06,
+      "loss": 0.0037,
+      "step": 4493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59909352695067e-06,
+      "loss": 0.0237,
+      "step": 4494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.595295531853742e-06,
+      "loss": 0.0196,
+      "step": 4495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59149759522936e-06,
+      "loss": 0.0039,
+      "step": 4496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.587699717626266e-06,
+      "loss": 0.0616,
+      "step": 4497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.583901899593176e-06,
+      "loss": 0.0735,
+      "step": 4498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58010414167881e-06,
+      "loss": 0.0083,
+      "step": 4499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576306444431877e-06,
+      "loss": 0.0162,
+      "step": 4500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572508808401078e-06,
+      "loss": 0.0102,
+      "step": 4501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.568711234135098e-06,
+      "loss": 0.0021,
+      "step": 4502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.564913722182621e-06,
+      "loss": 0.1503,
+      "step": 4503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561116273092322e-06,
+      "loss": 0.0369,
+      "step": 4504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557318887412865e-06,
+      "loss": 0.0008,
+      "step": 4505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553521565692902e-06,
+      "loss": 0.0704,
+      "step": 4506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549724308481073e-06,
+      "loss": 0.0293,
+      "step": 4507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54592711632602e-06,
+      "loss": 0.008,
+      "step": 4508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542129989776367e-06,
+      "loss": 0.1984,
+      "step": 4509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.538332929380734e-06,
+      "loss": 0.0149,
+      "step": 4510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534535935687723e-06,
+      "loss": 0.0058,
+      "step": 4511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.530739009245934e-06,
+      "loss": 0.0076,
+      "step": 4512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526942150603956e-06,
+      "loss": 0.0024,
+      "step": 4513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523145360310367e-06,
+      "loss": 0.0021,
+      "step": 4514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.519348638913732e-06,
+      "loss": 0.0033,
+      "step": 4515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515551986962614e-06,
+      "loss": 0.0298,
+      "step": 4516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511755405005557e-06,
+      "loss": 0.0437,
+      "step": 4517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507958893591104e-06,
+      "loss": 0.0277,
+      "step": 4518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.504162453267776e-06,
+      "loss": 0.105,
+      "step": 4519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500366084584102e-06,
+      "loss": 0.0623,
+      "step": 4520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.496569788088575e-06,
+      "loss": 0.0014,
+      "step": 4521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492773564329698e-06,
+      "loss": 0.0015,
+      "step": 4522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488977413855963e-06,
+      "loss": 0.0012,
+      "step": 4523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485181337215837e-06,
+      "loss": 0.0023,
+      "step": 4524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48138533495779e-06,
+      "loss": 0.0819,
+      "step": 4525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477589407630275e-06,
+      "loss": 0.0025,
+      "step": 4526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.473793555781737e-06,
+      "loss": 0.0046,
+      "step": 4527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469997779960605e-06,
+      "loss": 0.0172,
+      "step": 4528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466202080715304e-06,
+      "loss": 0.0252,
+      "step": 4529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462406458594245e-06,
+      "loss": 0.0442,
+      "step": 4530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458610914145826e-06,
+      "loss": 0.0128,
+      "step": 4531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454815447918435e-06,
+      "loss": 0.0411,
+      "step": 4532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451020060460446e-06,
+      "loss": 0.0236,
+      "step": 4533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447224752320234e-06,
+      "loss": 0.0079,
+      "step": 4534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.443429524046143e-06,
+      "loss": 0.0793,
+      "step": 4535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439634376186519e-06,
+      "loss": 0.0162,
+      "step": 4536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.435839309289695e-06,
+      "loss": 0.1553,
+      "step": 4537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432044323903985e-06,
+      "loss": 0.0678,
+      "step": 4538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.428249420577703e-06,
+      "loss": 0.0024,
+      "step": 4539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424454599859142e-06,
+      "loss": 0.1042,
+      "step": 4540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42065986229658e-06,
+      "loss": 0.0103,
+      "step": 4541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.416865208438297e-06,
+      "loss": 0.0368,
+      "step": 4542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413070638832548e-06,
+      "loss": 0.0573,
+      "step": 4543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409276154027584e-06,
+      "loss": 0.004,
+      "step": 4544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.405481754571635e-06,
+      "loss": 0.0174,
+      "step": 4545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401687441012928e-06,
+      "loss": 0.0007,
+      "step": 4546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397893213899671e-06,
+      "loss": 0.0016,
+      "step": 4547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.394099073780066e-06,
+      "loss": 0.0077,
+      "step": 4548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.390305021202291e-06,
+      "loss": 0.0716,
+      "step": 4549
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386511056714527e-06,
+      "loss": 0.0435,
+      "step": 4550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382717180864925e-06,
+      "loss": 0.1347,
+      "step": 4551
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378923394201638e-06,
+      "loss": 0.0024,
+      "step": 4552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.375129697272798e-06,
+      "loss": 0.0116,
+      "step": 4553
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37133609062653e-06,
+      "loss": 0.0073,
+      "step": 4554
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.367542574810937e-06,
+      "loss": 0.0081,
+      "step": 4555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363749150374115e-06,
+      "loss": 0.0179,
+      "step": 4556
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.359955817864148e-06,
+      "loss": 0.0362,
+      "step": 4557
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356162577829108e-06,
+      "loss": 0.0028,
+      "step": 4558
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.352369430817043e-06,
+      "loss": 0.0058,
+      "step": 4559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.348576377376e-06,
+      "loss": 0.1094,
+      "step": 4560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.344783418054005e-06,
+      "loss": 0.0176,
+      "step": 4561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.340990553399074e-06,
+      "loss": 0.0005,
+      "step": 4562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.337197783959207e-06,
+      "loss": 0.0824,
+      "step": 4563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33340511028239e-06,
+      "loss": 0.0498,
+      "step": 4564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.329612532916597e-06,
+      "loss": 0.0035,
+      "step": 4565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32582005240979e-06,
+      "loss": 0.0305,
+      "step": 4566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.322027669309913e-06,
+      "loss": 0.069,
+      "step": 4567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.318235384164896e-06,
+      "loss": 0.0762,
+      "step": 4568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314443197522658e-06,
+      "loss": 0.038,
+      "step": 4569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310651109931102e-06,
+      "loss": 0.1556,
+      "step": 4570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306859121938116e-06,
+      "loss": 0.0074,
+      "step": 4571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303067234091576e-06,
+      "loss": 0.0105,
+      "step": 4572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29927544693934e-06,
+      "loss": 0.0121,
+      "step": 4573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.295483761029253e-06,
+      "loss": 0.0331,
+      "step": 4574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29169217690915e-06,
+      "loss": 0.0615,
+      "step": 4575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.287900695126844e-06,
+      "loss": 0.0095,
+      "step": 4576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.284109316230133e-06,
+      "loss": 0.0336,
+      "step": 4577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280318040766805e-06,
+      "loss": 0.0948,
+      "step": 4578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276526869284636e-06,
+      "loss": 0.0145,
+      "step": 4579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.272735802331379e-06,
+      "loss": 0.0605,
+      "step": 4580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268944840454773e-06,
+      "loss": 0.0576,
+      "step": 4581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265153984202547e-06,
+      "loss": 0.1077,
+      "step": 4582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.261363234122412e-06,
+      "loss": 0.0041,
+      "step": 4583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257572590762065e-06,
+      "loss": 0.0622,
+      "step": 4584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253782054669182e-06,
+      "loss": 0.1584,
+      "step": 4585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249991626391429e-06,
+      "loss": 0.0455,
+      "step": 4586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246201306476455e-06,
+      "loss": 0.046,
+      "step": 4587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.242411095471897e-06,
+      "loss": 0.01,
+      "step": 4588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.238620993925368e-06,
+      "loss": 0.0042,
+      "step": 4589
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234831002384474e-06,
+      "loss": 0.0199,
+      "step": 4590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231041121396795e-06,
+      "loss": 0.002,
+      "step": 4591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.227251351509904e-06,
+      "loss": 0.0244,
+      "step": 4592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223461693271359e-06,
+      "loss": 0.0034,
+      "step": 4593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21967214722869e-06,
+      "loss": 0.0449,
+      "step": 4594
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.215882713929421e-06,
+      "loss": 0.0571,
+      "step": 4595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.212093393921061e-06,
+      "loss": 0.0032,
+      "step": 4596
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208304187751098e-06,
+      "loss": 0.0158,
+      "step": 4597
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.204515095967e-06,
+      "loss": 0.0228,
+      "step": 4598
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.200726119116227e-06,
+      "loss": 0.0297,
+      "step": 4599
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.196937257746216e-06,
+      "loss": 0.117,
+      "step": 4600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.193148512404394e-06,
+      "loss": 0.0326,
+      "step": 4601
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.189359883638162e-06,
+      "loss": 0.0084,
+      "step": 4602
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.185571371994915e-06,
+      "loss": 0.1317,
+      "step": 4603
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181782978022015e-06,
+      "loss": 0.0435,
+      "step": 4604
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.177994702266824e-06,
+      "loss": 0.0406,
+      "step": 4605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.174206545276678e-06,
+      "loss": 0.0226,
+      "step": 4606
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170418507598902e-06,
+      "loss": 0.0117,
+      "step": 4607
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166630589780793e-06,
+      "loss": 0.0048,
+      "step": 4608
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.162842792369642e-06,
+      "loss": 0.0083,
+      "step": 4609
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159055115912717e-06,
+      "loss": 0.0032,
+      "step": 4610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.155267560957268e-06,
+      "loss": 0.0522,
+      "step": 4611
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15148012805053e-06,
+      "loss": 0.0111,
+      "step": 4612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147692817739719e-06,
+      "loss": 0.0199,
+      "step": 4613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.143905630572034e-06,
+      "loss": 0.0159,
+      "step": 4614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140118567094655e-06,
+      "loss": 0.0094,
+      "step": 4615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136331627854745e-06,
+      "loss": 0.0023,
+      "step": 4616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132544813399454e-06,
+      "loss": 0.01,
+      "step": 4617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.128758124275898e-06,
+      "loss": 0.0039,
+      "step": 4618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.124971561031192e-06,
+      "loss": 0.0344,
+      "step": 4619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121185124212429e-06,
+      "loss": 0.0719,
+      "step": 4620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117398814366677e-06,
+      "loss": 0.0194,
+      "step": 4621
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11361263204099e-06,
+      "loss": 0.0016,
+      "step": 4622
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109826577782407e-06,
+      "loss": 0.077,
+      "step": 4623
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.106040652137944e-06,
+      "loss": 0.0227,
+      "step": 4624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102254855654598e-06,
+      "loss": 0.0792,
+      "step": 4625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098469188879348e-06,
+      "loss": 0.0017,
+      "step": 4626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.094683652359158e-06,
+      "loss": 0.0368,
+      "step": 4627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090898246640969e-06,
+      "loss": 0.0066,
+      "step": 4628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087112972271703e-06,
+      "loss": 0.001,
+      "step": 4629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.083327829798265e-06,
+      "loss": 0.0349,
+      "step": 4630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.079542819767544e-06,
+      "loss": 0.0091,
+      "step": 4631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0757579427264e-06,
+      "loss": 0.0012,
+      "step": 4632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071973199221687e-06,
+      "loss": 0.1115,
+      "step": 4633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068188589800223e-06,
+      "loss": 0.0145,
+      "step": 4634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.064404115008824e-06,
+      "loss": 0.001,
+      "step": 4635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060619775394276e-06,
+      "loss": 0.1608,
+      "step": 4636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.05683557150335e-06,
+      "loss": 0.0028,
+      "step": 4637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.053051503882794e-06,
+      "loss": 0.0594,
+      "step": 4638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049267573079339e-06,
+      "loss": 0.0424,
+      "step": 4639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.045483779639693e-06,
+      "loss": 0.0129,
+      "step": 4640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041700124110552e-06,
+      "loss": 0.0473,
+      "step": 4641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03791660703858e-06,
+      "loss": 0.0656,
+      "step": 4642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034133228970432e-06,
+      "loss": 0.0046,
+      "step": 4643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.030349990452735e-06,
+      "loss": 0.0401,
+      "step": 4644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.026566892032104e-06,
+      "loss": 0.0367,
+      "step": 4645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022783934255126e-06,
+      "loss": 0.007,
+      "step": 4646
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.019001117668368e-06,
+      "loss": 0.0007,
+      "step": 4647
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015218442818382e-06,
+      "loss": 0.0076,
+      "step": 4648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.011435910251694e-06,
+      "loss": 0.0015,
+      "step": 4649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007653520514819e-06,
+      "loss": 0.0012,
+      "step": 4650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003871274154237e-06,
+      "loss": 0.0663,
+      "step": 4651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000089171716422e-06,
+      "loss": 0.0226,
+      "step": 4652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996307213747812e-06,
+      "loss": 0.0075,
+      "step": 4653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.992525400794841e-06,
+      "loss": 0.1721,
+      "step": 4654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988743733403904e-06,
+      "loss": 0.0614,
+      "step": 4655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984962212121391e-06,
+      "loss": 0.0429,
+      "step": 4656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981180837493662e-06,
+      "loss": 0.0007,
+      "step": 4657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97739961006706e-06,
+      "loss": 0.0127,
+      "step": 4658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973618530387902e-06,
+      "loss": 0.0204,
+      "step": 4659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969837599002488e-06,
+      "loss": 0.0483,
+      "step": 4660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96605681645709e-06,
+      "loss": 0.0185,
+      "step": 4661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962276183297969e-06,
+      "loss": 0.0034,
+      "step": 4662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.958495700071359e-06,
+      "loss": 0.0051,
+      "step": 4663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.0593,
+      "step": 4664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.950935185600487e-06,
+      "loss": 0.0383,
+      "step": 4665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.947155155448588e-06,
+      "loss": 0.0181,
+      "step": 4666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.943375277413917e-06,
+      "loss": 0.0634,
+      "step": 4667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939595552042595e-06,
+      "loss": 0.0037,
+      "step": 4668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.935815979880727e-06,
+      "loss": 0.0048,
+      "step": 4669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.932036561474394e-06,
+      "loss": 0.0043,
+      "step": 4670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.928257297369657e-06,
+      "loss": 0.0276,
+      "step": 4671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924478188112547e-06,
+      "loss": 0.0785,
+      "step": 4672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92069923424908e-06,
+      "loss": 0.002,
+      "step": 4673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916920436325246e-06,
+      "loss": 0.0641,
+      "step": 4674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913141794887012e-06,
+      "loss": 0.0652,
+      "step": 4675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.909363310480329e-06,
+      "loss": 0.0403,
+      "step": 4676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.905584983651118e-06,
+      "loss": 0.0031,
+      "step": 4677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.901806814945276e-06,
+      "loss": 0.0619,
+      "step": 4678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.898028804908684e-06,
+      "loss": 0.0051,
+      "step": 4679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.894250954087199e-06,
+      "loss": 0.0061,
+      "step": 4680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.890473263026653e-06,
+      "loss": 0.0065,
+      "step": 4681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886695732272848e-06,
+      "loss": 0.0069,
+      "step": 4682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.882918362371573e-06,
+      "loss": 0.0171,
+      "step": 4683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879141153868595e-06,
+      "loss": 0.0034,
+      "step": 4684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.875364107309647e-06,
+      "loss": 0.0019,
+      "step": 4685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87158722324045e-06,
+      "loss": 0.0358,
+      "step": 4686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867810502206689e-06,
+      "loss": 0.0045,
+      "step": 4687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864033944754036e-06,
+      "loss": 0.0236,
+      "step": 4688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860257551428137e-06,
+      "loss": 0.0061,
+      "step": 4689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856481322774614e-06,
+      "loss": 0.0017,
+      "step": 4690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85270525933906e-06,
+      "loss": 0.1553,
+      "step": 4691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848929361667053e-06,
+      "loss": 0.0008,
+      "step": 4692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84515363030414e-06,
+      "loss": 0.0917,
+      "step": 4693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.841378065795849e-06,
+      "loss": 0.0064,
+      "step": 4694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837602668687679e-06,
+      "loss": 0.0355,
+      "step": 4695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.833827439525109e-06,
+      "loss": 0.0214,
+      "step": 4696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830052378853591e-06,
+      "loss": 0.1788,
+      "step": 4697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.826277487218555e-06,
+      "loss": 0.0422,
+      "step": 4698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.822502765165403e-06,
+      "loss": 0.0034,
+      "step": 4699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81872821323952e-06,
+      "loss": 0.1081,
+      "step": 4700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.814953831986253e-06,
+      "loss": 0.0062,
+      "step": 4701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811179621950937e-06,
+      "loss": 0.0305,
+      "step": 4702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80740558367888e-06,
+      "loss": 0.0298,
+      "step": 4703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.803631717715358e-06,
+      "loss": 0.019,
+      "step": 4704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799858024605628e-06,
+      "loss": 0.0128,
+      "step": 4705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796084504894923e-06,
+      "loss": 0.0588,
+      "step": 4706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.792311159128452e-06,
+      "loss": 0.022,
+      "step": 4707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.788537987851389e-06,
+      "loss": 0.002,
+      "step": 4708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784764991608894e-06,
+      "loss": 0.0011,
+      "step": 4709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.780992170946095e-06,
+      "loss": 0.0328,
+      "step": 4710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777219526408103e-06,
+      "loss": 0.0127,
+      "step": 4711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77344705853999e-06,
+      "loss": 0.0065,
+      "step": 4712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769674767886813e-06,
+      "loss": 0.0317,
+      "step": 4713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765902654993604e-06,
+      "loss": 0.0012,
+      "step": 4714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76213072040536e-06,
+      "loss": 0.0286,
+      "step": 4715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.758358964667061e-06,
+      "loss": 0.1216,
+      "step": 4716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.754587388323656e-06,
+      "loss": 0.0125,
+      "step": 4717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.750815991920071e-06,
+      "loss": 0.0009,
+      "step": 4718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747044776001208e-06,
+      "loss": 0.0269,
+      "step": 4719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743273741111939e-06,
+      "loss": 0.0068,
+      "step": 4720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739502887797108e-06,
+      "loss": 0.0174,
+      "step": 4721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.735732216601538e-06,
+      "loss": 0.0197,
+      "step": 4722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731961728070025e-06,
+      "loss": 0.0124,
+      "step": 4723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728191422747335e-06,
+      "loss": 0.0043,
+      "step": 4724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724421301178209e-06,
+      "loss": 0.001,
+      "step": 4725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.720651363907364e-06,
+      "loss": 0.0077,
+      "step": 4726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.716881611479487e-06,
+      "loss": 0.0525,
+      "step": 4727
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.713112044439246e-06,
+      "loss": 0.0564,
+      "step": 4728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709342663331264e-06,
+      "loss": 0.0049,
+      "step": 4729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70557346870016e-06,
+      "loss": 0.0041,
+      "step": 4730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701804461090506e-06,
+      "loss": 0.1136,
+      "step": 4731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698035641046862e-06,
+      "loss": 0.1139,
+      "step": 4732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694267009113753e-06,
+      "loss": 0.0005,
+      "step": 4733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.690498565835682e-06,
+      "loss": 0.0007,
+      "step": 4734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.686730311757115e-06,
+      "loss": 0.0029,
+      "step": 4735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.682962247422502e-06,
+      "loss": 0.0881,
+      "step": 4736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.679194373376264e-06,
+      "loss": 0.0024,
+      "step": 4737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675426690162784e-06,
+      "loss": 0.002,
+      "step": 4738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.671659198326429e-06,
+      "loss": 0.0058,
+      "step": 4739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66789189841153e-06,
+      "loss": 0.0887,
+      "step": 4740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.664124790962407e-06,
+      "loss": 0.088,
+      "step": 4741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.660357876523323e-06,
+      "loss": 0.0009,
+      "step": 4742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65659115563854e-06,
+      "loss": 0.0717,
+      "step": 4743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652824628852275e-06,
+      "loss": 0.0338,
+      "step": 4744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64905829670873e-06,
+      "loss": 0.0203,
+      "step": 4745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645292159752067e-06,
+      "loss": 0.0067,
+      "step": 4746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641526218526431e-06,
+      "loss": 0.0415,
+      "step": 4747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.637760473575929e-06,
+      "loss": 0.0407,
+      "step": 4748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.633994925444645e-06,
+      "loss": 0.0006,
+      "step": 4749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.630229574676632e-06,
+      "loss": 0.0482,
+      "step": 4750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.626464421815919e-06,
+      "loss": 0.0726,
+      "step": 4751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6226994674065e-06,
+      "loss": 0.0457,
+      "step": 4752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618934711992346e-06,
+      "loss": 0.1236,
+      "step": 4753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.615170156117394e-06,
+      "loss": 0.0177,
+      "step": 4754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611405800325559e-06,
+      "loss": 0.0004,
+      "step": 4755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607641645160721e-06,
+      "loss": 0.004,
+      "step": 4756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.603877691166728e-06,
+      "loss": 0.0573,
+      "step": 4757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600113938887409e-06,
+      "loss": 0.0007,
+      "step": 4758
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.596350388866557e-06,
+      "loss": 0.002,
+      "step": 4759
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592587041647943e-06,
+      "loss": 0.0293,
+      "step": 4760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588823897775296e-06,
+      "loss": 0.0551,
+      "step": 4761
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.585060957792326e-06,
+      "loss": 0.0072,
+      "step": 4762
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58129822224271e-06,
+      "loss": 0.0029,
+      "step": 4763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577535691670098e-06,
+      "loss": 0.003,
+      "step": 4764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573773366618106e-06,
+      "loss": 0.0132,
+      "step": 4765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57001124763032e-06,
+      "loss": 0.0011,
+      "step": 4766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.566249335250304e-06,
+      "loss": 0.075,
+      "step": 4767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.562487630021587e-06,
+      "loss": 0.0986,
+      "step": 4768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558726132487668e-06,
+      "loss": 0.0394,
+      "step": 4769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55496484319201e-06,
+      "loss": 0.0161,
+      "step": 4770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.551203762678056e-06,
+      "loss": 0.0669,
+      "step": 4771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547442891489214e-06,
+      "loss": 0.0427,
+      "step": 4772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.543682230168866e-06,
+      "loss": 0.0061,
+      "step": 4773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539921779260356e-06,
+      "loss": 0.0093,
+      "step": 4774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.536161539307e-06,
+      "loss": 0.0109,
+      "step": 4775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.532401510852092e-06,
+      "loss": 0.0031,
+      "step": 4776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528641694438887e-06,
+      "loss": 0.0024,
+      "step": 4777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.524882090610607e-06,
+      "loss": 0.0026,
+      "step": 4778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521122699910449e-06,
+      "loss": 0.0006,
+      "step": 4779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.51736352288158e-06,
+      "loss": 0.0888,
+      "step": 4780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513604560067135e-06,
+      "loss": 0.0469,
+      "step": 4781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50984581201021e-06,
+      "loss": 0.1093,
+      "step": 4782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506087279253887e-06,
+      "loss": 0.0218,
+      "step": 4783
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502328962341196e-06,
+      "loss": 0.0013,
+      "step": 4784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.498570861815151e-06,
+      "loss": 0.0051,
+      "step": 4785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.494812978218732e-06,
+      "loss": 0.0405,
+      "step": 4786
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491055312094883e-06,
+      "loss": 0.1089,
+      "step": 4787
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48729786398652e-06,
+      "loss": 0.0239,
+      "step": 4788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.483540634436528e-06,
+      "loss": 0.0776,
+      "step": 4789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.479783623987762e-06,
+      "loss": 0.0158,
+      "step": 4790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476026833183036e-06,
+      "loss": 0.0009,
+      "step": 4791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472270262565145e-06,
+      "loss": 0.1299,
+      "step": 4792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.468513912676843e-06,
+      "loss": 0.1592,
+      "step": 4793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.464757784060857e-06,
+      "loss": 0.0053,
+      "step": 4794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46100187725988e-06,
+      "loss": 0.0027,
+      "step": 4795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.457246192816571e-06,
+      "loss": 0.0531,
+      "step": 4796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.453490731273566e-06,
+      "loss": 0.0071,
+      "step": 4797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.449735493173455e-06,
+      "loss": 0.033,
+      "step": 4798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445980479058801e-06,
+      "loss": 0.0135,
+      "step": 4799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442225689472144e-06,
+      "loss": 0.0076,
+      "step": 4800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438471124955979e-06,
+      "loss": 0.0046,
+      "step": 4801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.434716786052771e-06,
+      "loss": 0.1185,
+      "step": 4802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430962673304959e-06,
+      "loss": 0.0022,
+      "step": 4803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.427208787254947e-06,
+      "loss": 0.0017,
+      "step": 4804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4234551284451e-06,
+      "loss": 0.0016,
+      "step": 4805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.419701697417753e-06,
+      "loss": 0.0017,
+      "step": 4806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415948494715215e-06,
+      "loss": 0.0131,
+      "step": 4807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.412195520879753e-06,
+      "loss": 0.0269,
+      "step": 4808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.408442776453606e-06,
+      "loss": 0.0096,
+      "step": 4809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.404690261978976e-06,
+      "loss": 0.1217,
+      "step": 4810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.400937977998043e-06,
+      "loss": 0.0517,
+      "step": 4811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39718592505293e-06,
+      "loss": 0.0171,
+      "step": 4812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.393434103685755e-06,
+      "loss": 0.125,
+      "step": 4813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.389682514438579e-06,
+      "loss": 0.0815,
+      "step": 4814
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385931157853445e-06,
+      "loss": 0.0032,
+      "step": 4815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.382180034472353e-06,
+      "loss": 0.0107,
+      "step": 4816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37842914483728e-06,
+      "loss": 0.0267,
+      "step": 4817
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374678489490155e-06,
+      "loss": 0.0677,
+      "step": 4818
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.370928068972885e-06,
+      "loss": 0.1247,
+      "step": 4819
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.367177883827336e-06,
+      "loss": 0.0422,
+      "step": 4820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.363427934595347e-06,
+      "loss": 0.0089,
+      "step": 4821
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359678221818712e-06,
+      "loss": 0.0006,
+      "step": 4822
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355928746039202e-06,
+      "loss": 0.0038,
+      "step": 4823
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.352179507798551e-06,
+      "loss": 0.0087,
+      "step": 4824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34843050763845e-06,
+      "loss": 0.0326,
+      "step": 4825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344681746100569e-06,
+      "loss": 0.0258,
+      "step": 4826
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34093322372653e-06,
+      "loss": 0.0077,
+      "step": 4827
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337184941057933e-06,
+      "loss": 0.0264,
+      "step": 4828
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333436898636333e-06,
+      "loss": 0.0046,
+      "step": 4829
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329689097003263e-06,
+      "loss": 0.055,
+      "step": 4830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325941536700206e-06,
+      "loss": 0.038,
+      "step": 4831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.322194218268618e-06,
+      "loss": 0.0107,
+      "step": 4832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318447142249922e-06,
+      "loss": 0.0711,
+      "step": 4833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314700309185503e-06,
+      "loss": 0.0565,
+      "step": 4834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310953719616712e-06,
+      "loss": 0.0112,
+      "step": 4835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30720737408486e-06,
+      "loss": 0.0398,
+      "step": 4836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30346127313123e-06,
+      "loss": 0.0003,
+      "step": 4837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299715417297072e-06,
+      "loss": 0.0594,
+      "step": 4838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.295969807123586e-06,
+      "loss": 0.0138,
+      "step": 4839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.292224443151947e-06,
+      "loss": 0.0333,
+      "step": 4840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.288479325923294e-06,
+      "loss": 0.0079,
+      "step": 4841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284734455978732e-06,
+      "loss": 0.0521,
+      "step": 4842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280989833859326e-06,
+      "loss": 0.0455,
+      "step": 4843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277245460106106e-06,
+      "loss": 0.0117,
+      "step": 4844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.273501335260066e-06,
+      "loss": 0.0013,
+      "step": 4845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269757459862169e-06,
+      "loss": 0.0009,
+      "step": 4846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266013834453335e-06,
+      "loss": 0.0645,
+      "step": 4847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262270459574451e-06,
+      "loss": 0.0631,
+      "step": 4848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258527335766367e-06,
+      "loss": 0.0294,
+      "step": 4849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254784463569901e-06,
+      "loss": 0.0051,
+      "step": 4850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.251041843525828e-06,
+      "loss": 0.0044,
+      "step": 4851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247299476174891e-06,
+      "loss": 0.0351,
+      "step": 4852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.243557362057793e-06,
+      "loss": 0.0535,
+      "step": 4853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.239815501715202e-06,
+      "loss": 0.004,
+      "step": 4854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.236073895687753e-06,
+      "loss": 0.0791,
+      "step": 4855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232332544516038e-06,
+      "loss": 0.004,
+      "step": 4856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22859144874062e-06,
+      "loss": 0.0034,
+      "step": 4857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.224850608902014e-06,
+      "loss": 0.0029,
+      "step": 4858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221110025540707e-06,
+      "loss": 0.0007,
+      "step": 4859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21736969919715e-06,
+      "loss": 0.0107,
+      "step": 4860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213629630411747e-06,
+      "loss": 0.0011,
+      "step": 4861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.209889819724875e-06,
+      "loss": 0.069,
+      "step": 4862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.206150267676866e-06,
+      "loss": 0.0296,
+      "step": 4863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202410974808027e-06,
+      "loss": 0.1616,
+      "step": 4864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.198671941658607e-06,
+      "loss": 0.0622,
+      "step": 4865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194933168768842e-06,
+      "loss": 0.0022,
+      "step": 4866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191194656678905e-06,
+      "loss": 0.0079,
+      "step": 4867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18745640592895e-06,
+      "loss": 0.0091,
+      "step": 4868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183718417059086e-06,
+      "loss": 0.0065,
+      "step": 4869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17998069060939e-06,
+      "loss": 0.0389,
+      "step": 4870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176243227119892e-06,
+      "loss": 0.0022,
+      "step": 4871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172506027130589e-06,
+      "loss": 0.0181,
+      "step": 4872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168769091181439e-06,
+      "loss": 0.006,
+      "step": 4873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165032419812366e-06,
+      "loss": 0.0214,
+      "step": 4874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161296013563249e-06,
+      "loss": 0.0483,
+      "step": 4875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157559872973932e-06,
+      "loss": 0.0029,
+      "step": 4876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153823998584222e-06,
+      "loss": 0.0309,
+      "step": 4877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150088390933883e-06,
+      "loss": 0.0013,
+      "step": 4878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.146353050562644e-06,
+      "loss": 0.0037,
+      "step": 4879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142617978010203e-06,
+      "loss": 0.0361,
+      "step": 4880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138883173816197e-06,
+      "loss": 0.0016,
+      "step": 4881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.135148638520247e-06,
+      "loss": 0.0007,
+      "step": 4882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131414372661925e-06,
+      "loss": 0.037,
+      "step": 4883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.127680376780765e-06,
+      "loss": 0.0411,
+      "step": 4884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.123946651416263e-06,
+      "loss": 0.166,
+      "step": 4885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.120213197107877e-06,
+      "loss": 0.0594,
+      "step": 4886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.116480014395021e-06,
+      "loss": 0.0386,
+      "step": 4887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112747103817076e-06,
+      "loss": 0.0064,
+      "step": 4888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109014465913378e-06,
+      "loss": 0.0525,
+      "step": 4889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.105282101223228e-06,
+      "loss": 0.1026,
+      "step": 4890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.101550010285887e-06,
+      "loss": 0.0138,
+      "step": 4891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097818193640571e-06,
+      "loss": 0.0009,
+      "step": 4892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094086651826472e-06,
+      "loss": 0.0379,
+      "step": 4893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090355385382715e-06,
+      "loss": 0.0009,
+      "step": 4894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08662439484841e-06,
+      "loss": 0.0585,
+      "step": 4895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082893680762619e-06,
+      "loss": 0.0016,
+      "step": 4896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07916324366436e-06,
+      "loss": 0.0019,
+      "step": 4897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.075433084092614e-06,
+      "loss": 0.0044,
+      "step": 4898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071703202586326e-06,
+      "loss": 0.0017,
+      "step": 4899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.067973599684398e-06,
+      "loss": 0.0035,
+      "step": 4900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064244275925684e-06,
+      "loss": 0.0114,
+      "step": 4901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060515231849009e-06,
+      "loss": 0.0465,
+      "step": 4902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.056786467993154e-06,
+      "loss": 0.0046,
+      "step": 4903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053057984896858e-06,
+      "loss": 0.0522,
+      "step": 4904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049329783098818e-06,
+      "loss": 0.0244,
+      "step": 4905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045601863137694e-06,
+      "loss": 0.0395,
+      "step": 4906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.041874225552106e-06,
+      "loss": 0.0372,
+      "step": 4907
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.038146870880626e-06,
+      "loss": 0.132,
+      "step": 4908
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034419799661792e-06,
+      "loss": 0.0049,
+      "step": 4909
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030693012434098e-06,
+      "loss": 0.015,
+      "step": 4910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026966509736001e-06,
+      "loss": 0.0838,
+      "step": 4911
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023240292105911e-06,
+      "loss": 0.113,
+      "step": 4912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.019514360082203e-06,
+      "loss": 0.0103,
+      "step": 4913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.015788714203205e-06,
+      "loss": 0.0039,
+      "step": 4914
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012063355007205e-06,
+      "loss": 0.0039,
+      "step": 4915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008338283032452e-06,
+      "loss": 0.0594,
+      "step": 4916
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.004613498817156e-06,
+      "loss": 0.0109,
+      "step": 4917
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.000889002899475e-06,
+      "loss": 0.0439,
+      "step": 4918
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997164795817534e-06,
+      "loss": 0.0043,
+      "step": 4919
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.993440878109417e-06,
+      "loss": 0.021,
+      "step": 4920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989717250313165e-06,
+      "loss": 0.0031,
+      "step": 4921
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.985993912966768e-06,
+      "loss": 0.1602,
+      "step": 4922
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.982270866608188e-06,
+      "loss": 0.0025,
+      "step": 4923
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978548111775335e-06,
+      "loss": 0.0019,
+      "step": 4924
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.974825649006082e-06,
+      "loss": 0.0148,
+      "step": 4925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.971103478838256e-06,
+      "loss": 0.0019,
+      "step": 4926
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967381601809649e-06,
+      "loss": 0.0158,
+      "step": 4927
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963660018458e-06,
+      "loss": 0.0495,
+      "step": 4928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.959938729321014e-06,
+      "loss": 0.0405,
+      "step": 4929
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.956217734936353e-06,
+      "loss": 0.0656,
+      "step": 4930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.952497035841629e-06,
+      "loss": 0.0325,
+      "step": 4931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948776632574417e-06,
+      "loss": 0.0038,
+      "step": 4932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.945056525672253e-06,
+      "loss": 0.0094,
+      "step": 4933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941336715672623e-06,
+      "loss": 0.0102,
+      "step": 4934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937617203112975e-06,
+      "loss": 0.0961,
+      "step": 4935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93389798853071e-06,
+      "loss": 0.0008,
+      "step": 4936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.930179072463184e-06,
+      "loss": 0.0843,
+      "step": 4937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926460455447718e-06,
+      "loss": 0.0028,
+      "step": 4938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922742138021586e-06,
+      "loss": 0.0398,
+      "step": 4939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919024120722021e-06,
+      "loss": 0.0869,
+      "step": 4940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.915306404086202e-06,
+      "loss": 0.0516,
+      "step": 4941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91158898865128e-06,
+      "loss": 0.0344,
+      "step": 4942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90787187495435e-06,
+      "loss": 0.0009,
+      "step": 4943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.904155063532472e-06,
+      "loss": 0.0023,
+      "step": 4944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900438554922657e-06,
+      "loss": 0.0073,
+      "step": 4945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.896722349661872e-06,
+      "loss": 0.0711,
+      "step": 4946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.893006448287046e-06,
+      "loss": 0.0223,
+      "step": 4947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889290851335058e-06,
+      "loss": 0.008,
+      "step": 4948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.885575559342751e-06,
+      "loss": 0.0811,
+      "step": 4949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881860572846905e-06,
+      "loss": 0.1228,
+      "step": 4950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.878145892384279e-06,
+      "loss": 0.0129,
+      "step": 4951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874431518491574e-06,
+      "loss": 0.0175,
+      "step": 4952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.870717451705455e-06,
+      "loss": 0.0375,
+      "step": 4953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.867003692562533e-06,
+      "loss": 0.0489,
+      "step": 4954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.863290241599383e-06,
+      "loss": 0.0156,
+      "step": 4955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.859577099352528e-06,
+      "loss": 0.0061,
+      "step": 4956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.855864266358459e-06,
+      "loss": 0.1045,
+      "step": 4957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852151743153606e-06,
+      "loss": 0.0762,
+      "step": 4958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848439530274363e-06,
+      "loss": 0.005,
+      "step": 4959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84472762825708e-06,
+      "loss": 0.0431,
+      "step": 4960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841016037638064e-06,
+      "loss": 0.0708,
+      "step": 4961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.837304758953565e-06,
+      "loss": 0.0037,
+      "step": 4962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.833593792739807e-06,
+      "loss": 0.041,
+      "step": 4963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829883139532948e-06,
+      "loss": 0.0318,
+      "step": 4964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826172799869116e-06,
+      "loss": 0.0231,
+      "step": 4965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822462774284389e-06,
+      "loss": 0.0038,
+      "step": 4966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818753063314795e-06,
+      "loss": 0.1026,
+      "step": 4967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815043667496323e-06,
+      "loss": 0.0087,
+      "step": 4968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.811334587364916e-06,
+      "loss": 0.2346,
+      "step": 4969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807625823456473e-06,
+      "loss": 0.0121,
+      "step": 4970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.803917376306836e-06,
+      "loss": 0.0016,
+      "step": 4971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800209246451812e-06,
+      "loss": 0.0249,
+      "step": 4972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796501434427161e-06,
+      "loss": 0.0729,
+      "step": 4973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792793940768597e-06,
+      "loss": 0.0009,
+      "step": 4974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789086766011782e-06,
+      "loss": 0.0379,
+      "step": 4975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.785379910692343e-06,
+      "loss": 0.0101,
+      "step": 4976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.781673375345844e-06,
+      "loss": 0.004,
+      "step": 4977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77796716050782e-06,
+      "loss": 0.0074,
+      "step": 4978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77426126671375e-06,
+      "loss": 0.1301,
+      "step": 4979
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.770555694499075e-06,
+      "loss": 0.1013,
+      "step": 4980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766850444399176e-06,
+      "loss": 0.0394,
+      "step": 4981
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.7631455169494e-06,
+      "loss": 0.0019,
+      "step": 4982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.759440912685043e-06,
+      "loss": 0.0364,
+      "step": 4983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75573663214135e-06,
+      "loss": 0.0024,
+      "step": 4984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.752032675853525e-06,
+      "loss": 0.0445,
+      "step": 4985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748329044356724e-06,
+      "loss": 0.0341,
+      "step": 4986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74462573818606e-06,
+      "loss": 0.0119,
+      "step": 4987
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740922757876586e-06,
+      "loss": 0.0058,
+      "step": 4988
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73722010396332e-06,
+      "loss": 0.0452,
+      "step": 4989
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733517776981235e-06,
+      "loss": 0.037,
+      "step": 4990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72981577746524e-06,
+      "loss": 0.0003,
+      "step": 4991
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726114105950212e-06,
+      "loss": 0.0082,
+      "step": 4992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72241276297098e-06,
+      "loss": 0.0077,
+      "step": 4993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.718711749062318e-06,
+      "loss": 0.012,
+      "step": 4994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.715011064758957e-06,
+      "loss": 0.0089,
+      "step": 4995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.711310710595578e-06,
+      "loss": 0.0112,
+      "step": 4996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707610687106823e-06,
+      "loss": 0.0025,
+      "step": 4997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.703910994827269e-06,
+      "loss": 0.0006,
+      "step": 4998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700211634291462e-06,
+      "loss": 0.0079,
+      "step": 4999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69651260603389e-06,
+      "loss": 0.0025,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692813910589001e-06,
+      "loss": 0.0008,
+      "step": 5001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689115548491186e-06,
+      "loss": 0.004,
+      "step": 5002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685417520274793e-06,
+      "loss": 0.0048,
+      "step": 5003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681719826474124e-06,
+      "loss": 0.0056,
+      "step": 5004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.678022467623424e-06,
+      "loss": 0.0034,
+      "step": 5005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.674325444256899e-06,
+      "loss": 0.0045,
+      "step": 5006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670628756908698e-06,
+      "loss": 0.0295,
+      "step": 5007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.666932406112932e-06,
+      "loss": 0.0812,
+      "step": 5008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.663236392403654e-06,
+      "loss": 0.0838,
+      "step": 5009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659540716314877e-06,
+      "loss": 0.0279,
+      "step": 5010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655845378380551e-06,
+      "loss": 0.0245,
+      "step": 5011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652150379134593e-06,
+      "loss": 0.0027,
+      "step": 5012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64845571911086e-06,
+      "loss": 0.0424,
+      "step": 5013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64476139884317e-06,
+      "loss": 0.0021,
+      "step": 5014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64106741886528e-06,
+      "loss": 0.0295,
+      "step": 5015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637373779710908e-06,
+      "loss": 0.009,
+      "step": 5016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.633680481913717e-06,
+      "loss": 0.0045,
+      "step": 5017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.629987526007326e-06,
+      "loss": 0.0021,
+      "step": 5018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.626294912525296e-06,
+      "loss": 0.1074,
+      "step": 5019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.622602642001143e-06,
+      "loss": 0.0591,
+      "step": 5020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.618910714968336e-06,
+      "loss": 0.0013,
+      "step": 5021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.615219131960293e-06,
+      "loss": 0.1577,
+      "step": 5022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611527893510384e-06,
+      "loss": 0.0016,
+      "step": 5023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607837000151921e-06,
+      "loss": 0.0064,
+      "step": 5024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604146452418175e-06,
+      "loss": 0.0288,
+      "step": 5025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.600456250842364e-06,
+      "loss": 0.0068,
+      "step": 5026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.596766395957659e-06,
+      "loss": 0.0008,
+      "step": 5027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.593076888297172e-06,
+      "loss": 0.0792,
+      "step": 5028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5893877283939754e-06,
+      "loss": 0.073,
+      "step": 5029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585698916781084e-06,
+      "loss": 0.0034,
+      "step": 5030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582010453991468e-06,
+      "loss": 0.069,
+      "step": 5031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.578322340558044e-06,
+      "loss": 0.0035,
+      "step": 5032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.574634577013675e-06,
+      "loss": 0.0792,
+      "step": 5033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.570947163891178e-06,
+      "loss": 0.0033,
+      "step": 5034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567260101723318e-06,
+      "loss": 0.0445,
+      "step": 5035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.563573391042813e-06,
+      "loss": 0.0352,
+      "step": 5036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5598870323823225e-06,
+      "loss": 0.0118,
+      "step": 5037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556201026274459e-06,
+      "loss": 0.1599,
+      "step": 5038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.552515373251787e-06,
+      "loss": 0.0023,
+      "step": 5039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.548830073846819e-06,
+      "loss": 0.017,
+      "step": 5040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545145128592009e-06,
+      "loss": 0.0757,
+      "step": 5041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54146053801977e-06,
+      "loss": 0.0554,
+      "step": 5042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.537776302662458e-06,
+      "loss": 0.0603,
+      "step": 5043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534092423052382e-06,
+      "loss": 0.0438,
+      "step": 5044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530408899721791e-06,
+      "loss": 0.0422,
+      "step": 5045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.526725733202894e-06,
+      "loss": 0.0852,
+      "step": 5046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523042924027837e-06,
+      "loss": 0.1025,
+      "step": 5047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.519360472728723e-06,
+      "loss": 0.0264,
+      "step": 5048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515678379837599e-06,
+      "loss": 0.0069,
+      "step": 5049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.511996645886464e-06,
+      "loss": 0.0515,
+      "step": 5050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.508315271407261e-06,
+      "loss": 0.0077,
+      "step": 5051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.50463425693188e-06,
+      "loss": 0.0013,
+      "step": 5052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5009536029921685e-06,
+      "loss": 0.0056,
+      "step": 5053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497273310119908e-06,
+      "loss": 0.0038,
+      "step": 5054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.493593378846837e-06,
+      "loss": 0.1211,
+      "step": 5055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489913809704643e-06,
+      "loss": 0.0458,
+      "step": 5056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486234603224956e-06,
+      "loss": 0.0404,
+      "step": 5057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.482555759939353e-06,
+      "loss": 0.0171,
+      "step": 5058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478877280379366e-06,
+      "loss": 0.081,
+      "step": 5059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4751991650764645e-06,
+      "loss": 0.0011,
+      "step": 5060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471521414562071e-06,
+      "loss": 0.0153,
+      "step": 5061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.467844029367555e-06,
+      "loss": 0.0301,
+      "step": 5062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.464167010024236e-06,
+      "loss": 0.0009,
+      "step": 5063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.460490357063372e-06,
+      "loss": 0.0176,
+      "step": 5064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.456814071016178e-06,
+      "loss": 0.016,
+      "step": 5065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.453138152413808e-06,
+      "loss": 0.0293,
+      "step": 5066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449462601787372e-06,
+      "loss": 0.0147,
+      "step": 5067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.445787419667916e-06,
+      "loss": 0.0629,
+      "step": 5068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.442112606586438e-06,
+      "loss": 0.0041,
+      "step": 5069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.438438163073884e-06,
+      "loss": 0.1085,
+      "step": 5070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434764089661151e-06,
+      "loss": 0.008,
+      "step": 5071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.431090386879066e-06,
+      "loss": 0.032,
+      "step": 5072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.427417055258424e-06,
+      "loss": 0.1205,
+      "step": 5073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423744095329946e-06,
+      "loss": 0.0085,
+      "step": 5074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.420071507624314e-06,
+      "loss": 0.0236,
+      "step": 5075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4163992926721505e-06,
+      "loss": 0.0097,
+      "step": 5076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.412727451004025e-06,
+      "loss": 0.0499,
+      "step": 5077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.409055983150451e-06,
+      "loss": 0.0274,
+      "step": 5078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.405384889641891e-06,
+      "loss": 0.0214,
+      "step": 5079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.401714171008756e-06,
+      "loss": 0.0042,
+      "step": 5080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.398043827781394e-06,
+      "loss": 0.0534,
+      "step": 5081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394373860490105e-06,
+      "loss": 0.0021,
+      "step": 5082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.390704269665135e-06,
+      "loss": 0.052,
+      "step": 5083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.387035055836673e-06,
+      "loss": 0.0041,
+      "step": 5084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3833662195348554e-06,
+      "loss": 0.0022,
+      "step": 5085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.379697761289763e-06,
+      "loss": 0.0463,
+      "step": 5086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3760296816314255e-06,
+      "loss": 0.0054,
+      "step": 5087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372361981089808e-06,
+      "loss": 0.0355,
+      "step": 5088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.368694660194833e-06,
+      "loss": 0.0026,
+      "step": 5089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3650277194763595e-06,
+      "loss": 0.0053,
+      "step": 5090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.361361159464195e-06,
+      "loss": 0.0317,
+      "step": 5091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357694980688095e-06,
+      "loss": 0.0709,
+      "step": 5092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354029183677754e-06,
+      "loss": 0.0753,
+      "step": 5093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350363768962814e-06,
+      "loss": 0.0014,
+      "step": 5094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346698737072861e-06,
+      "loss": 0.0121,
+      "step": 5095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34303408853743e-06,
+      "loss": 0.0351,
+      "step": 5096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3393698238859955e-06,
+      "loss": 0.0138,
+      "step": 5097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.335705943647976e-06,
+      "loss": 0.0015,
+      "step": 5098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3320424483527385e-06,
+      "loss": 0.0738,
+      "step": 5099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328379338529592e-06,
+      "loss": 0.0158,
+      "step": 5100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324716614707794e-06,
+      "loss": 0.0022,
+      "step": 5101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.321054277416535e-06,
+      "loss": 0.0325,
+      "step": 5102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3173923271849634e-06,
+      "loss": 0.0297,
+      "step": 5103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3137307645421605e-06,
+      "loss": 0.0057,
+      "step": 5104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310069590017159e-06,
+      "loss": 0.023,
+      "step": 5105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.306408804138937e-06,
+      "loss": 0.0359,
+      "step": 5106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302748407436405e-06,
+      "loss": 0.0165,
+      "step": 5107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.299088400438429e-06,
+      "loss": 0.0817,
+      "step": 5108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.295428783673813e-06,
+      "loss": 0.0115,
+      "step": 5109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.291769557671311e-06,
+      "loss": 0.0495,
+      "step": 5110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288110722959608e-06,
+      "loss": 0.0082,
+      "step": 5111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.284452280067344e-06,
+      "loss": 0.0303,
+      "step": 5112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280794229523098e-06,
+      "loss": 0.043,
+      "step": 5113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2771365718553965e-06,
+      "loss": 0.0228,
+      "step": 5114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2734793075927015e-06,
+      "loss": 0.1301,
+      "step": 5115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269822437263423e-06,
+      "loss": 0.0413,
+      "step": 5116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2661659613959105e-06,
+      "loss": 0.0022,
+      "step": 5117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.262509880518463e-06,
+      "loss": 0.0095,
+      "step": 5118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.258854195159315e-06,
+      "loss": 0.0317,
+      "step": 5119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2551989058466545e-06,
+      "loss": 0.0171,
+      "step": 5120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2515440131085975e-06,
+      "loss": 0.0057,
+      "step": 5121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.247889517473214e-06,
+      "loss": 0.0161,
+      "step": 5122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.244235419468515e-06,
+      "loss": 0.0191,
+      "step": 5123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240581719622451e-06,
+      "loss": 0.0061,
+      "step": 5124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236928418462914e-06,
+      "loss": 0.0066,
+      "step": 5125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.233275516517742e-06,
+      "loss": 0.0195,
+      "step": 5126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.229623014314716e-06,
+      "loss": 0.0447,
+      "step": 5127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225970912381557e-06,
+      "loss": 0.0377,
+      "step": 5128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222319211245925e-06,
+      "loss": 0.0384,
+      "step": 5129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218667911435425e-06,
+      "loss": 0.1042,
+      "step": 5130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.215017013477607e-06,
+      "loss": 0.0095,
+      "step": 5131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21136651789996e-06,
+      "loss": 0.1355,
+      "step": 5132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207716425229917e-06,
+      "loss": 0.003,
+      "step": 5133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204066735994848e-06,
+      "loss": 0.0115,
+      "step": 5134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200417450722068e-06,
+      "loss": 0.0113,
+      "step": 5135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1967685699388345e-06,
+      "loss": 0.0211,
+      "step": 5136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.193120094172347e-06,
+      "loss": 0.0476,
+      "step": 5137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.189472023949741e-06,
+      "loss": 0.0183,
+      "step": 5138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.185824359798099e-06,
+      "loss": 0.003,
+      "step": 5139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182177102244445e-06,
+      "loss": 0.1306,
+      "step": 5140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1785302518157415e-06,
+      "loss": 0.0488,
+      "step": 5141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174883809038896e-06,
+      "loss": 0.1171,
+      "step": 5142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171237774440747e-06,
+      "loss": 0.0231,
+      "step": 5143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.167592148548083e-06,
+      "loss": 0.0132,
+      "step": 5144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.163946931887636e-06,
+      "loss": 0.0683,
+      "step": 5145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160302124986073e-06,
+      "loss": 0.0129,
+      "step": 5146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.15665772837e-06,
+      "loss": 0.0544,
+      "step": 5147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1530137425659705e-06,
+      "loss": 0.0057,
+      "step": 5148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.149370168100473e-06,
+      "loss": 0.0054,
+      "step": 5149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.145727005499943e-06,
+      "loss": 0.0421,
+      "step": 5150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1420842552907465e-06,
+      "loss": 0.0367,
+      "step": 5151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1384419179991995e-06,
+      "loss": 0.0077,
+      "step": 5152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.134799994151553e-06,
+      "loss": 0.0091,
+      "step": 5153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1311584842740035e-06,
+      "loss": 0.0217,
+      "step": 5154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.127517388892678e-06,
+      "loss": 0.032,
+      "step": 5155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123876708533655e-06,
+      "loss": 0.0376,
+      "step": 5156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120236443722941e-06,
+      "loss": 0.0033,
+      "step": 5157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1165965949864934e-06,
+      "loss": 0.0515,
+      "step": 5158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112957162850208e-06,
+      "loss": 0.0027,
+      "step": 5159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109318147839912e-06,
+      "loss": 0.0206,
+      "step": 5160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.105679550481379e-06,
+      "loss": 0.0304,
+      "step": 5161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102041371300323e-06,
+      "loss": 0.0656,
+      "step": 5162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.098403610822398e-06,
+      "loss": 0.0018,
+      "step": 5163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094766269573189e-06,
+      "loss": 0.011,
+      "step": 5164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.091129348078231e-06,
+      "loss": 0.0673,
+      "step": 5165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087492846862992e-06,
+      "loss": 0.0311,
+      "step": 5166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.083856766452886e-06,
+      "loss": 0.0085,
+      "step": 5167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.080221107373255e-06,
+      "loss": 0.0136,
+      "step": 5168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0765858701493894e-06,
+      "loss": 0.0965,
+      "step": 5169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.072951055306521e-06,
+      "loss": 0.0311,
+      "step": 5170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0693166633698055e-06,
+      "loss": 0.0203,
+      "step": 5171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065682694864352e-06,
+      "loss": 0.0009,
+      "step": 5172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0620491503152065e-06,
+      "loss": 0.0027,
+      "step": 5173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.058416030247348e-06,
+      "loss": 0.0707,
+      "step": 5174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0547833351856976e-06,
+      "loss": 0.0415,
+      "step": 5175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0511510656551175e-06,
+      "loss": 0.0041,
+      "step": 5176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0475192221804014e-06,
+      "loss": 0.014,
+      "step": 5177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.043887805286287e-06,
+      "loss": 0.0098,
+      "step": 5178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0402568154974505e-06,
+      "loss": 0.013,
+      "step": 5179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.036626253338506e-06,
+      "loss": 0.0096,
+      "step": 5180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0329961193340014e-06,
+      "loss": 0.0466,
+      "step": 5181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029366414008425e-06,
+      "loss": 0.1062,
+      "step": 5182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.02573713788621e-06,
+      "loss": 0.0144,
+      "step": 5183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.02210829149172e-06,
+      "loss": 0.0008,
+      "step": 5184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0184798753492535e-06,
+      "loss": 0.0041,
+      "step": 5185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.014851889983058e-06,
+      "loss": 0.036,
+      "step": 5186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011224335917306e-06,
+      "loss": 0.0054,
+      "step": 5187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.007597213676118e-06,
+      "loss": 0.0703,
+      "step": 5188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.003970523783548e-06,
+      "loss": 0.0067,
+      "step": 5189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000344266763589e-06,
+      "loss": 0.0916,
+      "step": 5190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9967184431401665e-06,
+      "loss": 0.004,
+      "step": 5191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.99309305343715e-06,
+      "loss": 0.0011,
+      "step": 5192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.989468098178343e-06,
+      "loss": 0.0054,
+      "step": 5193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9858435778874865e-06,
+      "loss": 0.0107,
+      "step": 5194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982219493088258e-06,
+      "loss": 0.0026,
+      "step": 5195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978595844304272e-06,
+      "loss": 0.0093,
+      "step": 5196
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.974972632059085e-06,
+      "loss": 0.0599,
+      "step": 5197
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971349856876184e-06,
+      "loss": 0.0038,
+      "step": 5198
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967727519278995e-06,
+      "loss": 0.0017,
+      "step": 5199
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.964105619790877e-06,
+      "loss": 0.0322,
+      "step": 5200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9604841589351346e-06,
+      "loss": 0.0032,
+      "step": 5201
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956863137235002e-06,
+      "loss": 0.0009,
+      "step": 5202
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.953242555213655e-06,
+      "loss": 0.0606,
+      "step": 5203
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.949622413394199e-06,
+      "loss": 0.0271,
+      "step": 5204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.946002712299681e-06,
+      "loss": 0.0018,
+      "step": 5205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942383452453083e-06,
+      "loss": 0.0436,
+      "step": 5206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.938764634377325e-06,
+      "loss": 0.0649,
+      "step": 5207
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.935146258595258e-06,
+      "loss": 0.0485,
+      "step": 5208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.931528325629675e-06,
+      "loss": 0.042,
+      "step": 5209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.927910836003301e-06,
+      "loss": 0.005,
+      "step": 5210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.924293790238803e-06,
+      "loss": 0.0042,
+      "step": 5211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.920677188858776e-06,
+      "loss": 0.0045,
+      "step": 5212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.917061032385749e-06,
+      "loss": 0.0031,
+      "step": 5213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.913445321342198e-06,
+      "loss": 0.006,
+      "step": 5214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.1148,
+      "step": 5215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906215237633079e-06,
+      "loss": 0.1151,
+      "step": 5216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.902600866012128e-06,
+      "loss": 0.0591,
+      "step": 5217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.898986941909886e-06,
+      "loss": 0.004,
+      "step": 5218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.895373465848501e-06,
+      "loss": 0.0279,
+      "step": 5219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.891760438350057e-06,
+      "loss": 0.0061,
+      "step": 5220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.888147859936569e-06,
+      "loss": 0.0079,
+      "step": 5221
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884535731129991e-06,
+      "loss": 0.0435,
+      "step": 5222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8809240524522115e-06,
+      "loss": 0.0042,
+      "step": 5223
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.877312824425056e-06,
+      "loss": 0.0448,
+      "step": 5224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.873702047570279e-06,
+      "loss": 0.1029,
+      "step": 5225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.870091722409573e-06,
+      "loss": 0.0555,
+      "step": 5226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8664818494645645e-06,
+      "loss": 0.0022,
+      "step": 5227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.862872429256817e-06,
+      "loss": 0.0388,
+      "step": 5228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.859263462307828e-06,
+      "loss": 0.0321,
+      "step": 5229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8556549491390254e-06,
+      "loss": 0.0069,
+      "step": 5230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.852046890271778e-06,
+      "loss": 0.0004,
+      "step": 5231
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.848439286227385e-06,
+      "loss": 0.0081,
+      "step": 5232
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8448321375270824e-06,
+      "loss": 0.0027,
+      "step": 5233
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.841225444692032e-06,
+      "loss": 0.035,
+      "step": 5234
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.837619208243342e-06,
+      "loss": 0.0342,
+      "step": 5235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834013428702049e-06,
+      "loss": 0.0304,
+      "step": 5236
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.830408106589123e-06,
+      "loss": 0.0771,
+      "step": 5237
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826803242425466e-06,
+      "loss": 0.0066,
+      "step": 5238
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.823198836731921e-06,
+      "loss": 0.0173,
+      "step": 5239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8195948900292555e-06,
+      "loss": 0.0014,
+      "step": 5240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.815991402838176e-06,
+      "loss": 0.0521,
+      "step": 5241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812388375679322e-06,
+      "loss": 0.0144,
+      "step": 5242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808785809073268e-06,
+      "loss": 0.0031,
+      "step": 5243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.80518370354052e-06,
+      "loss": 0.007,
+      "step": 5244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.801582059601517e-06,
+      "loss": 0.0785,
+      "step": 5245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797980877776631e-06,
+      "loss": 0.0063,
+      "step": 5246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.794380158586174e-06,
+      "loss": 0.0665,
+      "step": 5247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790779902550376e-06,
+      "loss": 0.0694,
+      "step": 5248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.787180110189416e-06,
+      "loss": 0.1085,
+      "step": 5249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7835807820234e-06,
+      "loss": 0.0211,
+      "step": 5250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.779981918572364e-06,
+      "loss": 0.0235,
+      "step": 5251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7763835203562785e-06,
+      "loss": 0.0299,
+      "step": 5252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772785587895055e-06,
+      "loss": 0.0205,
+      "step": 5253
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7691881217085176e-06,
+      "loss": 0.001,
+      "step": 5254
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.765591122316444e-06,
+      "loss": 0.181,
+      "step": 5255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.761994590238536e-06,
+      "loss": 0.0009,
+      "step": 5256
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758398525994425e-06,
+      "loss": 0.0028,
+      "step": 5257
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7548029301036814e-06,
+      "loss": 0.0008,
+      "step": 5258
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7512078030858e-06,
+      "loss": 0.0028,
+      "step": 5259
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7476131454602185e-06,
+      "loss": 0.0245,
+      "step": 5260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744018957746295e-06,
+      "loss": 0.0295,
+      "step": 5261
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740425240463326e-06,
+      "loss": 0.0088,
+      "step": 5262
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.736831994130542e-06,
+      "loss": 0.0015,
+      "step": 5263
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.733239219267103e-06,
+      "loss": 0.0177,
+      "step": 5264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729646916392096e-06,
+      "loss": 0.0089,
+      "step": 5265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.72605508602455e-06,
+      "loss": 0.0154,
+      "step": 5266
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.722463728683419e-06,
+      "loss": 0.0005,
+      "step": 5267
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718872844887585e-06,
+      "loss": 0.0484,
+      "step": 5268
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715282435155873e-06,
+      "loss": 0.0226,
+      "step": 5269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711692500007027e-06,
+      "loss": 0.0151,
+      "step": 5270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7081030399597306e-06,
+      "loss": 0.0057,
+      "step": 5271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704514055532597e-06,
+      "loss": 0.0376,
+      "step": 5272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.700925547244173e-06,
+      "loss": 0.0025,
+      "step": 5273
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697337515612928e-06,
+      "loss": 0.0067,
+      "step": 5274
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693749961157272e-06,
+      "loss": 0.0384,
+      "step": 5275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690162884395541e-06,
+      "loss": 0.0463,
+      "step": 5276
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.686576285846007e-06,
+      "loss": 0.0012,
+      "step": 5277
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.682990166026864e-06,
+      "loss": 0.0116,
+      "step": 5278
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6794045254562434e-06,
+      "loss": 0.0055,
+      "step": 5279
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.675819364652212e-06,
+      "loss": 0.0407,
+      "step": 5280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6722346841327515e-06,
+      "loss": 0.0321,
+      "step": 5281
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66865048441579e-06,
+      "loss": 0.0226,
+      "step": 5282
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.665066766019179e-06,
+      "loss": 0.0466,
+      "step": 5283
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661483529460702e-06,
+      "loss": 0.0367,
+      "step": 5284
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.657900775258071e-06,
+      "loss": 0.0038,
+      "step": 5285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.654318503928934e-06,
+      "loss": 0.004,
+      "step": 5286
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.65073671599086e-06,
+      "loss": 0.0399,
+      "step": 5287
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647155411961354e-06,
+      "loss": 0.0658,
+      "step": 5288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643574592357852e-06,
+      "loss": 0.1146,
+      "step": 5289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63999425769772e-06,
+      "loss": 0.0389,
+      "step": 5290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.636414408498249e-06,
+      "loss": 0.0035,
+      "step": 5291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632835045276664e-06,
+      "loss": 0.0639,
+      "step": 5292
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629256168550119e-06,
+      "loss": 0.0017,
+      "step": 5293
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.625677778835702e-06,
+      "loss": 0.0043,
+      "step": 5294
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6220998766504185e-06,
+      "loss": 0.0036,
+      "step": 5295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6185224625112155e-06,
+      "loss": 0.0025,
+      "step": 5296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614945536934963e-06,
+      "loss": 0.0636,
+      "step": 5297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611369100438465e-06,
+      "loss": 0.0027,
+      "step": 5298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.607793153538451e-06,
+      "loss": 0.0152,
+      "step": 5299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604217696751585e-06,
+      "loss": 0.1501,
+      "step": 5300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.600642730594451e-06,
+      "loss": 0.0202,
+      "step": 5301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.59706825558357e-06,
+      "loss": 0.0032,
+      "step": 5302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593494272235392e-06,
+      "loss": 0.0049,
+      "step": 5303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.589920781066291e-06,
+      "loss": 0.0322,
+      "step": 5304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5863477825925705e-06,
+      "loss": 0.0873,
+      "step": 5305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582775277330467e-06,
+      "loss": 0.0039,
+      "step": 5306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579203265796148e-06,
+      "loss": 0.0037,
+      "step": 5307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5756317485057e-06,
+      "loss": 0.0705,
+      "step": 5308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.572060725975143e-06,
+      "loss": 0.035,
+      "step": 5309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.568490198720426e-06,
+      "loss": 0.0007,
+      "step": 5310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564920167257427e-06,
+      "loss": 0.1083,
+      "step": 5311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561350632101952e-06,
+      "loss": 0.0192,
+      "step": 5312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5577815937697375e-06,
+      "loss": 0.0119,
+      "step": 5313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554213052776441e-06,
+      "loss": 0.0455,
+      "step": 5314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.550645009637654e-06,
+      "loss": 0.0011,
+      "step": 5315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5470774648688965e-06,
+      "loss": 0.0028,
+      "step": 5316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.543510418985615e-06,
+      "loss": 0.0024,
+      "step": 5317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5399438725031805e-06,
+      "loss": 0.0056,
+      "step": 5318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.536377825936897e-06,
+      "loss": 0.0007,
+      "step": 5319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532812279801997e-06,
+      "loss": 0.1172,
+      "step": 5320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529247234613634e-06,
+      "loss": 0.0065,
+      "step": 5321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.525682690886898e-06,
+      "loss": 0.0445,
+      "step": 5322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.522118649136796e-06,
+      "loss": 0.0107,
+      "step": 5323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518555109878271e-06,
+      "loss": 0.0048,
+      "step": 5324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.514992073626188e-06,
+      "loss": 0.0045,
+      "step": 5325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.511429540895349e-06,
+      "loss": 0.0067,
+      "step": 5326
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507867512200467e-06,
+      "loss": 0.0527,
+      "step": 5327
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.504305988056199e-06,
+      "loss": 0.0011,
+      "step": 5328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500744968977116e-06,
+      "loss": 0.0026,
+      "step": 5329
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497184455477728e-06,
+      "loss": 0.0665,
+      "step": 5330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4936244480724575e-06,
+      "loss": 0.0312,
+      "step": 5331
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.490064947275665e-06,
+      "loss": 0.0444,
+      "step": 5332
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.486505953601636e-06,
+      "loss": 0.0501,
+      "step": 5333
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4829474675645844e-06,
+      "loss": 0.0028,
+      "step": 5334
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47938948967864e-06,
+      "loss": 0.0894,
+      "step": 5335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.475832020457875e-06,
+      "loss": 0.0165,
+      "step": 5336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.472275060416273e-06,
+      "loss": 0.0031,
+      "step": 5337
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4687186100677524e-06,
+      "loss": 0.0618,
+      "step": 5338
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4651626699261614e-06,
+      "loss": 0.0023,
+      "step": 5339
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.461607240505263e-06,
+      "loss": 0.0057,
+      "step": 5340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458052322318758e-06,
+      "loss": 0.0019,
+      "step": 5341
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4544979158802645e-06,
+      "loss": 0.0507,
+      "step": 5342
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.450944021703336e-06,
+      "loss": 0.0008,
+      "step": 5343
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447390640301443e-06,
+      "loss": 0.0027,
+      "step": 5344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.443837772187984e-06,
+      "loss": 0.0037,
+      "step": 5345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.440285417876288e-06,
+      "loss": 0.0972,
+      "step": 5346
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4367335778796055e-06,
+      "loss": 0.0047,
+      "step": 5347
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.433182252711113e-06,
+      "loss": 0.068,
+      "step": 5348
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429631442883914e-06,
+      "loss": 0.13,
+      "step": 5349
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.426081148911041e-06,
+      "loss": 0.0024,
+      "step": 5350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42253137130544e-06,
+      "loss": 0.0113,
+      "step": 5351
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418982110579997e-06,
+      "loss": 0.0242,
+      "step": 5352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.41543336724751e-06,
+      "loss": 0.0046,
+      "step": 5353
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411885141820713e-06,
+      "loss": 0.0071,
+      "step": 5354
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.408337434812261e-06,
+      "loss": 0.0116,
+      "step": 5355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4047902467347355e-06,
+      "loss": 0.0249,
+      "step": 5356
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.401243578100638e-06,
+      "loss": 0.0045,
+      "step": 5357
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397697429422401e-06,
+      "loss": 0.0033,
+      "step": 5358
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39415180121238e-06,
+      "loss": 0.011,
+      "step": 5359
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3906066939828546e-06,
+      "loss": 0.0084,
+      "step": 5360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387062108246026e-06,
+      "loss": 0.001,
+      "step": 5361
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383518044514028e-06,
+      "loss": 0.1628,
+      "step": 5362
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.379974503298914e-06,
+      "loss": 0.0035,
+      "step": 5363
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.376431485112657e-06,
+      "loss": 0.0026,
+      "step": 5364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372888990467165e-06,
+      "loss": 0.0805,
+      "step": 5365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.369347019874264e-06,
+      "loss": 0.0473,
+      "step": 5366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365805573845701e-06,
+      "loss": 0.0028,
+      "step": 5367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3622646528931574e-06,
+      "loss": 0.0106,
+      "step": 5368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35872425752823e-06,
+      "loss": 0.0677,
+      "step": 5369
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.355184388262445e-06,
+      "loss": 0.0665,
+      "step": 5370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3516450456072465e-06,
+      "loss": 0.0564,
+      "step": 5371
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348106230074007e-06,
+      "loss": 0.0051,
+      "step": 5372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.344567942174025e-06,
+      "loss": 0.0019,
+      "step": 5373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.341030182418516e-06,
+      "loss": 0.0027,
+      "step": 5374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.337492951318624e-06,
+      "loss": 0.0116,
+      "step": 5375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333956249385418e-06,
+      "loss": 0.0008,
+      "step": 5376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3304200771298886e-06,
+      "loss": 0.0385,
+      "step": 5377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.326884435062943e-06,
+      "loss": 0.0465,
+      "step": 5378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.323349323695424e-06,
+      "loss": 0.0023,
+      "step": 5379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3198147435380885e-06,
+      "loss": 0.0214,
+      "step": 5380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.316280695101621e-06,
+      "loss": 0.0014,
+      "step": 5381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.312747178896629e-06,
+      "loss": 0.0007,
+      "step": 5382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309214195433644e-06,
+      "loss": 0.0348,
+      "step": 5383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.305681745223114e-06,
+      "loss": 0.0007,
+      "step": 5384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302149828775419e-06,
+      "loss": 0.0043,
+      "step": 5385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.298618446600856e-06,
+      "loss": 0.0155,
+      "step": 5386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295087599209648e-06,
+      "loss": 0.0133,
+      "step": 5387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291557287111938e-06,
+      "loss": 0.0542,
+      "step": 5388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2880275108177915e-06,
+      "loss": 0.1263,
+      "step": 5389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2844982708372e-06,
+      "loss": 0.0458,
+      "step": 5390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.28096956768008e-06,
+      "loss": 0.0377,
+      "step": 5391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277441401856257e-06,
+      "loss": 0.0438,
+      "step": 5392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273913773875491e-06,
+      "loss": 0.055,
+      "step": 5393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.27038668424746e-06,
+      "loss": 0.1124,
+      "step": 5394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266860133481769e-06,
+      "loss": 0.009,
+      "step": 5395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26333412208794e-06,
+      "loss": 0.0053,
+      "step": 5396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259808650575416e-06,
+      "loss": 0.0222,
+      "step": 5397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.256283719453567e-06,
+      "loss": 0.0219,
+      "step": 5398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252759329231681e-06,
+      "loss": 0.0105,
+      "step": 5399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.249235480418974e-06,
+      "loss": 0.1532,
+      "step": 5400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.245712173524572e-06,
+      "loss": 0.0183,
+      "step": 5401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242189409057533e-06,
+      "loss": 0.0015,
+      "step": 5402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238667187526832e-06,
+      "loss": 0.008,
+      "step": 5403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235145509441372e-06,
+      "loss": 0.0514,
+      "step": 5404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.231624375309971e-06,
+      "loss": 0.0053,
+      "step": 5405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2281037856413625e-06,
+      "loss": 0.0094,
+      "step": 5406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224583740944214e-06,
+      "loss": 0.0157,
+      "step": 5407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221064241727108e-06,
+      "loss": 0.0354,
+      "step": 5408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2175452884985536e-06,
+      "loss": 0.0027,
+      "step": 5409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214026881766968e-06,
+      "loss": 0.0068,
+      "step": 5410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.210509022040705e-06,
+      "loss": 0.0543,
+      "step": 5411
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206991709828028e-06,
+      "loss": 0.0068,
+      "step": 5412
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2034749456371315e-06,
+      "loss": 0.0595,
+      "step": 5413
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199958729976118e-06,
+      "loss": 0.1112,
+      "step": 5414
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19644306335302e-06,
+      "loss": 0.0268,
+      "step": 5415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1929279462757905e-06,
+      "loss": 0.0124,
+      "step": 5416
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1894133792523005e-06,
+      "loss": 0.0006,
+      "step": 5417
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18589936279034e-06,
+      "loss": 0.0605,
+      "step": 5418
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1823858973976255e-06,
+      "loss": 0.0385,
+      "step": 5419
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.178872983581783e-06,
+      "loss": 0.0284,
+      "step": 5420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175360621850371e-06,
+      "loss": 0.1343,
+      "step": 5421
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17184881271086e-06,
+      "loss": 0.0397,
+      "step": 5422
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168337556670647e-06,
+      "loss": 0.0056,
+      "step": 5423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164826854237042e-06,
+      "loss": 0.0553,
+      "step": 5424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16131670591728e-06,
+      "loss": 0.0661,
+      "step": 5425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.157807112218517e-06,
+      "loss": 0.0021,
+      "step": 5426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154298073647822e-06,
+      "loss": 0.0627,
+      "step": 5427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.15078959071219e-06,
+      "loss": 0.1174,
+      "step": 5428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.147281663918534e-06,
+      "loss": 0.0307,
+      "step": 5429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14377429377369e-06,
+      "loss": 0.0003,
+      "step": 5430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.140267480784405e-06,
+      "loss": 0.0047,
+      "step": 5431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136761225457357e-06,
+      "loss": 0.0596,
+      "step": 5432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133255528299128e-06,
+      "loss": 0.0352,
+      "step": 5433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.129750389816234e-06,
+      "loss": 0.0011,
+      "step": 5434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1262458105151034e-06,
+      "loss": 0.1282,
+      "step": 5435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.122741790902089e-06,
+      "loss": 0.0267,
+      "step": 5436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.119238331483455e-06,
+      "loss": 0.0076,
+      "step": 5437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115735432765387e-06,
+      "loss": 0.0841,
+      "step": 5438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.112233095253995e-06,
+      "loss": 0.0073,
+      "step": 5439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.108731319455303e-06,
+      "loss": 0.0499,
+      "step": 5440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105230105875254e-06,
+      "loss": 0.0491,
+      "step": 5441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101729455019711e-06,
+      "loss": 0.0007,
+      "step": 5442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.098229367394455e-06,
+      "loss": 0.061,
+      "step": 5443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094729843505187e-06,
+      "loss": 0.0026,
+      "step": 5444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091230883857525e-06,
+      "loss": 0.0268,
+      "step": 5445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08773248895701e-06,
+      "loss": 0.0588,
+      "step": 5446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.084234659309088e-06,
+      "loss": 0.0058,
+      "step": 5447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.080737395419139e-06,
+      "loss": 0.0801,
+      "step": 5448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077240697792457e-06,
+      "loss": 0.0143,
+      "step": 5449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.073744566934247e-06,
+      "loss": 0.0048,
+      "step": 5450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.070249003349639e-06,
+      "loss": 0.0037,
+      "step": 5451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066754007543681e-06,
+      "loss": 0.0058,
+      "step": 5452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063259580021337e-06,
+      "loss": 0.0882,
+      "step": 5453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059765721287487e-06,
+      "loss": 0.0351,
+      "step": 5454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0562724318469326e-06,
+      "loss": 0.0027,
+      "step": 5455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052779712204391e-06,
+      "loss": 0.0052,
+      "step": 5456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049287562864501e-06,
+      "loss": 0.0799,
+      "step": 5457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.04579598433181e-06,
+      "loss": 0.1013,
+      "step": 5458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042304977110793e-06,
+      "loss": 0.1337,
+      "step": 5459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.03881454170584e-06,
+      "loss": 0.0069,
+      "step": 5460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.035324678621248e-06,
+      "loss": 0.0672,
+      "step": 5461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0318353883612465e-06,
+      "loss": 0.0388,
+      "step": 5462
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0283466714299725e-06,
+      "loss": 0.0238,
+      "step": 5463
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.024858528331483e-06,
+      "loss": 0.0776,
+      "step": 5464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.021370959569753e-06,
+      "loss": 0.0026,
+      "step": 5465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.017883965648676e-06,
+      "loss": 0.0288,
+      "step": 5466
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.014397547072058e-06,
+      "loss": 0.0586,
+      "step": 5467
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.010911704343624e-06,
+      "loss": 0.0579,
+      "step": 5468
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0074264379670155e-06,
+      "loss": 0.0578,
+      "step": 5469
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.003941748445794e-06,
+      "loss": 0.0007,
+      "step": 5470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0004576362834295e-06,
+      "loss": 0.1161,
+      "step": 5471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996974101983318e-06,
+      "loss": 0.0287,
+      "step": 5472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.993491146048766e-06,
+      "loss": 0.0037,
+      "step": 5473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9900087689830045e-06,
+      "loss": 0.0006,
+      "step": 5474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986526971289166e-06,
+      "loss": 0.0022,
+      "step": 5475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983045753470308e-06,
+      "loss": 0.0042,
+      "step": 5476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979565116029408e-06,
+      "loss": 0.0782,
+      "step": 5477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976085059469354e-06,
+      "loss": 0.0741,
+      "step": 5478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972605584292957e-06,
+      "loss": 0.0022,
+      "step": 5479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9691266910029296e-06,
+      "loss": 0.0497,
+      "step": 5480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.965648380101916e-06,
+      "loss": 0.0813,
+      "step": 5481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.962170652092468e-06,
+      "loss": 0.009,
+      "step": 5482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.958693507477058e-06,
+      "loss": 0.0006,
+      "step": 5483
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955216946758067e-06,
+      "loss": 0.0469,
+      "step": 5484
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951740970437796e-06,
+      "loss": 0.1495,
+      "step": 5485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948265579018464e-06,
+      "loss": 0.0623,
+      "step": 5486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.944790773002203e-06,
+      "loss": 0.0845,
+      "step": 5487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.941316552891058e-06,
+      "loss": 0.001,
+      "step": 5488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.937842919186992e-06,
+      "loss": 0.0872,
+      "step": 5489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.934369872391883e-06,
+      "loss": 0.0008,
+      "step": 5490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.930897413007523e-06,
+      "loss": 0.0273,
+      "step": 5491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.927425541535621e-06,
+      "loss": 0.1017,
+      "step": 5492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923954258477804e-06,
+      "loss": 0.0509,
+      "step": 5493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920483564335604e-06,
+      "loss": 0.0063,
+      "step": 5494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.917013459610477e-06,
+      "loss": 0.1127,
+      "step": 5495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.913543944803795e-06,
+      "loss": 0.0357,
+      "step": 5496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9100750204168325e-06,
+      "loss": 0.0036,
+      "step": 5497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.906606686950792e-06,
+      "loss": 0.0572,
+      "step": 5498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903138944906785e-06,
+      "loss": 0.0018,
+      "step": 5499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8996717947858395e-06,
+      "loss": 0.0318,
+      "step": 5500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.896205237088894e-06,
+      "loss": 0.028,
+      "step": 5501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8927392723168095e-06,
+      "loss": 0.0044,
+      "step": 5502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.889273900970347e-06,
+      "loss": 0.0051,
+      "step": 5503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.885809123550194e-06,
+      "loss": 0.0405,
+      "step": 5504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8823449405569525e-06,
+      "loss": 0.0225,
+      "step": 5505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8788813524911324e-06,
+      "loss": 0.0708,
+      "step": 5506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.875418359853158e-06,
+      "loss": 0.0565,
+      "step": 5507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.871955963143372e-06,
+      "loss": 0.0177,
+      "step": 5508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868494162862028e-06,
+      "loss": 0.0709,
+      "step": 5509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8650329595092984e-06,
+      "loss": 0.0142,
+      "step": 5510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.861572353585259e-06,
+      "loss": 0.0125,
+      "step": 5511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858112345589906e-06,
+      "loss": 0.0251,
+      "step": 5512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.854652936023151e-06,
+      "loss": 0.052,
+      "step": 5513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.851194125384818e-06,
+      "loss": 0.0036,
+      "step": 5514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.847735914174643e-06,
+      "loss": 0.0169,
+      "step": 5515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8442783028922725e-06,
+      "loss": 0.003,
+      "step": 5516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.840821292037268e-06,
+      "loss": 0.0084,
+      "step": 5517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.837364882109111e-06,
+      "loss": 0.0262,
+      "step": 5518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833909073607187e-06,
+      "loss": 0.0369,
+      "step": 5519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8304538670308e-06,
+      "loss": 0.0175,
+      "step": 5520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.826999262879164e-06,
+      "loss": 0.0407,
+      "step": 5521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823545261651409e-06,
+      "loss": 0.0294,
+      "step": 5522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.820091863846577e-06,
+      "loss": 0.0023,
+      "step": 5523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816639069963619e-06,
+      "loss": 0.0791,
+      "step": 5524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813186880501409e-06,
+      "loss": 0.058,
+      "step": 5525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.809735295958718e-06,
+      "loss": 0.0139,
+      "step": 5526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806284316834242e-06,
+      "loss": 0.0097,
+      "step": 5527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.802833943626585e-06,
+      "loss": 0.0881,
+      "step": 5528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.799384176834268e-06,
+      "loss": 0.011,
+      "step": 5529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795935016955714e-06,
+      "loss": 0.0488,
+      "step": 5530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7924864644892675e-06,
+      "loss": 0.002,
+      "step": 5531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7890385199331834e-06,
+      "loss": 0.0027,
+      "step": 5532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.785591183785626e-06,
+      "loss": 0.0013,
+      "step": 5533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.782144456544681e-06,
+      "loss": 0.0342,
+      "step": 5534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.778698338708329e-06,
+      "loss": 0.0132,
+      "step": 5535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.775252830774475e-06,
+      "loss": 0.0023,
+      "step": 5536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7718079332409345e-06,
+      "loss": 0.0503,
+      "step": 5537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768363646605435e-06,
+      "loss": 0.0028,
+      "step": 5538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76491997136561e-06,
+      "loss": 0.0033,
+      "step": 5539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7614769080190145e-06,
+      "loss": 0.0268,
+      "step": 5540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.758034457063105e-06,
+      "loss": 0.0343,
+      "step": 5541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.75459261899526e-06,
+      "loss": 0.0276,
+      "step": 5542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751151394312755e-06,
+      "loss": 0.0273,
+      "step": 5543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7477107835127945e-06,
+      "loss": 0.0011,
+      "step": 5544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.744270787092474e-06,
+      "loss": 0.005,
+      "step": 5545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740831405548819e-06,
+      "loss": 0.0022,
+      "step": 5546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.737392639378757e-06,
+      "loss": 0.0209,
+      "step": 5547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.733954489079127e-06,
+      "loss": 0.0071,
+      "step": 5548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7305169551466835e-06,
+      "loss": 0.0241,
+      "step": 5549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727080038078084e-06,
+      "loss": 0.0057,
+      "step": 5550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.723643738369909e-06,
+      "loss": 0.0048,
+      "step": 5551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7202080565186335e-06,
+      "loss": 0.0063,
+      "step": 5552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.716772993020656e-06,
+      "loss": 0.0021,
+      "step": 5553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71333854837228e-06,
+      "loss": 0.0319,
+      "step": 5554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709904723069723e-06,
+      "loss": 0.0003,
+      "step": 5555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.706471517609111e-06,
+      "loss": 0.1207,
+      "step": 5556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7030389324864845e-06,
+      "loss": 0.0032,
+      "step": 5557
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6996069681977825e-06,
+      "loss": 0.0086,
+      "step": 5558
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6961756252388664e-06,
+      "loss": 0.0069,
+      "step": 5559
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.692744904105505e-06,
+      "loss": 0.0309,
+      "step": 5560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689314805293378e-06,
+      "loss": 0.0042,
+      "step": 5561
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685885329298066e-06,
+      "loss": 0.066,
+      "step": 5562
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6824564766150724e-06,
+      "loss": 0.0339,
+      "step": 5563
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679028247739803e-06,
+      "loss": 0.0886,
+      "step": 5564
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6756006431675755e-06,
+      "loss": 0.0012,
+      "step": 5565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672173663393618e-06,
+      "loss": 0.1402,
+      "step": 5566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66874730891307e-06,
+      "loss": 0.0665,
+      "step": 5567
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66532158022098e-06,
+      "loss": 0.004,
+      "step": 5568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.661896477812294e-06,
+      "loss": 0.0391,
+      "step": 5569
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.658472002181886e-06,
+      "loss": 0.0407,
+      "step": 5570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.655048153824534e-06,
+      "loss": 0.0309,
+      "step": 5571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651624933234915e-06,
+      "loss": 0.0647,
+      "step": 5572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.648202340907627e-06,
+      "loss": 0.0073,
+      "step": 5573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6447803773371725e-06,
+      "loss": 0.1152,
+      "step": 5574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.641359043017964e-06,
+      "loss": 0.0063,
+      "step": 5575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.637938338444325e-06,
+      "loss": 0.0564,
+      "step": 5576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.634518264110485e-06,
+      "loss": 0.0051,
+      "step": 5577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.631098820510587e-06,
+      "loss": 0.0995,
+      "step": 5578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.627680008138671e-06,
+      "loss": 0.0804,
+      "step": 5579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.624261827488698e-06,
+      "loss": 0.0089,
+      "step": 5580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.620844279054536e-06,
+      "loss": 0.0395,
+      "step": 5581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617427363329957e-06,
+      "loss": 0.0552,
+      "step": 5582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614011080808647e-06,
+      "loss": 0.0742,
+      "step": 5583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610595431984195e-06,
+      "loss": 0.0092,
+      "step": 5584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6071804173501065e-06,
+      "loss": 0.008,
+      "step": 5585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60376603739978e-06,
+      "loss": 0.0159,
+      "step": 5586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6003522926265434e-06,
+      "loss": 0.0074,
+      "step": 5587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.596939183523611e-06,
+      "loss": 0.0134,
+      "step": 5588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593526710584123e-06,
+      "loss": 0.0455,
+      "step": 5589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5901148743011156e-06,
+      "loss": 0.0937,
+      "step": 5590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.586703675167544e-06,
+      "loss": 0.0016,
+      "step": 5591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.58329311367626e-06,
+      "loss": 0.0064,
+      "step": 5592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5798831903200336e-06,
+      "loss": 0.0478,
+      "step": 5593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.576473905591534e-06,
+      "loss": 0.0873,
+      "step": 5594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.573065259983347e-06,
+      "loss": 0.0439,
+      "step": 5595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569657253987952e-06,
+      "loss": 0.029,
+      "step": 5596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5662498880977526e-06,
+      "loss": 0.0199,
+      "step": 5597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562843162805051e-06,
+      "loss": 0.0038,
+      "step": 5598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.559437078602052e-06,
+      "loss": 0.0015,
+      "step": 5599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.556031635980879e-06,
+      "loss": 0.0574,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.552626835433556e-06,
+      "loss": 0.0128,
+      "step": 5601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549222677452017e-06,
+      "loss": 0.017,
+      "step": 5602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5458191625281e-06,
+      "loss": 0.0011,
+      "step": 5603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.542416291153558e-06,
+      "loss": 0.0434,
+      "step": 5604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539014063820035e-06,
+      "loss": 0.0785,
+      "step": 5605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535612481019098e-06,
+      "loss": 0.0664,
+      "step": 5606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532211543242214e-06,
+      "loss": 0.0019,
+      "step": 5607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528811250980757e-06,
+      "loss": 0.0642,
+      "step": 5608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525411604726008e-06,
+      "loss": 0.0091,
+      "step": 5609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.522012604969156e-06,
+      "loss": 0.0436,
+      "step": 5610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518614252201295e-06,
+      "loss": 0.003,
+      "step": 5611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.515216546913431e-06,
+      "loss": 0.0187,
+      "step": 5612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.511819489596464e-06,
+      "loss": 0.001,
+      "step": 5613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5084230807412135e-06,
+      "loss": 0.005,
+      "step": 5614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.505027320838394e-06,
+      "loss": 0.1375,
+      "step": 5615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5016322103786345e-06,
+      "loss": 0.0382,
+      "step": 5616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.498237749852469e-06,
+      "loss": 0.0369,
+      "step": 5617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494843939750335e-06,
+      "loss": 0.003,
+      "step": 5618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491450780562579e-06,
+      "loss": 0.0235,
+      "step": 5619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.488058272779449e-06,
+      "loss": 0.04,
+      "step": 5620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.484666416891109e-06,
+      "loss": 0.0304,
+      "step": 5621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481275213387611e-06,
+      "loss": 0.0078,
+      "step": 5622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477884662758928e-06,
+      "loss": 0.0378,
+      "step": 5623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.474494765494934e-06,
+      "loss": 0.0038,
+      "step": 5624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471105522085408e-06,
+      "loss": 0.0086,
+      "step": 5625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.46771693302004e-06,
+      "loss": 0.0186,
+      "step": 5626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4643289987884105e-06,
+      "loss": 0.003,
+      "step": 5627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4609417198800206e-06,
+      "loss": 0.0357,
+      "step": 5628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457555096784272e-06,
+      "loss": 0.0162,
+      "step": 5629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.45416912999047e-06,
+      "loss": 0.0562,
+      "step": 5630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.45078381998783e-06,
+      "loss": 0.0032,
+      "step": 5631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4473991672654624e-06,
+      "loss": 0.004,
+      "step": 5632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.444015172312391e-06,
+      "loss": 0.02,
+      "step": 5633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440631835617543e-06,
+      "loss": 0.1541,
+      "step": 5634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43724915766975e-06,
+      "loss": 0.022,
+      "step": 5635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433867138957751e-06,
+      "loss": 0.0945,
+      "step": 5636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430485779970182e-06,
+      "loss": 0.0013,
+      "step": 5637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.427105081195594e-06,
+      "loss": 0.036,
+      "step": 5638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423725043122437e-06,
+      "loss": 0.0861,
+      "step": 5639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.420345666239064e-06,
+      "loss": 0.1276,
+      "step": 5640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416966951033731e-06,
+      "loss": 0.2064,
+      "step": 5641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413588897994606e-06,
+      "loss": 0.0052,
+      "step": 5642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410211507609755e-06,
+      "loss": 0.0273,
+      "step": 5643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406834780367154e-06,
+      "loss": 0.0408,
+      "step": 5644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.403458716754675e-06,
+      "loss": 0.0259,
+      "step": 5645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.400083317260103e-06,
+      "loss": 0.0314,
+      "step": 5646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396708582371119e-06,
+      "loss": 0.0111,
+      "step": 5647
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39333451257532e-06,
+      "loss": 0.0631,
+      "step": 5648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389961108360188e-06,
+      "loss": 0.0106,
+      "step": 5649
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.386588370213124e-06,
+      "loss": 0.0664,
+      "step": 5650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383216298621427e-06,
+      "loss": 0.0389,
+      "step": 5651
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379844894072303e-06,
+      "loss": 0.0153,
+      "step": 5652
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376474157052858e-06,
+      "loss": 0.0071,
+      "step": 5653
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3731040880501095e-06,
+      "loss": 0.0024,
+      "step": 5654
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.369734687550961e-06,
+      "loss": 0.0505,
+      "step": 5655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.366365956042236e-06,
+      "loss": 0.0232,
+      "step": 5656
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362997894010661e-06,
+      "loss": 0.0409,
+      "step": 5657
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.359630501942851e-06,
+      "loss": 0.0415,
+      "step": 5658
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35626378032534e-06,
+      "loss": 0.0078,
+      "step": 5659
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352897729644555e-06,
+      "loss": 0.0294,
+      "step": 5660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349532350386832e-06,
+      "loss": 0.0246,
+      "step": 5661
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3461676430384115e-06,
+      "loss": 0.0038,
+      "step": 5662
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.342803608085429e-06,
+      "loss": 0.0307,
+      "step": 5663
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3394402460139275e-06,
+      "loss": 0.0024,
+      "step": 5664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.33607755730986e-06,
+      "loss": 0.1297,
+      "step": 5665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3327155424590635e-06,
+      "loss": 0.0101,
+      "step": 5666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3293542019472975e-06,
+      "loss": 0.0076,
+      "step": 5667
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32599353626021e-06,
+      "loss": 0.0004,
+      "step": 5668
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.322633545883358e-06,
+      "loss": 0.0344,
+      "step": 5669
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3192742313022005e-06,
+      "loss": 0.0735,
+      "step": 5670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3159155930021e-06,
+      "loss": 0.0272,
+      "step": 5671
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.312557631468318e-06,
+      "loss": 0.0014,
+      "step": 5672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309200347186021e-06,
+      "loss": 0.0711,
+      "step": 5673
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305843740640279e-06,
+      "loss": 0.012,
+      "step": 5674
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302487812316054e-06,
+      "loss": 0.1165,
+      "step": 5675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299132562698222e-06,
+      "loss": 0.0331,
+      "step": 5676
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295777992271557e-06,
+      "loss": 0.0059,
+      "step": 5677
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2924241015207326e-06,
+      "loss": 0.0075,
+      "step": 5678
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.289070890930328e-06,
+      "loss": 0.0075,
+      "step": 5679
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28571836098482e-06,
+      "loss": 0.0317,
+      "step": 5680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.282366512168595e-06,
+      "loss": 0.0251,
+      "step": 5681
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279015344965928e-06,
+      "loss": 0.1271,
+      "step": 5682
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.275664859861003e-06,
+      "loss": 0.0111,
+      "step": 5683
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.272315057337911e-06,
+      "loss": 0.0765,
+      "step": 5684
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.268965937880632e-06,
+      "loss": 0.0238,
+      "step": 5685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265617501973055e-06,
+      "loss": 0.0355,
+      "step": 5686
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.262269750098972e-06,
+      "loss": 0.0212,
+      "step": 5687
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25892268274207e-06,
+      "loss": 0.0146,
+      "step": 5688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.255576300385943e-06,
+      "loss": 0.018,
+      "step": 5689
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252230603514082e-06,
+      "loss": 0.0066,
+      "step": 5690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.248885592609885e-06,
+      "loss": 0.0004,
+      "step": 5691
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.245541268156636e-06,
+      "loss": 0.0029,
+      "step": 5692
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242197630637538e-06,
+      "loss": 0.015,
+      "step": 5693
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238854680535683e-06,
+      "loss": 0.0446,
+      "step": 5694
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235512418334073e-06,
+      "loss": 0.0027,
+      "step": 5695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2321708445155985e-06,
+      "loss": 0.0194,
+      "step": 5696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228829959563057e-06,
+      "loss": 0.0035,
+      "step": 5697
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225489763959151e-06,
+      "loss": 0.0228,
+      "step": 5698
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2221502581864755e-06,
+      "loss": 0.0298,
+      "step": 5699
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2188114427275315e-06,
+      "loss": 0.0068,
+      "step": 5700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.215473318064722e-06,
+      "loss": 0.0387,
+      "step": 5701
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212135884680338e-06,
+      "loss": 0.1694,
+      "step": 5702
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208799143056581e-06,
+      "loss": 0.0481,
+      "step": 5703
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.205463093675552e-06,
+      "loss": 0.0061,
+      "step": 5704
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202127737019251e-06,
+      "loss": 0.0027,
+      "step": 5705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198793073569577e-06,
+      "loss": 0.0091,
+      "step": 5706
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195459103808327e-06,
+      "loss": 0.0078,
+      "step": 5707
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192125828217203e-06,
+      "loss": 0.0084,
+      "step": 5708
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1887932472778056e-06,
+      "loss": 0.0006,
+      "step": 5709
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.185461361471631e-06,
+      "loss": 0.0189,
+      "step": 5710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.182130171280071e-06,
+      "loss": 0.1195,
+      "step": 5711
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178799677184428e-06,
+      "loss": 0.0339,
+      "step": 5712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175469879665898e-06,
+      "loss": 0.0394,
+      "step": 5713
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.172140779205576e-06,
+      "loss": 0.0025,
+      "step": 5714
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.168812376284462e-06,
+      "loss": 0.0089,
+      "step": 5715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.165484671383445e-06,
+      "loss": 0.0313,
+      "step": 5716
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162157664983322e-06,
+      "loss": 0.0228,
+      "step": 5717
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.158831357564791e-06,
+      "loss": 0.0588,
+      "step": 5718
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1555057496084335e-06,
+      "loss": 0.0031,
+      "step": 5719
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152180841594745e-06,
+      "loss": 0.051,
+      "step": 5720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148856634004113e-06,
+      "loss": 0.0679,
+      "step": 5721
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.145533127316831e-06,
+      "loss": 0.0165,
+      "step": 5722
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.142210322013085e-06,
+      "loss": 0.0806,
+      "step": 5723
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1388882185729546e-06,
+      "loss": 0.0381,
+      "step": 5724
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135566817476432e-06,
+      "loss": 0.0056,
+      "step": 5725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1322461192033945e-06,
+      "loss": 0.0536,
+      "step": 5726
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12892612423363e-06,
+      "loss": 0.0475,
+      "step": 5727
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12560683304681e-06,
+      "loss": 0.0081,
+      "step": 5728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.122288246122517e-06,
+      "loss": 0.047,
+      "step": 5729
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.118970363940229e-06,
+      "loss": 0.0227,
+      "step": 5730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.115653186979318e-06,
+      "loss": 0.0181,
+      "step": 5731
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112336715719057e-06,
+      "loss": 0.1356,
+      "step": 5732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.109020950638618e-06,
+      "loss": 0.0777,
+      "step": 5733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105705892217069e-06,
+      "loss": 0.0835,
+      "step": 5734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.10239154093338e-06,
+      "loss": 0.0069,
+      "step": 5735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099077897266407e-06,
+      "loss": 0.0206,
+      "step": 5736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.095764961694923e-06,
+      "loss": 0.0165,
+      "step": 5737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092452734697577e-06,
+      "loss": 0.0498,
+      "step": 5738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089141216752931e-06,
+      "loss": 0.0091,
+      "step": 5739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.085830408339441e-06,
+      "loss": 0.044,
+      "step": 5740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.082520309935459e-06,
+      "loss": 0.0383,
+      "step": 5741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.079210922019235e-06,
+      "loss": 0.0154,
+      "step": 5742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.075902245068917e-06,
+      "loss": 0.0017,
+      "step": 5743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072594279562552e-06,
+      "loss": 0.0239,
+      "step": 5744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069287025978075e-06,
+      "loss": 0.015,
+      "step": 5745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.065980484793328e-06,
+      "loss": 0.0296,
+      "step": 5746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.062674656486049e-06,
+      "loss": 0.0655,
+      "step": 5747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059369541533868e-06,
+      "loss": 0.0275,
+      "step": 5748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.056065140414317e-06,
+      "loss": 0.0098,
+      "step": 5749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.052761453604824e-06,
+      "loss": 0.0102,
+      "step": 5750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.04945848158271e-06,
+      "loss": 0.0022,
+      "step": 5751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.046156224825195e-06,
+      "loss": 0.0036,
+      "step": 5752
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.042854683809397e-06,
+      "loss": 0.0378,
+      "step": 5753
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.039553859012334e-06,
+      "loss": 0.0426,
+      "step": 5754
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0362537509109065e-06,
+      "loss": 0.0682,
+      "step": 5755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032954359981927e-06,
+      "loss": 0.0348,
+      "step": 5756
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0296556867020995e-06,
+      "loss": 0.0056,
+      "step": 5757
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.026357731548019e-06,
+      "loss": 0.0524,
+      "step": 5758
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.023060494996184e-06,
+      "loss": 0.0046,
+      "step": 5759
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019763977522987e-06,
+      "loss": 0.0054,
+      "step": 5760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.016468179604712e-06,
+      "loss": 0.1474,
+      "step": 5761
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.01317310171755e-06,
+      "loss": 0.0059,
+      "step": 5762
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.009878744337571e-06,
+      "loss": 0.0028,
+      "step": 5763
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0065851079407605e-06,
+      "loss": 0.0089,
+      "step": 5764
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.00329219300298e-06,
+      "loss": 0.0205,
+      "step": 5765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.0052,
+      "step": 5766
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996708529407489e-06,
+      "loss": 0.0507,
+      "step": 5767
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.993417781701e-06,
+      "loss": 0.0113,
+      "step": 5768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990127757355989e-06,
+      "loss": 0.0029,
+      "step": 5769
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.986838456847805e-06,
+      "loss": 0.044,
+      "step": 5770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9835498806516965e-06,
+      "loss": 0.0041,
+      "step": 5771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980262029242798e-06,
+      "loss": 0.0009,
+      "step": 5772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976974903096148e-06,
+      "loss": 0.0044,
+      "step": 5773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973688502686676e-06,
+      "loss": 0.026,
+      "step": 5774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.97040282848921e-06,
+      "loss": 0.0043,
+      "step": 5775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.96711788097847e-06,
+      "loss": 0.0008,
+      "step": 5776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963833660629073e-06,
+      "loss": 0.1317,
+      "step": 5777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.960550167915533e-06,
+      "loss": 0.0342,
+      "step": 5778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957267403312249e-06,
+      "loss": 0.0025,
+      "step": 5779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.953985367293528e-06,
+      "loss": 0.0058,
+      "step": 5780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.95070406033356e-06,
+      "loss": 0.0034,
+      "step": 5781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947423482906436e-06,
+      "loss": 0.0015,
+      "step": 5782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944143635486144e-06,
+      "loss": 0.0224,
+      "step": 5783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94086451854656e-06,
+      "loss": 0.0768,
+      "step": 5784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.93758613256146e-06,
+      "loss": 0.0341,
+      "step": 5785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.934308478004511e-06,
+      "loss": 0.0861,
+      "step": 5786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931031555349276e-06,
+      "loss": 0.0009,
+      "step": 5787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.927755365069215e-06,
+      "loss": 0.061,
+      "step": 5788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924479907637671e-06,
+      "loss": 0.0173,
+      "step": 5789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921205183527893e-06,
+      "loss": 0.0014,
+      "step": 5790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.91793119321302e-06,
+      "loss": 0.0099,
+      "step": 5791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9146579371660895e-06,
+      "loss": 0.0497,
+      "step": 5792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.911385415860021e-06,
+      "loss": 0.0757,
+      "step": 5793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.908113629767637e-06,
+      "loss": 0.0755,
+      "step": 5794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904842579361653e-06,
+      "loss": 0.0629,
+      "step": 5795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.901572265114677e-06,
+      "loss": 0.0035,
+      "step": 5796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.898302687499215e-06,
+      "loss": 0.1054,
+      "step": 5797
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8950338469876556e-06,
+      "loss": 0.0056,
+      "step": 5798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891765744052291e-06,
+      "loss": 0.0293,
+      "step": 5799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.888498379165303e-06,
+      "loss": 0.0165,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885231752798767e-06,
+      "loss": 0.0015,
+      "step": 5801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881965865424653e-06,
+      "loss": 0.019,
+      "step": 5802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.878700717514824e-06,
+      "loss": 0.0046,
+      "step": 5803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875436309541033e-06,
+      "loss": 0.0556,
+      "step": 5804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.872172641974934e-06,
+      "loss": 0.0311,
+      "step": 5805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86890971528806e-06,
+      "loss": 0.0385,
+      "step": 5806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.865647529951854e-06,
+      "loss": 0.1066,
+      "step": 5807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.862386086437635e-06,
+      "loss": 0.1552,
+      "step": 5808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.859125385216628e-06,
+      "loss": 0.0706,
+      "step": 5809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8558654267599445e-06,
+      "loss": 0.0116,
+      "step": 5810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852606211538591e-06,
+      "loss": 0.0598,
+      "step": 5811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8493477400234665e-06,
+      "loss": 0.0006,
+      "step": 5812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8460900126853605e-06,
+      "loss": 0.0035,
+      "step": 5813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84283302999496e-06,
+      "loss": 0.0387,
+      "step": 5814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.839576792422833e-06,
+      "loss": 0.0189,
+      "step": 5815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836321300439454e-06,
+      "loss": 0.0114,
+      "step": 5816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833066554515179e-06,
+      "loss": 0.0205,
+      "step": 5817
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.829812555120263e-06,
+      "loss": 0.0286,
+      "step": 5818
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826559302724856e-06,
+      "loss": 0.037,
+      "step": 5819
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.823306797798983e-06,
+      "loss": 0.0838,
+      "step": 5820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.82005504081258e-06,
+      "loss": 0.0331,
+      "step": 5821
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.816804032235466e-06,
+      "loss": 0.0372,
+      "step": 5822
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.813553772537354e-06,
+      "loss": 0.0444,
+      "step": 5823
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8103042621878515e-06,
+      "loss": 0.0077,
+      "step": 5824
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807055501656448e-06,
+      "loss": 0.0807,
+      "step": 5825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8038074914125345e-06,
+      "loss": 0.1295,
+      "step": 5826
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.800560231925391e-06,
+      "loss": 0.0899,
+      "step": 5827
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.797313723664186e-06,
+      "loss": 0.0055,
+      "step": 5828
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794067967097985e-06,
+      "loss": 0.0036,
+      "step": 5829
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.79082296269574e-06,
+      "loss": 0.0027,
+      "step": 5830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.787578710926296e-06,
+      "loss": 0.0017,
+      "step": 5831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.784335212258393e-06,
+      "loss": 0.0051,
+      "step": 5832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.781092467160655e-06,
+      "loss": 0.0744,
+      "step": 5833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777850476101597e-06,
+      "loss": 0.0037,
+      "step": 5834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774609239549632e-06,
+      "loss": 0.0026,
+      "step": 5835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771368757973062e-06,
+      "loss": 0.0312,
+      "step": 5836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768129031840076e-06,
+      "loss": 0.0198,
+      "step": 5837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764890061618759e-06,
+      "loss": 0.0524,
+      "step": 5838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7616518477770824e-06,
+      "loss": 0.0045,
+      "step": 5839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758414390782912e-06,
+      "loss": 0.0076,
+      "step": 5840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.755177691104004e-06,
+      "loss": 0.0602,
+      "step": 5841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751941749207996e-06,
+      "loss": 0.0104,
+      "step": 5842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.748706565562429e-06,
+      "loss": 0.03,
+      "step": 5843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.745472140634728e-06,
+      "loss": 0.0039,
+      "step": 5844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742238474892211e-06,
+      "loss": 0.2396,
+      "step": 5845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739005568802083e-06,
+      "loss": 0.0298,
+      "step": 5846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.735773422831446e-06,
+      "loss": 0.0848,
+      "step": 5847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732542037447278e-06,
+      "loss": 0.0788,
+      "step": 5848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.729311413116463e-06,
+      "loss": 0.0453,
+      "step": 5849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.72608155030577e-06,
+      "loss": 0.0474,
+      "step": 5850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722852449481848e-06,
+      "loss": 0.0452,
+      "step": 5851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7196241111112506e-06,
+      "loss": 0.0783,
+      "step": 5852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.716396535660412e-06,
+      "loss": 0.0717,
+      "step": 5853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713169723595662e-06,
+      "loss": 0.0092,
+      "step": 5854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.709943675383216e-06,
+      "loss": 0.0407,
+      "step": 5855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.706718391489179e-06,
+      "loss": 0.0054,
+      "step": 5856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.703493872379548e-06,
+      "loss": 0.0096,
+      "step": 5857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700270118520211e-06,
+      "loss": 0.0771,
+      "step": 5858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.697047130376937e-06,
+      "loss": 0.0562,
+      "step": 5859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693824908415393e-06,
+      "loss": 0.0025,
+      "step": 5860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.690603453101134e-06,
+      "loss": 0.0579,
+      "step": 5861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6873827648996e-06,
+      "loss": 0.0713,
+      "step": 5862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684162844276121e-06,
+      "loss": 0.1061,
+      "step": 5863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680943691695922e-06,
+      "loss": 0.0115,
+      "step": 5864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6777253076241105e-06,
+      "loss": 0.0764,
+      "step": 5865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674507692525685e-06,
+      "loss": 0.0649,
+      "step": 5866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6712908468655395e-06,
+      "loss": 0.0154,
+      "step": 5867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668074771108442e-06,
+      "loss": 0.0132,
+      "step": 5868
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6648594657190605e-06,
+      "loss": 0.0248,
+      "step": 5869
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66164493116195e-06,
+      "loss": 0.0108,
+      "step": 5870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658431167901553e-06,
+      "loss": 0.0055,
+      "step": 5871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655218176402202e-06,
+      "loss": 0.0234,
+      "step": 5872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6520059571281154e-06,
+      "loss": 0.0701,
+      "step": 5873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6487945105434005e-06,
+      "loss": 0.001,
+      "step": 5874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645583837112059e-06,
+      "loss": 0.0012,
+      "step": 5875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642373937297968e-06,
+      "loss": 0.0023,
+      "step": 5876
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639164811564908e-06,
+      "loss": 0.0786,
+      "step": 5877
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.635956460376532e-06,
+      "loss": 0.0633,
+      "step": 5878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.632748884196394e-06,
+      "loss": 0.0043,
+      "step": 5879
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629542083487931e-06,
+      "loss": 0.0034,
+      "step": 5880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626336058714468e-06,
+      "loss": 0.0591,
+      "step": 5881
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.623130810339219e-06,
+      "loss": 0.0447,
+      "step": 5882
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619926338825284e-06,
+      "loss": 0.0103,
+      "step": 5883
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616722644635651e-06,
+      "loss": 0.006,
+      "step": 5884
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613519728233204e-06,
+      "loss": 0.0223,
+      "step": 5885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610317590080694e-06,
+      "loss": 0.0392,
+      "step": 5886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.607116230640781e-06,
+      "loss": 0.0396,
+      "step": 5887
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603915650376002e-06,
+      "loss": 0.1168,
+      "step": 5888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600715849748787e-06,
+      "loss": 0.0144,
+      "step": 5889
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5975168292214445e-06,
+      "loss": 0.0295,
+      "step": 5890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.594318589256177e-06,
+      "loss": 0.0064,
+      "step": 5891
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5911211303150736e-06,
+      "loss": 0.02,
+      "step": 5892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587924452860111e-06,
+      "loss": 0.0131,
+      "step": 5893
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584728557353154e-06,
+      "loss": 0.031,
+      "step": 5894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581533444255945e-06,
+      "loss": 0.0133,
+      "step": 5895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.578339114030127e-06,
+      "loss": 0.0084,
+      "step": 5896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5751455671372205e-06,
+      "loss": 0.0032,
+      "step": 5897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571952804038635e-06,
+      "loss": 0.0012,
+      "step": 5898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568760825195671e-06,
+      "loss": 0.0144,
+      "step": 5899
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5655696310695105e-06,
+      "loss": 0.0079,
+      "step": 5900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.562379222121224e-06,
+      "loss": 0.0037,
+      "step": 5901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.559189598811773e-06,
+      "loss": 0.056,
+      "step": 5902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556000761601995e-06,
+      "loss": 0.0268,
+      "step": 5903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5528127109526186e-06,
+      "loss": 0.0126,
+      "step": 5904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.549625447324262e-06,
+      "loss": 0.0202,
+      "step": 5905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546438971177429e-06,
+      "loss": 0.0015,
+      "step": 5906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543253282972507e-06,
+      "loss": 0.0575,
+      "step": 5907
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5400683831697735e-06,
+      "loss": 0.0519,
+      "step": 5908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536884272229386e-06,
+      "loss": 0.0236,
+      "step": 5909
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.533700950611394e-06,
+      "loss": 0.0609,
+      "step": 5910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530518418775734e-06,
+      "loss": 0.0037,
+      "step": 5911
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527336677182216e-06,
+      "loss": 0.0872,
+      "step": 5912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524155726290549e-06,
+      "loss": 0.0032,
+      "step": 5913
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.520975566560324e-06,
+      "loss": 0.0018,
+      "step": 5914
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.517796198451016e-06,
+      "loss": 0.0039,
+      "step": 5915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.514617622421993e-06,
+      "loss": 0.0359,
+      "step": 5916
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.511439838932492e-06,
+      "loss": 0.0081,
+      "step": 5917
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508262848441651e-06,
+      "loss": 0.0067,
+      "step": 5918
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.505086651408489e-06,
+      "loss": 0.0057,
+      "step": 5919
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501911248291911e-06,
+      "loss": 0.1069,
+      "step": 5920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4987366395506996e-06,
+      "loss": 0.0012,
+      "step": 5921
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.495562825643533e-06,
+      "loss": 0.0372,
+      "step": 5922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492389807028971e-06,
+      "loss": 0.023,
+      "step": 5923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4892175841654564e-06,
+      "loss": 0.0139,
+      "step": 5924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.486046157511319e-06,
+      "loss": 0.1592,
+      "step": 5925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482875527524774e-06,
+      "loss": 0.032,
+      "step": 5926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.479705694663921e-06,
+      "loss": 0.0994,
+      "step": 5927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4765366593867454e-06,
+      "loss": 0.0908,
+      "step": 5928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.473368422151111e-06,
+      "loss": 0.0988,
+      "step": 5929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470200983414778e-06,
+      "loss": 0.0027,
+      "step": 5930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.467034343635377e-06,
+      "loss": 0.0113,
+      "step": 5931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463868503270435e-06,
+      "loss": 0.0038,
+      "step": 5932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460703462777359e-06,
+      "loss": 0.0316,
+      "step": 5933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457539222613439e-06,
+      "loss": 0.0504,
+      "step": 5934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454375783235855e-06,
+      "loss": 0.0383,
+      "step": 5935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.451213145101664e-06,
+      "loss": 0.0025,
+      "step": 5936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44805130866781e-06,
+      "loss": 0.01,
+      "step": 5937
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444890274391128e-06,
+      "loss": 0.0035,
+      "step": 5938
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441730042728323e-06,
+      "loss": 0.0402,
+      "step": 5939
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.438570614135994e-06,
+      "loss": 0.0054,
+      "step": 5940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435411989070624e-06,
+      "loss": 0.0021,
+      "step": 5941
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.432254167988576e-06,
+      "loss": 0.0106,
+      "step": 5942
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.429097151346098e-06,
+      "loss": 0.002,
+      "step": 5943
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425940939599329e-06,
+      "loss": 0.058,
+      "step": 5944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422785533204275e-06,
+      "loss": 0.1002,
+      "step": 5945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419630932616841e-06,
+      "loss": 0.0004,
+      "step": 5946
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4164771382928115e-06,
+      "loss": 0.0004,
+      "step": 5947
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4133241506878495e-06,
+      "loss": 0.0184,
+      "step": 5948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.410171970257508e-06,
+      "loss": 0.0005,
+      "step": 5949
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.407020597457218e-06,
+      "loss": 0.0041,
+      "step": 5950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403870032742299e-06,
+      "loss": 0.074,
+      "step": 5951
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40072027656795e-06,
+      "loss": 0.0647,
+      "step": 5952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397571329389255e-06,
+      "loss": 0.0041,
+      "step": 5953
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39442319166118e-06,
+      "loss": 0.0084,
+      "step": 5954
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3912758638385796e-06,
+      "loss": 0.0374,
+      "step": 5955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.388129346376177e-06,
+      "loss": 0.0592,
+      "step": 5956
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384983639728593e-06,
+      "loss": 0.0197,
+      "step": 5957
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381838744350328e-06,
+      "loss": 0.0459,
+      "step": 5958
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.378694660695756e-06,
+      "loss": 0.0681,
+      "step": 5959
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.375551389219146e-06,
+      "loss": 0.0455,
+      "step": 5960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372408930374642e-06,
+      "loss": 0.0011,
+      "step": 5961
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.369267284616275e-06,
+      "loss": 0.1901,
+      "step": 5962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366126452397957e-06,
+      "loss": 0.0126,
+      "step": 5963
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.362986434173483e-06,
+      "loss": 0.0046,
+      "step": 5964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359847230396525e-06,
+      "loss": 0.001,
+      "step": 5965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356708841520643e-06,
+      "loss": 0.0064,
+      "step": 5966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353571267999279e-06,
+      "loss": 0.0487,
+      "step": 5967
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.350434510285757e-06,
+      "loss": 0.0184,
+      "step": 5968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347298568833281e-06,
+      "loss": 0.0403,
+      "step": 5969
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344163444094939e-06,
+      "loss": 0.0036,
+      "step": 5970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341029136523701e-06,
+      "loss": 0.012,
+      "step": 5971
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337895646572421e-06,
+      "loss": 0.0073,
+      "step": 5972
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.334762974693828e-06,
+      "loss": 0.0504,
+      "step": 5973
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331631121340535e-06,
+      "loss": 0.0377,
+      "step": 5974
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328500086965041e-06,
+      "loss": 0.0021,
+      "step": 5975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.325369872019727e-06,
+      "loss": 0.0067,
+      "step": 5976
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322240476956848e-06,
+      "loss": 0.0391,
+      "step": 5977
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319111902228551e-06,
+      "loss": 0.0313,
+      "step": 5978
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315984148286855e-06,
+      "loss": 0.0521,
+      "step": 5979
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312857215583667e-06,
+      "loss": 0.0271,
+      "step": 5980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309731104570775e-06,
+      "loss": 0.0088,
+      "step": 5981
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306605815699838e-06,
+      "loss": 0.0022,
+      "step": 5982
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.303481349422411e-06,
+      "loss": 0.0859,
+      "step": 5983
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30035770618992e-06,
+      "loss": 0.0635,
+      "step": 5984
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297234886453682e-06,
+      "loss": 0.1773,
+      "step": 5985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2941128906648786e-06,
+      "loss": 0.0729,
+      "step": 5986
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.290991719274589e-06,
+      "loss": 0.0293,
+      "step": 5987
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287871372733763e-06,
+      "loss": 0.0147,
+      "step": 5988
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2847518514932374e-06,
+      "loss": 0.0491,
+      "step": 5989
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2816331560037306e-06,
+      "loss": 0.0099,
+      "step": 5990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.27851528671583e-06,
+      "loss": 0.0741,
+      "step": 5991
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275398244080015e-06,
+      "loss": 0.0558,
+      "step": 5992
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272282028546644e-06,
+      "loss": 0.0509,
+      "step": 5993
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.269166640565953e-06,
+      "loss": 0.0086,
+      "step": 5994
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266052080588062e-06,
+      "loss": 0.0248,
+      "step": 5995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262938349062967e-06,
+      "loss": 0.004,
+      "step": 5996
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259825446440547e-06,
+      "loss": 0.0072,
+      "step": 5997
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.256713373170565e-06,
+      "loss": 0.0844,
+      "step": 5998
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253602129702651e-06,
+      "loss": 0.1118,
+      "step": 5999
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250491716486333e-06,
+      "loss": 0.001,
+      "step": 6000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247382133971002e-06,
+      "loss": 0.0231,
+      "step": 6001
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.24427338260594e-06,
+      "loss": 0.0545,
+      "step": 6002
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241165462840306e-06,
+      "loss": 0.0122,
+      "step": 6003
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23805837512314e-06,
+      "loss": 0.0038,
+      "step": 6004
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23495211990336e-06,
+      "loss": 0.0021,
+      "step": 6005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.231846697629763e-06,
+      "loss": 0.012,
+      "step": 6006
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.228742108751029e-06,
+      "loss": 0.004,
+      "step": 6007
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225638353715719e-06,
+      "loss": 0.0006,
+      "step": 6008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2225354329722604e-06,
+      "loss": 0.0034,
+      "step": 6009
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2194333469689755e-06,
+      "loss": 0.0429,
+      "step": 6010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2163320961540585e-06,
+      "loss": 0.0327,
+      "step": 6011
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.213231680975586e-06,
+      "loss": 0.0437,
+      "step": 6012
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210132101881516e-06,
+      "loss": 0.007,
+      "step": 6013
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.207033359319673e-06,
+      "loss": 0.0342,
+      "step": 6014
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203935453737776e-06,
+      "loss": 0.008,
+      "step": 6015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.200838385583417e-06,
+      "loss": 0.0255,
+      "step": 6016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197742155304069e-06,
+      "loss": 0.0043,
+      "step": 6017
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194646763347074e-06,
+      "loss": 0.0644,
+      "step": 6018
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1915522101596664e-06,
+      "loss": 0.052,
+      "step": 6019
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188458496188952e-06,
+      "loss": 0.009,
+      "step": 6020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1853656218819175e-06,
+      "loss": 0.0339,
+      "step": 6021
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182273587685428e-06,
+      "loss": 0.0032,
+      "step": 6022
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179182394046228e-06,
+      "loss": 0.0148,
+      "step": 6023
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176092041410937e-06,
+      "loss": 0.0273,
+      "step": 6024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.173002530226063e-06,
+      "loss": 0.0173,
+      "step": 6025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169913860937975e-06,
+      "loss": 0.0052,
+      "step": 6026
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166826033992939e-06,
+      "loss": 0.0057,
+      "step": 6027
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163739049837082e-06,
+      "loss": 0.0046,
+      "step": 6028
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.160652908916422e-06,
+      "loss": 0.0339,
+      "step": 6029
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157567611676852e-06,
+      "loss": 0.0221,
+      "step": 6030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154483158564143e-06,
+      "loss": 0.067,
+      "step": 6031
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151399550023941e-06,
+      "loss": 0.1012,
+      "step": 6032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.148316786501774e-06,
+      "loss": 0.0082,
+      "step": 6033
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.145234868443048e-06,
+      "loss": 0.0097,
+      "step": 6034
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142153796293039e-06,
+      "loss": 0.0232,
+      "step": 6035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.13907357049691e-06,
+      "loss": 0.0271,
+      "step": 6036
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135994191499699e-06,
+      "loss": 0.0018,
+      "step": 6037
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132915659746321e-06,
+      "loss": 0.0081,
+      "step": 6038
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129837975681567e-06,
+      "loss": 0.0162,
+      "step": 6039
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126761139750109e-06,
+      "loss": 0.069,
+      "step": 6040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.123685152396498e-06,
+      "loss": 0.0992,
+      "step": 6041
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.120610014065151e-06,
+      "loss": 0.0043,
+      "step": 6042
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.11753572520038e-06,
+      "loss": 0.0012,
+      "step": 6043
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114462286246356e-06,
+      "loss": 0.0364,
+      "step": 6044
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111389697647138e-06,
+      "loss": 0.0783,
+      "step": 6045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1083179598466625e-06,
+      "loss": 0.0752,
+      "step": 6046
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105247073288739e-06,
+      "loss": 0.0306,
+      "step": 6047
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102177038417056e-06,
+      "loss": 0.063,
+      "step": 6048
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09910785567518e-06,
+      "loss": 0.0011,
+      "step": 6049
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09603952550655e-06,
+      "loss": 0.0305,
+      "step": 6050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.092972048354491e-06,
+      "loss": 0.0093,
+      "step": 6051
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.08990542466219e-06,
+      "loss": 0.0061,
+      "step": 6052
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.086839654872723e-06,
+      "loss": 0.007,
+      "step": 6053
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083774739429043e-06,
+      "loss": 0.0237,
+      "step": 6054
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080710678773967e-06,
+      "loss": 0.06,
+      "step": 6055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.077647473350201e-06,
+      "loss": 0.011,
+      "step": 6056
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0745851236003256e-06,
+      "loss": 0.0219,
+      "step": 6057
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071523629966792e-06,
+      "loss": 0.004,
+      "step": 6058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068462992891933e-06,
+      "loss": 0.0121,
+      "step": 6059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0654032128179554e-06,
+      "loss": 0.036,
+      "step": 6060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.062344290186947e-06,
+      "loss": 0.0985,
+      "step": 6061
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.059286225440859e-06,
+      "loss": 0.0456,
+      "step": 6062
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056229019021531e-06,
+      "loss": 0.0915,
+      "step": 6063
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053172671370675e-06,
+      "loss": 0.0072,
+      "step": 6064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050117182929879e-06,
+      "loss": 0.0147,
+      "step": 6065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047062554140605e-06,
+      "loss": 0.0188,
+      "step": 6066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044008785444193e-06,
+      "loss": 0.0061,
+      "step": 6067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.040955877281861e-06,
+      "loss": 0.0011,
+      "step": 6068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.037903830094693e-06,
+      "loss": 0.0155,
+      "step": 6069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034852644323661e-06,
+      "loss": 0.0109,
+      "step": 6070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0318023204096015e-06,
+      "loss": 0.0076,
+      "step": 6071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.028752858793234e-06,
+      "loss": 0.0903,
+      "step": 6072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.025704259915151e-06,
+      "loss": 0.0029,
+      "step": 6073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02265652421582e-06,
+      "loss": 0.0358,
+      "step": 6074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0196096521355866e-06,
+      "loss": 0.0515,
+      "step": 6075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0165636441146676e-06,
+      "loss": 0.0046,
+      "step": 6076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0135185005931555e-06,
+      "loss": 0.0398,
+      "step": 6077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010474222011025e-06,
+      "loss": 0.0797,
+      "step": 6078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007430808808111e-06,
+      "loss": 0.0385,
+      "step": 6079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0043882614241355e-06,
+      "loss": 0.0394,
+      "step": 6080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001346580298694e-06,
+      "loss": 0.0082,
+      "step": 6081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998305765871258e-06,
+      "loss": 0.0176,
+      "step": 6082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.995265818581161e-06,
+      "loss": 0.0527,
+      "step": 6083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9922267388676284e-06,
+      "loss": 0.0047,
+      "step": 6084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989188527169749e-06,
+      "loss": 0.0582,
+      "step": 6085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986151183926494e-06,
+      "loss": 0.0832,
+      "step": 6086
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983114709576706e-06,
+      "loss": 0.009,
+      "step": 6087
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.980079104559093e-06,
+      "loss": 0.0127,
+      "step": 6088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977044369312253e-06,
+      "loss": 0.0039,
+      "step": 6089
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974010504274647e-06,
+      "loss": 0.0053,
+      "step": 6090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970977509884617e-06,
+      "loss": 0.0499,
+      "step": 6091
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.967945386580374e-06,
+      "loss": 0.0092,
+      "step": 6092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964914134800007e-06,
+      "loss": 0.0794,
+      "step": 6093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.961883754981479e-06,
+      "loss": 0.0648,
+      "step": 6094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958854247562626e-06,
+      "loss": 0.0038,
+      "step": 6095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.955825612981156e-06,
+      "loss": 0.0243,
+      "step": 6096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.952797851674648e-06,
+      "loss": 0.0096,
+      "step": 6097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949770964080564e-06,
+      "loss": 0.0067,
+      "step": 6098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.946744950636233e-06,
+      "loss": 0.0104,
+      "step": 6099
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.943719811778862e-06,
+      "loss": 0.0027,
+      "step": 6100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940695547945528e-06,
+      "loss": 0.0016,
+      "step": 6101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937672159573184e-06,
+      "loss": 0.0023,
+      "step": 6102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.934649647098655e-06,
+      "loss": 0.0538,
+      "step": 6103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931628010958643e-06,
+      "loss": 0.1077,
+      "step": 6104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.928607251589714e-06,
+      "loss": 0.0392,
+      "step": 6105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925587369428316e-06,
+      "loss": 0.0012,
+      "step": 6106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922568364910769e-06,
+      "loss": 0.0083,
+      "step": 6107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.919550238473265e-06,
+      "loss": 0.0875,
+      "step": 6108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916532990551869e-06,
+      "loss": 0.0168,
+      "step": 6109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913516621582522e-06,
+      "loss": 0.0373,
+      "step": 6110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910501132001031e-06,
+      "loss": 0.0018,
+      "step": 6111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90748652224308e-06,
+      "loss": 0.0068,
+      "step": 6112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.904472792744233e-06,
+      "loss": 0.0078,
+      "step": 6113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90145994393991e-06,
+      "loss": 0.0048,
+      "step": 6114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8984479762654205e-06,
+      "loss": 0.0506,
+      "step": 6115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895436890155936e-06,
+      "loss": 0.002,
+      "step": 6116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.892426686046508e-06,
+      "loss": 0.0033,
+      "step": 6117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8894173643720544e-06,
+      "loss": 0.0354,
+      "step": 6118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886408925567371e-06,
+      "loss": 0.1103,
+      "step": 6119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.88340137006712e-06,
+      "loss": 0.0053,
+      "step": 6120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880394698305846e-06,
+      "loss": 0.0054,
+      "step": 6121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.87738891071795e-06,
+      "loss": 0.0162,
+      "step": 6122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874384007737721e-06,
+      "loss": 0.0487,
+      "step": 6123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.871379989799313e-06,
+      "loss": 0.0235,
+      "step": 6124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8683768573367484e-06,
+      "loss": 0.0243,
+      "step": 6125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8653746107839285e-06,
+      "loss": 0.0578,
+      "step": 6126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862373250574626e-06,
+      "loss": 0.0623,
+      "step": 6127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.859372777142482e-06,
+      "loss": 0.1314,
+      "step": 6128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85637319092101e-06,
+      "loss": 0.004,
+      "step": 6129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8533744923436e-06,
+      "loss": 0.0043,
+      "step": 6130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850376681843511e-06,
+      "loss": 0.049,
+      "step": 6131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8473797598538665e-06,
+      "loss": 0.0592,
+      "step": 6132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844383726807673e-06,
+      "loss": 0.0005,
+      "step": 6133
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8413885831378e-06,
+      "loss": 0.0295,
+      "step": 6134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.838394329276998e-06,
+      "loss": 0.0015,
+      "step": 6135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8354009656578775e-06,
+      "loss": 0.0013,
+      "step": 6136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832408492712931e-06,
+      "loss": 0.0023,
+      "step": 6137
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829416910874513e-06,
+      "loss": 0.0017,
+      "step": 6138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826426220574853e-06,
+      "loss": 0.1194,
+      "step": 6139
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.823436422246057e-06,
+      "loss": 0.0424,
+      "step": 6140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820447516320093e-06,
+      "loss": 0.0081,
+      "step": 6141
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817459503228804e-06,
+      "loss": 0.0553,
+      "step": 6142
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8144723834039076e-06,
+      "loss": 0.0034,
+      "step": 6143
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.811486157276987e-06,
+      "loss": 0.0011,
+      "step": 6144
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808500825279499e-06,
+      "loss": 0.0018,
+      "step": 6145
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8055163878427703e-06,
+      "loss": 0.0046,
+      "step": 6146
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8025328453979992e-06,
+      "loss": 0.0035,
+      "step": 6147
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799550198376257e-06,
+      "loss": 0.0033,
+      "step": 6148
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7965684472084775e-06,
+      "loss": 0.032,
+      "step": 6149
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.793587592325472e-06,
+      "loss": 0.0014,
+      "step": 6150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7906076341579235e-06,
+      "loss": 0.0026,
+      "step": 6151
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7876285731363784e-06,
+      "loss": 0.0288,
+      "step": 6152
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7846504096912597e-06,
+      "loss": 0.0202,
+      "step": 6153
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7816731442528577e-06,
+      "loss": 0.0037,
+      "step": 6154
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.778696777251335e-06,
+      "loss": 0.0018,
+      "step": 6155
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7757213091167243e-06,
+      "loss": 0.0565,
+      "step": 6156
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7727467402789284e-06,
+      "loss": 0.0138,
+      "step": 6157
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7697730711677148e-06,
+      "loss": 0.0037,
+      "step": 6158
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7668003022127274e-06,
+      "loss": 0.0577,
+      "step": 6159
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7638284338434784e-06,
+      "loss": 0.0004,
+      "step": 6160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7608574664893504e-06,
+      "loss": 0.0116,
+      "step": 6161
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.757887400579593e-06,
+      "loss": 0.0483,
+      "step": 6162
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75491823654333e-06,
+      "loss": 0.0561,
+      "step": 6163
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75194997480955e-06,
+      "loss": 0.1735,
+      "step": 6164
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7489826158071196e-06,
+      "loss": 0.0119,
+      "step": 6165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.746016159964764e-06,
+      "loss": 0.0361,
+      "step": 6166
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7430506077110794e-06,
+      "loss": 0.0846,
+      "step": 6167
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7400859594745388e-06,
+      "loss": 0.0015,
+      "step": 6168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.737122215683482e-06,
+      "loss": 0.0052,
+      "step": 6169
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7341593767661144e-06,
+      "loss": 0.0016,
+      "step": 6170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.731197443150515e-06,
+      "loss": 0.0014,
+      "step": 6171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.72823641526463e-06,
+      "loss": 0.0362,
+      "step": 6172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.725276293536274e-06,
+      "loss": 0.0128,
+      "step": 6173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.722317078393136e-06,
+      "loss": 0.0087,
+      "step": 6174
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7193587702627632e-06,
+      "loss": 0.0041,
+      "step": 6175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7164013695725798e-06,
+      "loss": 0.0402,
+      "step": 6176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7134448767498775e-06,
+      "loss": 0.0024,
+      "step": 6177
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7104892922218184e-06,
+      "loss": 0.0008,
+      "step": 6178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7075346164154335e-06,
+      "loss": 0.0007,
+      "step": 6179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.704580849757614e-06,
+      "loss": 0.056,
+      "step": 6180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.70162799267513e-06,
+      "loss": 0.069,
+      "step": 6181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.698676045594616e-06,
+      "loss": 0.0051,
+      "step": 6182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6957250089425754e-06,
+      "loss": 0.0018,
+      "step": 6183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6927748831453835e-06,
+      "loss": 0.078,
+      "step": 6184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6898256686292753e-06,
+      "loss": 0.0325,
+      "step": 6185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6868773658203617e-06,
+      "loss": 0.0019,
+      "step": 6186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.683929975144619e-06,
+      "loss": 0.0025,
+      "step": 6187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6809834970278934e-06,
+      "loss": 0.0135,
+      "step": 6188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.678037931895898e-06,
+      "loss": 0.0128,
+      "step": 6189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6750932801742135e-06,
+      "loss": 0.1179,
+      "step": 6190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6721495422882948e-06,
+      "loss": 0.036,
+      "step": 6191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6692067186634495e-06,
+      "loss": 0.0354,
+      "step": 6192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6662648097248723e-06,
+      "loss": 0.0059,
+      "step": 6193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6633238158976083e-06,
+      "loss": 0.0482,
+      "step": 6194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6603837376065822e-06,
+      "loss": 0.0153,
+      "step": 6195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.657444575276583e-06,
+      "loss": 0.0248,
+      "step": 6196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.654506329332266e-06,
+      "loss": 0.0071,
+      "step": 6197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6515690001981563e-06,
+      "loss": 0.0332,
+      "step": 6198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.648632588298644e-06,
+      "loss": 0.0027,
+      "step": 6199
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6456970940579895e-06,
+      "loss": 0.1249,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6427625179003223e-06,
+      "loss": 0.0386,
+      "step": 6201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6398288602496278e-06,
+      "loss": 0.0033,
+      "step": 6202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6368961215297726e-06,
+      "loss": 0.003,
+      "step": 6203
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6339643021644834e-06,
+      "loss": 0.0093,
+      "step": 6204
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6310334025773553e-06,
+      "loss": 0.0698,
+      "step": 6205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6281034231918554e-06,
+      "loss": 0.0792,
+      "step": 6206
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6251743644313064e-06,
+      "loss": 0.0061,
+      "step": 6207
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6222462267189074e-06,
+      "loss": 0.0654,
+      "step": 6208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.619319010477722e-06,
+      "loss": 0.0038,
+      "step": 6209
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6163927161306854e-06,
+      "loss": 0.0243,
+      "step": 6210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6134673441005864e-06,
+      "loss": 0.0035,
+      "step": 6211
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6105428948100917e-06,
+      "loss": 0.0024,
+      "step": 6212
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6076193686817327e-06,
+      "loss": 0.0186,
+      "step": 6213
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6046967661379074e-06,
+      "loss": 0.0055,
+      "step": 6214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6017750876008796e-06,
+      "loss": 0.0217,
+      "step": 6215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5988543334927773e-06,
+      "loss": 0.0045,
+      "step": 6216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.595934504235599e-06,
+      "loss": 0.0138,
+      "step": 6217
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5930156002512105e-06,
+      "loss": 0.0396,
+      "step": 6218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.590097621961335e-06,
+      "loss": 0.0025,
+      "step": 6219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.587180569787575e-06,
+      "loss": 0.0227,
+      "step": 6220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.584264444151384e-06,
+      "loss": 0.0041,
+      "step": 6221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.581349245474095e-06,
+      "loss": 0.1207,
+      "step": 6222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578434974176901e-06,
+      "loss": 0.0613,
+      "step": 6223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5755216306808614e-06,
+      "loss": 0.0116,
+      "step": 6224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.572609215406905e-06,
+      "loss": 0.014,
+      "step": 6225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5696977287758205e-06,
+      "loss": 0.0186,
+      "step": 6226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.56678717120827e-06,
+      "loss": 0.0155,
+      "step": 6227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.563877543124772e-06,
+      "loss": 0.0183,
+      "step": 6228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.560968844945716e-06,
+      "loss": 0.0008,
+      "step": 6229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5580610770913593e-06,
+      "loss": 0.0769,
+      "step": 6230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5551542399818204e-06,
+      "loss": 0.0747,
+      "step": 6231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5522483340370863e-06,
+      "loss": 0.0485,
+      "step": 6232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5493433596770088e-06,
+      "loss": 0.0018,
+      "step": 6233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546439317321306e-06,
+      "loss": 0.0291,
+      "step": 6234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543536207389556e-06,
+      "loss": 0.0118,
+      "step": 6235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5406340303012075e-06,
+      "loss": 0.0049,
+      "step": 6236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5377327864755774e-06,
+      "loss": 0.0048,
+      "step": 6237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5348324763318363e-06,
+      "loss": 0.0028,
+      "step": 6238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.531933100289031e-06,
+      "loss": 0.0351,
+      "step": 6239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5290346587660695e-06,
+      "loss": 0.0045,
+      "step": 6240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5261371521817247e-06,
+      "loss": 0.0099,
+      "step": 6241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.523240580954632e-06,
+      "loss": 0.0507,
+      "step": 6242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.520344945503298e-06,
+      "loss": 0.0114,
+      "step": 6243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5174502462460914e-06,
+      "loss": 0.0036,
+      "step": 6244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.514556483601237e-06,
+      "loss": 0.0133,
+      "step": 6245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.511663657986837e-06,
+      "loss": 0.0033,
+      "step": 6246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5087717698208513e-06,
+      "loss": 0.0087,
+      "step": 6247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5058808195211092e-06,
+      "loss": 0.0186,
+      "step": 6248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5029908075052966e-06,
+      "loss": 0.0042,
+      "step": 6249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5001017341909694e-06,
+      "loss": 0.1192,
+      "step": 6250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.497213599995549e-06,
+      "loss": 0.0071,
+      "step": 6251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.494326405336318e-06,
+      "loss": 0.3546,
+      "step": 6252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.491440150630423e-06,
+      "loss": 0.0313,
+      "step": 6253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.488554836294882e-06,
+      "loss": 0.1576,
+      "step": 6254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.485670462746563e-06,
+      "loss": 0.0227,
+      "step": 6255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4827870304022116e-06,
+      "loss": 0.004,
+      "step": 6256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4799045396784293e-06,
+      "loss": 0.0081,
+      "step": 6257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.477022990991685e-06,
+      "loss": 0.0469,
+      "step": 6258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4741423847583134e-06,
+      "loss": 0.0101,
+      "step": 6259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4712627213945084e-06,
+      "loss": 0.0022,
+      "step": 6260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.468384001316333e-06,
+      "loss": 0.0023,
+      "step": 6261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.465506224939704e-06,
+      "loss": 0.0263,
+      "step": 6262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4626293926804165e-06,
+      "loss": 0.1295,
+      "step": 6263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4597535049541142e-06,
+      "loss": 0.0801,
+      "step": 6264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4568785621763136e-06,
+      "loss": 0.01,
+      "step": 6265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.454004564762393e-06,
+      "loss": 0.0023,
+      "step": 6266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4511315131275947e-06,
+      "loss": 0.0697,
+      "step": 6267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.448259407687021e-06,
+      "loss": 0.0032,
+      "step": 6268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4453882488556424e-06,
+      "loss": 0.0995,
+      "step": 6269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.442518037048287e-06,
+      "loss": 0.0289,
+      "step": 6270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4396487726796545e-06,
+      "loss": 0.0333,
+      "step": 6271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4367804561642957e-06,
+      "loss": 0.0032,
+      "step": 6272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.433913087916633e-06,
+      "loss": 0.0173,
+      "step": 6273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.431046668350949e-06,
+      "loss": 0.0197,
+      "step": 6274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4281811978813928e-06,
+      "loss": 0.0399,
+      "step": 6275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4253166769219746e-06,
+      "loss": 0.0023,
+      "step": 6276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4224531058865605e-06,
+      "loss": 0.0066,
+      "step": 6277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4195904851888896e-06,
+      "loss": 0.0016,
+      "step": 6278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.416728815242557e-06,
+      "loss": 0.0547,
+      "step": 6279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.413868096461028e-06,
+      "loss": 0.0977,
+      "step": 6280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.411008329257618e-06,
+      "loss": 0.1196,
+      "step": 6281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4081495140455156e-06,
+      "loss": 0.0017,
+      "step": 6282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4052916512377678e-06,
+      "loss": 0.0021,
+      "step": 6283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4024347412472857e-06,
+      "loss": 0.0216,
+      "step": 6284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.39957878448684e-06,
+      "loss": 0.0182,
+      "step": 6285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.396723781369067e-06,
+      "loss": 0.0587,
+      "step": 6286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.393869732306463e-06,
+      "loss": 0.1517,
+      "step": 6287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3910166377113894e-06,
+      "loss": 0.0013,
+      "step": 6288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.388164497996065e-06,
+      "loss": 0.0482,
+      "step": 6289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.385313313572569e-06,
+      "loss": 0.053,
+      "step": 6290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3824630848528505e-06,
+      "loss": 0.0006,
+      "step": 6291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3796138122487155e-06,
+      "loss": 0.069,
+      "step": 6292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3767654961718344e-06,
+      "loss": 0.0138,
+      "step": 6293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.373918137033736e-06,
+      "loss": 0.0004,
+      "step": 6294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3710717352458143e-06,
+      "loss": 0.0088,
+      "step": 6295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.368226291219323e-06,
+      "loss": 0.0619,
+      "step": 6296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.36538180536538e-06,
+      "loss": 0.0203,
+      "step": 6297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3625382780949576e-06,
+      "loss": 0.0167,
+      "step": 6298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3596957098188964e-06,
+      "loss": 0.09,
+      "step": 6299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.356854100947897e-06,
+      "loss": 0.0551,
+      "step": 6300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3540134518925226e-06,
+      "loss": 0.0629,
+      "step": 6301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3511737630631933e-06,
+      "loss": 0.1621,
+      "step": 6302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3483350348701993e-06,
+      "loss": 0.0092,
+      "step": 6303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3454972677236773e-06,
+      "loss": 0.0078,
+      "step": 6304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3426604620336388e-06,
+      "loss": 0.0351,
+      "step": 6305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.33982461820995e-06,
+      "loss": 0.0048,
+      "step": 6306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.336989736662345e-06,
+      "loss": 0.1021,
+      "step": 6307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.334155817800404e-06,
+      "loss": 0.0291,
+      "step": 6308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3313228620335826e-06,
+      "loss": 0.0625,
+      "step": 6309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3284908697711903e-06,
+      "loss": 0.0005,
+      "step": 6310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3256598414224027e-06,
+      "loss": 0.0809,
+      "step": 6311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3228297773962493e-06,
+      "loss": 0.0281,
+      "step": 6312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.320000678101626e-06,
+      "loss": 0.0007,
+      "step": 6313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.317172543947289e-06,
+      "loss": 0.099,
+      "step": 6314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3143453753418465e-06,
+      "loss": 0.0288,
+      "step": 6315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311519172693778e-06,
+      "loss": 0.0744,
+      "step": 6316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 0.0332,
+      "step": 6317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3058696669029666e-06,
+      "loss": 0.0025,
+      "step": 6318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3030463645764733e-06,
+      "loss": 0.1001,
+      "step": 6319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.300224029839858e-06,
+      "loss": 0.0614,
+      "step": 6320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.297402663100897e-06,
+      "loss": 0.0618,
+      "step": 6321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2945822647672287e-06,
+      "loss": 0.0035,
+      "step": 6322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2917628352463493e-06,
+      "loss": 0.0672,
+      "step": 6323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2889443749456185e-06,
+      "loss": 0.064,
+      "step": 6324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.286126884272248e-06,
+      "loss": 0.0715,
+      "step": 6325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2833103636333176e-06,
+      "loss": 0.0116,
+      "step": 6326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.280494813435764e-06,
+      "loss": 0.0691,
+      "step": 6327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2776802340863834e-06,
+      "loss": 0.0107,
+      "step": 6328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2748666259918325e-06,
+      "loss": 0.0017,
+      "step": 6329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2720539895586265e-06,
+      "loss": 0.0315,
+      "step": 6330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.269242325193147e-06,
+      "loss": 0.0062,
+      "step": 6331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2664316333016187e-06,
+      "loss": 0.0836,
+      "step": 6332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.263621914290146e-06,
+      "loss": 0.0033,
+      "step": 6333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.260813168564675e-06,
+      "loss": 0.0102,
+      "step": 6334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.258005396531022e-06,
+      "loss": 0.0065,
+      "step": 6335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.255198598594862e-06,
+      "loss": 0.0104,
+      "step": 6336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2523927751617233e-06,
+      "loss": 0.0171,
+      "step": 6337
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.249587926637001e-06,
+      "loss": 0.0261,
+      "step": 6338
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2467840534259433e-06,
+      "loss": 0.106,
+      "step": 6339
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.24398115593366e-06,
+      "loss": 0.0089,
+      "step": 6340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241179234565123e-06,
+      "loss": 0.0024,
+      "step": 6341
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2383782897251526e-06,
+      "loss": 0.0391,
+      "step": 6342
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.235578321818439e-06,
+      "loss": 0.0084,
+      "step": 6343
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.232779331249528e-06,
+      "loss": 0.0429,
+      "step": 6344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2299813184228268e-06,
+      "loss": 0.0188,
+      "step": 6345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2271842837425917e-06,
+      "loss": 0.0544,
+      "step": 6346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.224388227612946e-06,
+      "loss": 0.0282,
+      "step": 6347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2215931504378704e-06,
+      "loss": 0.032,
+      "step": 6348
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2187990526212033e-06,
+      "loss": 0.0054,
+      "step": 6349
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.216005934566646e-06,
+      "loss": 0.0345,
+      "step": 6350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2132137966777476e-06,
+      "loss": 0.0622,
+      "step": 6351
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2104226393579242e-06,
+      "loss": 0.0477,
+      "step": 6352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2076324630104483e-06,
+      "loss": 0.0045,
+      "step": 6353
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.204843268038451e-06,
+      "loss": 0.0096,
+      "step": 6354
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2020550548449213e-06,
+      "loss": 0.0134,
+      "step": 6355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199267823832705e-06,
+      "loss": 0.0033,
+      "step": 6356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1964815754045073e-06,
+      "loss": 0.0044,
+      "step": 6357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.193696309962895e-06,
+      "loss": 0.0847,
+      "step": 6358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1909120279102812e-06,
+      "loss": 0.0865,
+      "step": 6359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.188128729648954e-06,
+      "loss": 0.0003,
+      "step": 6360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185346415581041e-06,
+      "loss": 0.2544,
+      "step": 6361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.182565086108541e-06,
+      "loss": 0.0594,
+      "step": 6362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.179784741633307e-06,
+      "loss": 0.0105,
+      "step": 6363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.177005382557048e-06,
+      "loss": 0.0127,
+      "step": 6364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.174227009281332e-06,
+      "loss": 0.0524,
+      "step": 6365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171449622207584e-06,
+      "loss": 0.0321,
+      "step": 6366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1686732217370898e-06,
+      "loss": 0.0133,
+      "step": 6367
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1658978082709833e-06,
+      "loss": 0.0425,
+      "step": 6368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1631233822102668e-06,
+      "loss": 0.008,
+      "step": 6369
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.160349943955793e-06,
+      "loss": 0.0164,
+      "step": 6370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.157577493908274e-06,
+      "loss": 0.1079,
+      "step": 6371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1548060324682827e-06,
+      "loss": 0.0821,
+      "step": 6372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1520355600362406e-06,
+      "loss": 0.0525,
+      "step": 6373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1492660770124337e-06,
+      "loss": 0.0032,
+      "step": 6374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1464975837970035e-06,
+      "loss": 0.0051,
+      "step": 6375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1437300807899462e-06,
+      "loss": 0.0212,
+      "step": 6376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.14096356839112e-06,
+      "loss": 0.0036,
+      "step": 6377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1381980470002305e-06,
+      "loss": 0.0083,
+      "step": 6378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1354335170168483e-06,
+      "loss": 0.0395,
+      "step": 6379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1326699788404e-06,
+      "loss": 0.0378,
+      "step": 6380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.129907432870166e-06,
+      "loss": 0.1085,
+      "step": 6381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1271458795052867e-06,
+      "loss": 0.0534,
+      "step": 6382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.124385319144755e-06,
+      "loss": 0.0013,
+      "step": 6383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.121625752187425e-06,
+      "loss": 0.0048,
+      "step": 6384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1188671790320013e-06,
+      "loss": 0.0767,
+      "step": 6385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.116109600077052e-06,
+      "loss": 0.055,
+      "step": 6386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1133530157209936e-06,
+      "loss": 0.0258,
+      "step": 6387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.110597426362104e-06,
+      "loss": 0.0014,
+      "step": 6388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.107842832398519e-06,
+      "loss": 0.0136,
+      "step": 6389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105089234228227e-06,
+      "loss": 0.0013,
+      "step": 6390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1023366322490733e-06,
+      "loss": 0.021,
+      "step": 6391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.099585026858759e-06,
+      "loss": 0.0044,
+      "step": 6392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.096834418454844e-06,
+      "loss": 0.0849,
+      "step": 6393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.094084807434743e-06,
+      "loss": 0.0087,
+      "step": 6394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091336194195721e-06,
+      "loss": 0.0005,
+      "step": 6395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0885885791349045e-06,
+      "loss": 0.061,
+      "step": 6396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0858419626492763e-06,
+      "loss": 0.0041,
+      "step": 6397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.083096345135672e-06,
+      "loss": 0.0448,
+      "step": 6398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0803517269907856e-06,
+      "loss": 0.0248,
+      "step": 6399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0776081086111687e-06,
+      "loss": 0.0058,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0748654903932164e-06,
+      "loss": 0.0474,
+      "step": 6401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0721238727331925e-06,
+      "loss": 0.0023,
+      "step": 6402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.069383256027215e-06,
+      "loss": 0.0276,
+      "step": 6403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0666436406712485e-06,
+      "loss": 0.003,
+      "step": 6404
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.063905027061119e-06,
+      "loss": 0.014,
+      "step": 6405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0611674155925085e-06,
+      "loss": 0.0205,
+      "step": 6406
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0584308066609537e-06,
+      "loss": 0.0075,
+      "step": 6407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0556952006618446e-06,
+      "loss": 0.0775,
+      "step": 6408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0529605979904275e-06,
+      "loss": 0.0176,
+      "step": 6409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.050226999041803e-06,
+      "loss": 0.0028,
+      "step": 6410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.047494404210931e-06,
+      "loss": 0.004,
+      "step": 6411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0447628138926153e-06,
+      "loss": 0.0233,
+      "step": 6412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0420322284815275e-06,
+      "loss": 0.0019,
+      "step": 6413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0393026483721877e-06,
+      "loss": 0.0206,
+      "step": 6414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0365740739589677e-06,
+      "loss": 0.0049,
+      "step": 6415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0338465056361e-06,
+      "loss": 0.0175,
+      "step": 6416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.03111994379767e-06,
+      "loss": 0.0805,
+      "step": 6417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0283943888376157e-06,
+      "loss": 0.0065,
+      "step": 6418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.025669841149732e-06,
+      "loss": 0.0052,
+      "step": 6419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0229463011276703e-06,
+      "loss": 0.0205,
+      "step": 6420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0202237691649263e-06,
+      "loss": 0.0048,
+      "step": 6421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017502245654862e-06,
+      "loss": 0.0679,
+      "step": 6422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0147817309906867e-06,
+      "loss": 0.049,
+      "step": 6423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0120622255654673e-06,
+      "loss": 0.0305,
+      "step": 6424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.009343729772124e-06,
+      "loss": 0.0608,
+      "step": 6425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.00662624400343e-06,
+      "loss": 0.0024,
+      "step": 6426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0039097686520134e-06,
+      "loss": 0.0241,
+      "step": 6427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0011943041103585e-06,
+      "loss": 0.0376,
+      "step": 6428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.998479850770798e-06,
+      "loss": 0.0029,
+      "step": 6429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9957664090255247e-06,
+      "loss": 0.0448,
+      "step": 6430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9930539792665767e-06,
+      "loss": 0.0054,
+      "step": 6431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9903425618858572e-06,
+      "loss": 0.0027,
+      "step": 6432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9876321572751143e-06,
+      "loss": 0.0336,
+      "step": 6433
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.984922765825955e-06,
+      "loss": 0.0064,
+      "step": 6434
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.982214387929837e-06,
+      "loss": 0.0185,
+      "step": 6435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9795070239780734e-06,
+      "loss": 0.0353,
+      "step": 6436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9768006743618306e-06,
+      "loss": 0.0008,
+      "step": 6437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.974095339472124e-06,
+      "loss": 0.0356,
+      "step": 6438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9713910196998273e-06,
+      "loss": 0.0133,
+      "step": 6439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9686877154356676e-06,
+      "loss": 0.0029,
+      "step": 6440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.965985427070226e-06,
+      "loss": 0.0178,
+      "step": 6441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9632841549939296e-06,
+      "loss": 0.0066,
+      "step": 6442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.960583899597066e-06,
+      "loss": 0.0316,
+      "step": 6443
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9578846612697743e-06,
+      "loss": 0.0531,
+      "step": 6444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.955186440402046e-06,
+      "loss": 0.0086,
+      "step": 6445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9524892373837253e-06,
+      "loss": 0.0081,
+      "step": 6446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.949793052604515e-06,
+      "loss": 0.0134,
+      "step": 6447
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9470978864539557e-06,
+      "loss": 0.0013,
+      "step": 6448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9444037393214565e-06,
+      "loss": 0.0041,
+      "step": 6449
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.941710611596271e-06,
+      "loss": 0.0556,
+      "step": 6450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.939018503667511e-06,
+      "loss": 0.0358,
+      "step": 6451
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.936327415924136e-06,
+      "loss": 0.0016,
+      "step": 6452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9336373487549606e-06,
+      "loss": 0.008,
+      "step": 6453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9309483025486527e-06,
+      "loss": 0.0159,
+      "step": 6454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9282602776937287e-06,
+      "loss": 0.0789,
+      "step": 6455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9255732745785627e-06,
+      "loss": 0.0513,
+      "step": 6456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9228872935913742e-06,
+      "loss": 0.0688,
+      "step": 6457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9202023351202415e-06,
+      "loss": 0.0646,
+      "step": 6458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9175183995530943e-06,
+      "loss": 0.0024,
+      "step": 6459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9148354872777117e-06,
+      "loss": 0.0321,
+      "step": 6460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9121535986817284e-06,
+      "loss": 0.0149,
+      "step": 6461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9094727341526275e-06,
+      "loss": 0.1186,
+      "step": 6462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.906792894077746e-06,
+      "loss": 0.0014,
+      "step": 6463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9041140788442766e-06,
+      "loss": 0.0135,
+      "step": 6464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9014362888392544e-06,
+      "loss": 0.0398,
+      "step": 6465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8987595244495736e-06,
+      "loss": 0.0698,
+      "step": 6466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8960837860619807e-06,
+      "loss": 0.0021,
+      "step": 6467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8934090740630703e-06,
+      "loss": 0.0067,
+      "step": 6468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8907353888392955e-06,
+      "loss": 0.0881,
+      "step": 6469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.888062730776947e-06,
+      "loss": 0.0795,
+      "step": 6470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8853911002621803e-06,
+      "loss": 0.0011,
+      "step": 6471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8827204976809997e-06,
+      "loss": 0.0002,
+      "step": 6472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8800509234192607e-06,
+      "loss": 0.0007,
+      "step": 6473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.877382377862663e-06,
+      "loss": 0.0824,
+      "step": 6474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874714861396768e-06,
+      "loss": 0.0068,
+      "step": 6475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8720483744069817e-06,
+      "loss": 0.073,
+      "step": 6476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.869382917278566e-06,
+      "loss": 0.0022,
+      "step": 6477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8667184903966316e-06,
+      "loss": 0.0193,
+      "step": 6478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8640550941461388e-06,
+      "loss": 0.0561,
+      "step": 6479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8613927289119014e-06,
+      "loss": 0.0271,
+      "step": 6480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8587313950785876e-06,
+      "loss": 0.0434,
+      "step": 6481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8560710930307055e-06,
+      "loss": 0.0375,
+      "step": 6482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8534118231526263e-06,
+      "loss": 0.0154,
+      "step": 6483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850753585828563e-06,
+      "loss": 0.0354,
+      "step": 6484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8480963814425842e-06,
+      "loss": 0.0025,
+      "step": 6485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8454402103786096e-06,
+      "loss": 0.1585,
+      "step": 6486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8427850730204076e-06,
+      "loss": 0.0318,
+      "step": 6487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8401309697515987e-06,
+      "loss": 0.0881,
+      "step": 6488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.837477900955653e-06,
+      "loss": 0.1313,
+      "step": 6489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8348258670158944e-06,
+      "loss": 0.0243,
+      "step": 6490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8321748683154893e-06,
+      "loss": 0.0169,
+      "step": 6491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.829524905237461e-06,
+      "loss": 0.0012,
+      "step": 6492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8268759781646817e-06,
+      "loss": 0.0368,
+      "step": 6493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8242280874798746e-06,
+      "loss": 0.0487,
+      "step": 6494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8215812335656124e-06,
+      "loss": 0.0013,
+      "step": 6495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8189354168043183e-06,
+      "loss": 0.0421,
+      "step": 6496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8162906375782684e-06,
+      "loss": 0.0085,
+      "step": 6497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8136468962695796e-06,
+      "loss": 0.0366,
+      "step": 6498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.811004193260228e-06,
+      "loss": 0.0459,
+      "step": 6499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8083625289320417e-06,
+      "loss": 0.0033,
+      "step": 6500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.805721903666685e-06,
+      "loss": 0.008,
+      "step": 6501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8030823178456858e-06,
+      "loss": 0.0401,
+      "step": 6502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8004437718504172e-06,
+      "loss": 0.0023,
+      "step": 6503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7978062660620996e-06,
+      "loss": 0.073,
+      "step": 6504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.795169800861809e-06,
+      "loss": 0.0364,
+      "step": 6505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7925343766304637e-06,
+      "loss": 0.0484,
+      "step": 6506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7898999937488415e-06,
+      "loss": 0.1094,
+      "step": 6507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7872666525975543e-06,
+      "loss": 0.0019,
+      "step": 6508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7846343535570787e-06,
+      "loss": 0.0641,
+      "step": 6509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.782003097007733e-06,
+      "loss": 0.0061,
+      "step": 6510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7793728833296906e-06,
+      "loss": 0.053,
+      "step": 6511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.776743712902964e-06,
+      "loss": 0.0016,
+      "step": 6512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774115586107423e-06,
+      "loss": 0.0017,
+      "step": 6513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.771488503322787e-06,
+      "loss": 0.001,
+      "step": 6514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768862464928621e-06,
+      "loss": 0.004,
+      "step": 6515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7662374713043414e-06,
+      "loss": 0.1462,
+      "step": 6516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.763613522829216e-06,
+      "loss": 0.0056,
+      "step": 6517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76099061988235e-06,
+      "loss": 0.0007,
+      "step": 6518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7583687628427115e-06,
+      "loss": 0.0018,
+      "step": 6519
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7557479520891104e-06,
+      "loss": 0.0034,
+      "step": 6520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.753128188000208e-06,
+      "loss": 0.0055,
+      "step": 6521
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7505094709545134e-06,
+      "loss": 0.0241,
+      "step": 6522
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7478918013303833e-06,
+      "loss": 0.0012,
+      "step": 6523
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7452751795060273e-06,
+      "loss": 0.1259,
+      "step": 6524
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7426596058594944e-06,
+      "loss": 0.0034,
+      "step": 6525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.740045080768694e-06,
+      "loss": 0.0014,
+      "step": 6526
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7374316046113737e-06,
+      "loss": 0.001,
+      "step": 6527
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7348191777651347e-06,
+      "loss": 0.0407,
+      "step": 6528
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7322078006074283e-06,
+      "loss": 0.0152,
+      "step": 6529
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7295974735155493e-06,
+      "loss": 0.003,
+      "step": 6530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7269881968666446e-06,
+      "loss": 0.0044,
+      "step": 6531
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.724379971037708e-06,
+      "loss": 0.0865,
+      "step": 6532
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7217727964055805e-06,
+      "loss": 0.0023,
+      "step": 6533
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.719166673346957e-06,
+      "loss": 0.0066,
+      "step": 6534
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.716561602238368e-06,
+      "loss": 0.0658,
+      "step": 6535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.713957583456203e-06,
+      "loss": 0.0045,
+      "step": 6536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.711354617376696e-06,
+      "loss": 0.0015,
+      "step": 6537
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.708752704375932e-06,
+      "loss": 0.0013,
+      "step": 6538
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.706151844829834e-06,
+      "loss": 0.0162,
+      "step": 6539
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7035520391141856e-06,
+      "loss": 0.0446,
+      "step": 6540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7009532876046084e-06,
+      "loss": 0.0073,
+      "step": 6541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.698355590676578e-06,
+      "loss": 0.1479,
+      "step": 6542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.695758948705418e-06,
+      "loss": 0.0842,
+      "step": 6543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.693163362066289e-06,
+      "loss": 0.146,
+      "step": 6544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6905688311342105e-06,
+      "loss": 0.0578,
+      "step": 6545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6879753562840473e-06,
+      "loss": 0.0695,
+      "step": 6546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6853829378905073e-06,
+      "loss": 0.0157,
+      "step": 6547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.682791576328151e-06,
+      "loss": 0.015,
+      "step": 6548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.680201271971383e-06,
+      "loss": 0.0146,
+      "step": 6549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.677612025194455e-06,
+      "loss": 0.0842,
+      "step": 6550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.675023836371471e-06,
+      "loss": 0.0825,
+      "step": 6551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6724367058763713e-06,
+      "loss": 0.057,
+      "step": 6552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6698506340829554e-06,
+      "loss": 0.0014,
+      "step": 6553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6672656213648594e-06,
+      "loss": 0.0013,
+      "step": 6554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6646816680955733e-06,
+      "loss": 0.0858,
+      "step": 6555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.662098774648434e-06,
+      "loss": 0.044,
+      "step": 6556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6595169413966215e-06,
+      "loss": 0.0674,
+      "step": 6557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6569361687131643e-06,
+      "loss": 0.0496,
+      "step": 6558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.654356456970939e-06,
+      "loss": 0.0619,
+      "step": 6559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6517778065426715e-06,
+      "loss": 0.0185,
+      "step": 6560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6492002178009224e-06,
+      "loss": 0.0064,
+      "step": 6561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.646623691118111e-06,
+      "loss": 0.0068,
+      "step": 6562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6440482268664993e-06,
+      "loss": 0.1503,
+      "step": 6563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.641473825418197e-06,
+      "loss": 0.0574,
+      "step": 6564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6389004871451573e-06,
+      "loss": 0.0778,
+      "step": 6565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6363282124191847e-06,
+      "loss": 0.0059,
+      "step": 6566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633757001611922e-06,
+      "loss": 0.0056,
+      "step": 6567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6311868550948636e-06,
+      "loss": 0.0062,
+      "step": 6568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.628617773239354e-06,
+      "loss": 0.0034,
+      "step": 6569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6260497564165776e-06,
+      "loss": 0.0601,
+      "step": 6570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6234828049975645e-06,
+      "loss": 0.0362,
+      "step": 6571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6209169193531948e-06,
+      "loss": 0.0065,
+      "step": 6572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6183520998541923e-06,
+      "loss": 0.0185,
+      "step": 6573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6157883468711296e-06,
+      "loss": 0.0393,
+      "step": 6574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.613225660774421e-06,
+      "loss": 0.0031,
+      "step": 6575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61066404193433e-06,
+      "loss": 0.0228,
+      "step": 6576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6081034907209667e-06,
+      "loss": 0.0746,
+      "step": 6577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6055440075042793e-06,
+      "loss": 0.011,
+      "step": 6578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6029855926540705e-06,
+      "loss": 0.0013,
+      "step": 6579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.600428246539989e-06,
+      "loss": 0.099,
+      "step": 6580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597871969531518e-06,
+      "loss": 0.0306,
+      "step": 6581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.595316761997998e-06,
+      "loss": 0.0022,
+      "step": 6582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.59276262430861e-06,
+      "loss": 0.0035,
+      "step": 6583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5902095568323815e-06,
+      "loss": 0.0013,
+      "step": 6584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.587657559938186e-06,
+      "loss": 0.0768,
+      "step": 6585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5851066339947397e-06,
+      "loss": 0.0564,
+      "step": 6586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5825567793706088e-06,
+      "loss": 0.0094,
+      "step": 6587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.580007996434197e-06,
+      "loss": 0.0198,
+      "step": 6588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.57746028555376e-06,
+      "loss": 0.0039,
+      "step": 6589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.574913647097396e-06,
+      "loss": 0.0092,
+      "step": 6590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.572368081433049e-06,
+      "loss": 0.054,
+      "step": 6591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.569823588928508e-06,
+      "loss": 0.0044,
+      "step": 6592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5672801699514094e-06,
+      "loss": 0.0008,
+      "step": 6593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.564737824869226e-06,
+      "loss": 0.002,
+      "step": 6594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.562196554049283e-06,
+      "loss": 0.0493,
+      "step": 6595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.559656357858753e-06,
+      "loss": 0.0184,
+      "step": 6596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5571172366646412e-06,
+      "loss": 0.0021,
+      "step": 6597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55457919083381e-06,
+      "loss": 0.0018,
+      "step": 6598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.552042220732961e-06,
+      "loss": 0.0131,
+      "step": 6599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5495063267286404e-06,
+      "loss": 0.0018,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.546971509187238e-06,
+      "loss": 0.0158,
+      "step": 6601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.544437768474993e-06,
+      "loss": 0.0052,
+      "step": 6602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5419051049579833e-06,
+      "loss": 0.0384,
+      "step": 6603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5393735190021374e-06,
+      "loss": 0.0691,
+      "step": 6604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.536843010973218e-06,
+      "loss": 0.0003,
+      "step": 6605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.534313581236839e-06,
+      "loss": 0.032,
+      "step": 6606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5317852301584642e-06,
+      "loss": 0.29,
+      "step": 6607
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.529257958103387e-06,
+      "loss": 0.0196,
+      "step": 6608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5267317654367563e-06,
+      "loss": 0.0373,
+      "step": 6609
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.524206652523562e-06,
+      "loss": 0.0068,
+      "step": 6610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.521682619728637e-06,
+      "loss": 0.043,
+      "step": 6611
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5191596674166596e-06,
+      "loss": 0.0096,
+      "step": 6612
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5166377959521537e-06,
+      "loss": 0.0304,
+      "step": 6613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.51411700569948e-06,
+      "loss": 0.0006,
+      "step": 6614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5115972970228488e-06,
+      "loss": 0.0011,
+      "step": 6615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5090786702863126e-06,
+      "loss": 0.1025,
+      "step": 6616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.506561125853769e-06,
+      "loss": 0.0014,
+      "step": 6617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5040446640889583e-06,
+      "loss": 0.0242,
+      "step": 6618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.501529285355463e-06,
+      "loss": 0.0944,
+      "step": 6619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4990149900167116e-06,
+      "loss": 0.0046,
+      "step": 6620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496501778435977e-06,
+      "loss": 0.0769,
+      "step": 6621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4939896509763674e-06,
+      "loss": 0.0005,
+      "step": 6622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.491478608000846e-06,
+      "loss": 0.0067,
+      "step": 6623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4889686498722075e-06,
+      "loss": 0.008,
+      "step": 6624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4864597769530997e-06,
+      "loss": 0.0333,
+      "step": 6625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4839519896060094e-06,
+      "loss": 0.0017,
+      "step": 6626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481445288193266e-06,
+      "loss": 0.0306,
+      "step": 6627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4789396730770443e-06,
+      "loss": 0.0028,
+      "step": 6628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.476435144619361e-06,
+      "loss": 0.0041,
+      "step": 6629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4739317031820774e-06,
+      "loss": 0.0008,
+      "step": 6630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4714293491268904e-06,
+      "loss": 0.0524,
+      "step": 6631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.468928082815348e-06,
+      "loss": 0.0103,
+      "step": 6632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.466427904608839e-06,
+      "loss": 0.0056,
+      "step": 6633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4639288148685945e-06,
+      "loss": 0.1388,
+      "step": 6634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.461430813955691e-06,
+      "loss": 0.1849,
+      "step": 6635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4589339022310386e-06,
+      "loss": 0.0713,
+      "step": 6636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4564380800554e-06,
+      "loss": 0.0054,
+      "step": 6637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.453943347789376e-06,
+      "loss": 0.0093,
+      "step": 6638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4514497057934117e-06,
+      "loss": 0.0909,
+      "step": 6639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4489571544277944e-06,
+      "loss": 0.1448,
+      "step": 6640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4464656940526508e-06,
+      "loss": 0.0132,
+      "step": 6641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.443975325027953e-06,
+      "loss": 0.0007,
+      "step": 6642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.441486047713515e-06,
+      "loss": 0.0011,
+      "step": 6643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4389978624689923e-06,
+      "loss": 0.0164,
+      "step": 6644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4365107696538835e-06,
+      "loss": 0.0074,
+      "step": 6645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4340247696275297e-06,
+      "loss": 0.0375,
+      "step": 6646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4315398627491137e-06,
+      "loss": 0.0117,
+      "step": 6647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.429056049377662e-06,
+      "loss": 0.0478,
+      "step": 6648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4265733298720373e-06,
+      "loss": 0.0099,
+      "step": 6649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4240917045909463e-06,
+      "loss": 0.1366,
+      "step": 6650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4216111738929426e-06,
+      "loss": 0.0656,
+      "step": 6651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.419131738136419e-06,
+      "loss": 0.0286,
+      "step": 6652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.416653397679607e-06,
+      "loss": 0.0184,
+      "step": 6653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.414176152880585e-06,
+      "loss": 0.0045,
+      "step": 6654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4117000040972704e-06,
+      "loss": 0.0038,
+      "step": 6655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409224951687421e-06,
+      "loss": 0.0012,
+      "step": 6656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.406750996008641e-06,
+      "loss": 0.002,
+      "step": 6657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4042781374183676e-06,
+      "loss": 0.0695,
+      "step": 6658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4018063762738863e-06,
+      "loss": 0.0009,
+      "step": 6659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3993357129323224e-06,
+      "loss": 0.0359,
+      "step": 6660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.396866147750644e-06,
+      "loss": 0.0046,
+      "step": 6661
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3943976810856573e-06,
+      "loss": 0.0045,
+      "step": 6662
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3919303132940162e-06,
+      "loss": 0.0167,
+      "step": 6663
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.389464044732204e-06,
+      "loss": 0.1276,
+      "step": 6664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.386998875756554e-06,
+      "loss": 0.0017,
+      "step": 6665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3845348067232455e-06,
+      "loss": 0.0276,
+      "step": 6666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3820718379882836e-06,
+      "loss": 0.0731,
+      "step": 6667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.379609969907527e-06,
+      "loss": 0.0061,
+      "step": 6668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3771492028366717e-06,
+      "loss": 0.0651,
+      "step": 6669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.374689537131254e-06,
+      "loss": 0.0884,
+      "step": 6670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.372230973146652e-06,
+      "loss": 0.0636,
+      "step": 6671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.369773511238084e-06,
+      "loss": 0.0706,
+      "step": 6672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3673171517606096e-06,
+      "loss": 0.0152,
+      "step": 6673
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3648618950691305e-06,
+      "loss": 0.0264,
+      "step": 6674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3624077415183834e-06,
+      "loss": 0.0128,
+      "step": 6675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3599546914629534e-06,
+      "loss": 0.0023,
+      "step": 6676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3575027452572574e-06,
+      "loss": 0.0137,
+      "step": 6677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.355051903255562e-06,
+      "loss": 0.0863,
+      "step": 6678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352602165811967e-06,
+      "loss": 0.0034,
+      "step": 6679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.350153533280418e-06,
+      "loss": 0.012,
+      "step": 6680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3477060060146973e-06,
+      "loss": 0.0032,
+      "step": 6681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.345259584368429e-06,
+      "loss": 0.0029,
+      "step": 6682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3428142686950815e-06,
+      "loss": 0.0077,
+      "step": 6683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340370059347952e-06,
+      "loss": 0.0183,
+      "step": 6684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3379269566801866e-06,
+      "loss": 0.0126,
+      "step": 6685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3354849610447728e-06,
+      "loss": 0.0398,
+      "step": 6686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3330440727945323e-06,
+      "loss": 0.0074,
+      "step": 6687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3306042922821328e-06,
+      "loss": 0.0059,
+      "step": 6688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3281656198600767e-06,
+      "loss": 0.0031,
+      "step": 6689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3257280558807117e-06,
+      "loss": 0.0038,
+      "step": 6690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323291600696217e-06,
+      "loss": 0.0316,
+      "step": 6691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.320856254658619e-06,
+      "loss": 0.0268,
+      "step": 6692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3184220181197846e-06,
+      "loss": 0.0024,
+      "step": 6693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.315988891431412e-06,
+      "loss": 0.033,
+      "step": 6694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3135568749450467e-06,
+      "loss": 0.0494,
+      "step": 6695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3111259690120726e-06,
+      "loss": 0.0108,
+      "step": 6696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.308696173983711e-06,
+      "loss": 0.0231,
+      "step": 6697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.306267490211025e-06,
+      "loss": 0.001,
+      "step": 6698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3038399180449144e-06,
+      "loss": 0.0058,
+      "step": 6699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3014134578361237e-06,
+      "loss": 0.0006,
+      "step": 6700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.298988109935227e-06,
+      "loss": 0.0582,
+      "step": 6701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2965638746926466e-06,
+      "loss": 0.1375,
+      "step": 6702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.29414075245864e-06,
+      "loss": 0.0692,
+      "step": 6703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.29171874358331e-06,
+      "loss": 0.0022,
+      "step": 6704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.289297848416586e-06,
+      "loss": 0.0447,
+      "step": 6705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.286878067308248e-06,
+      "loss": 0.0043,
+      "step": 6706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.28445940060791e-06,
+      "loss": 0.0471,
+      "step": 6707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2820418486650264e-06,
+      "loss": 0.0181,
+      "step": 6708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.279625411828891e-06,
+      "loss": 0.0707,
+      "step": 6709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2772100904486373e-06,
+      "loss": 0.0062,
+      "step": 6710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.274795884873231e-06,
+      "loss": 0.0403,
+      "step": 6711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2723827954514842e-06,
+      "loss": 0.0024,
+      "step": 6712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.269970822532045e-06,
+      "loss": 0.0685,
+      "step": 6713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2675599664634007e-06,
+      "loss": 0.039,
+      "step": 6714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2651502275938763e-06,
+      "loss": 0.0199,
+      "step": 6715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2627416062716366e-06,
+      "loss": 0.1483,
+      "step": 6716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2603341028446833e-06,
+      "loss": 0.0283,
+      "step": 6717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2579277176608607e-06,
+      "loss": 0.0194,
+      "step": 6718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2555224510678466e-06,
+      "loss": 0.0012,
+      "step": 6719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2531183034131544e-06,
+      "loss": 0.0386,
+      "step": 6720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.250715275044144e-06,
+      "loss": 0.0506,
+      "step": 6721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2483133663080105e-06,
+      "loss": 0.0021,
+      "step": 6722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2459125775517854e-06,
+      "loss": 0.0432,
+      "step": 6723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.243512909122342e-06,
+      "loss": 0.0442,
+      "step": 6724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2411143613663866e-06,
+      "loss": 0.0602,
+      "step": 6725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2387169346304683e-06,
+      "loss": 0.0013,
+      "step": 6726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.236320629260974e-06,
+      "loss": 0.0808,
+      "step": 6727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2339254456041227e-06,
+      "loss": 0.0505,
+      "step": 6728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.231531384005977e-06,
+      "loss": 0.0366,
+      "step": 6729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2291384448124364e-06,
+      "loss": 0.0258,
+      "step": 6730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2267466283692386e-06,
+      "loss": 0.0376,
+      "step": 6731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.22435593502196e-06,
+      "loss": 0.023,
+      "step": 6732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2219663651160074e-06,
+      "loss": 0.0637,
+      "step": 6733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.219577918996634e-06,
+      "loss": 0.007,
+      "step": 6734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.217190597008927e-06,
+      "loss": 0.0126,
+      "step": 6735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.214804399497815e-06,
+      "loss": 0.0059,
+      "step": 6736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2124193268080552e-06,
+      "loss": 0.0074,
+      "step": 6737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.21003537928425e-06,
+      "loss": 0.055,
+      "step": 6738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2076525572708386e-06,
+      "loss": 0.0007,
+      "step": 6739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2052708611120954e-06,
+      "loss": 0.0574,
+      "step": 6740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2028902911521324e-06,
+      "loss": 0.0224,
+      "step": 6741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.200510847734899e-06,
+      "loss": 0.0997,
+      "step": 6742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1981325312041834e-06,
+      "loss": 0.044,
+      "step": 6743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1957553419036117e-06,
+      "loss": 0.0026,
+      "step": 6744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1933792801766397e-06,
+      "loss": 0.0032,
+      "step": 6745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.191004346366571e-06,
+      "loss": 0.0492,
+      "step": 6746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1886305408165363e-06,
+      "loss": 0.0035,
+      "step": 6747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186257863869511e-06,
+      "loss": 0.0351,
+      "step": 6748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183886315868302e-06,
+      "loss": 0.0019,
+      "step": 6749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1815158971555595e-06,
+      "loss": 0.019,
+      "step": 6750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.179146608073763e-06,
+      "loss": 0.0062,
+      "step": 6751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1767784489652345e-06,
+      "loss": 0.0017,
+      "step": 6752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.174411420172132e-06,
+      "loss": 0.0169,
+      "step": 6753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1720455220364443e-06,
+      "loss": 0.0079,
+      "step": 6754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.169680754900003e-06,
+      "loss": 0.0364,
+      "step": 6755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1673171191044764e-06,
+      "loss": 0.0035,
+      "step": 6756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1649546149913647e-06,
+      "loss": 0.0007,
+      "step": 6757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16259324290201e-06,
+      "loss": 0.005,
+      "step": 6758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16023300317759e-06,
+      "loss": 0.0213,
+      "step": 6759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1578738961591116e-06,
+      "loss": 0.0034,
+      "step": 6760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1555159221874267e-06,
+      "loss": 0.0057,
+      "step": 6761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.15315908160322e-06,
+      "loss": 0.0668,
+      "step": 6762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.150803374747015e-06,
+      "loss": 0.0029,
+      "step": 6763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1484488019591644e-06,
+      "loss": 0.0013,
+      "step": 6764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1460953635798642e-06,
+      "loss": 0.0042,
+      "step": 6765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.143743059949144e-06,
+      "loss": 0.04,
+      "step": 6766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.14139189140687e-06,
+      "loss": 0.0652,
+      "step": 6767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.139041858292744e-06,
+      "loss": 0.0086,
+      "step": 6768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1366929609463028e-06,
+      "loss": 0.013,
+      "step": 6769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1343451997069208e-06,
+      "loss": 0.0093,
+      "step": 6770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1319985749138107e-06,
+      "loss": 0.0069,
+      "step": 6771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1296530869060106e-06,
+      "loss": 0.003,
+      "step": 6772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1273087360224086e-06,
+      "loss": 0.0204,
+      "step": 6773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1249655226017153e-06,
+      "loss": 0.0498,
+      "step": 6774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1226234469824857e-06,
+      "loss": 0.0384,
+      "step": 6775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1202825095031077e-06,
+      "loss": 0.0036,
+      "step": 6776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.117942710501806e-06,
+      "loss": 0.009,
+      "step": 6777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1156040503166376e-06,
+      "loss": 0.0017,
+      "step": 6778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.113266529285498e-06,
+      "loss": 0.0371,
+      "step": 6779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.110930147746121e-06,
+      "loss": 0.0178,
+      "step": 6780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1085949060360654e-06,
+      "loss": 0.0016,
+      "step": 6781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1062608044927345e-06,
+      "loss": 0.0207,
+      "step": 6782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.103927843453364e-06,
+      "loss": 0.1061,
+      "step": 6783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1015960232550247e-06,
+      "loss": 0.0523,
+      "step": 6784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0992653442346243e-06,
+      "loss": 0.0017,
+      "step": 6785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0969358067289035e-06,
+      "loss": 0.1263,
+      "step": 6786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0946074110744408e-06,
+      "loss": 0.0065,
+      "step": 6787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0922801576076434e-06,
+      "loss": 0.0006,
+      "step": 6788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0899540466647615e-06,
+      "loss": 0.0393,
+      "step": 6789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.087629078581872e-06,
+      "loss": 0.0016,
+      "step": 6790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.085305253694895e-06,
+      "loss": 0.0547,
+      "step": 6791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0829825723395803e-06,
+      "loss": 0.0328,
+      "step": 6792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0806610348515135e-06,
+      "loss": 0.009,
+      "step": 6793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0783406415661158e-06,
+      "loss": 0.0225,
+      "step": 6794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0760213928186423e-06,
+      "loss": 0.0041,
+      "step": 6795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.073703288944183e-06,
+      "loss": 0.0023,
+      "step": 6796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.071386330277665e-06,
+      "loss": 0.0127,
+      "step": 6797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.069070517153842e-06,
+      "loss": 0.002,
+      "step": 6798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0667558499073104e-06,
+      "loss": 0.01,
+      "step": 6799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064442328872498e-06,
+      "loss": 0.0189,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0621299543836692e-06,
+      "loss": 0.0299,
+      "step": 6801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.059818726774917e-06,
+      "loss": 0.0766,
+      "step": 6802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.057508646380173e-06,
+      "loss": 0.0017,
+      "step": 6803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0551997135332046e-06,
+      "loss": 0.0423,
+      "step": 6804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0528919285676097e-06,
+      "loss": 0.0455,
+      "step": 6805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0505852918168244e-06,
+      "loss": 0.0968,
+      "step": 6806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.048279803614113e-06,
+      "loss": 0.027,
+      "step": 6807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.045975464292578e-06,
+      "loss": 0.0821,
+      "step": 6808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0436722741851555e-06,
+      "loss": 0.0451,
+      "step": 6809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0413702336246156e-06,
+      "loss": 0.0016,
+      "step": 6810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0390693429435626e-06,
+      "loss": 0.0693,
+      "step": 6811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0367696024744322e-06,
+      "loss": 0.0069,
+      "step": 6812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.034471012549497e-06,
+      "loss": 0.0041,
+      "step": 6813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0321735735008654e-06,
+      "loss": 0.0223,
+      "step": 6814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0298772856604696e-06,
+      "loss": 0.0594,
+      "step": 6815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027582149360088e-06,
+      "loss": 0.0122,
+      "step": 6816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0252881649313205e-06,
+      "loss": 0.0882,
+      "step": 6817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0229953327056106e-06,
+      "loss": 0.0052,
+      "step": 6818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.020703653014231e-06,
+      "loss": 0.0561,
+      "step": 6819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.018413126188288e-06,
+      "loss": 0.0131,
+      "step": 6820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.016123752558722e-06,
+      "loss": 0.0025,
+      "step": 6821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0138355324563075e-06,
+      "loss": 0.1496,
+      "step": 6822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0115484662116523e-06,
+      "loss": 0.0083,
+      "step": 6823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0092625541551924e-06,
+      "loss": 0.0226,
+      "step": 6824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.006977796617203e-06,
+      "loss": 0.0045,
+      "step": 6825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.004694193927791e-06,
+      "loss": 0.0518,
+      "step": 6826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.002411746416897e-06,
+      "loss": 0.0027,
+      "step": 6827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.000130454414295e-06,
+      "loss": 0.0666,
+      "step": 6828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997850318249586e-06,
+      "loss": 0.0069,
+      "step": 6829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9955713382522113e-06,
+      "loss": 0.0109,
+      "step": 6830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9932935147514445e-06,
+      "loss": 0.0083,
+      "step": 6831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.991016848076387e-06,
+      "loss": 0.0163,
+      "step": 6832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9887413385559818e-06,
+      "loss": 0.0403,
+      "step": 6833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9864669865189934e-06,
+      "loss": 0.004,
+      "step": 6834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9841937922940267e-06,
+      "loss": 0.0029,
+      "step": 6835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9819217562095184e-06,
+      "loss": 0.0255,
+      "step": 6836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9796508785937364e-06,
+      "loss": 0.012,
+      "step": 6837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9773811597747817e-06,
+      "loss": 0.0077,
+      "step": 6838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97511260008059e-06,
+      "loss": 0.0042,
+      "step": 6839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.972845199838924e-06,
+      "loss": 0.0052,
+      "step": 6840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970578959377388e-06,
+      "loss": 0.0502,
+      "step": 6841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.96831387902341e-06,
+      "loss": 0.0122,
+      "step": 6842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9660499591042502e-06,
+      "loss": 0.0054,
+      "step": 6843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9637871999470082e-06,
+      "loss": 0.0273,
+      "step": 6844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961525601878611e-06,
+      "loss": 0.1031,
+      "step": 6845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95926516522582e-06,
+      "loss": 0.1341,
+      "step": 6846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9570058903152255e-06,
+      "loss": 0.0017,
+      "step": 6847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.954747777473256e-06,
+      "loss": 0.0551,
+      "step": 6848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9524908270261655e-06,
+      "loss": 0.0091,
+      "step": 6849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9502350393000467e-06,
+      "loss": 0.0132,
+      "step": 6850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9479804146208147e-06,
+      "loss": 0.0189,
+      "step": 6851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9457269533142265e-06,
+      "loss": 0.0355,
+      "step": 6852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943474655705865e-06,
+      "loss": 0.0009,
+      "step": 6853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9412235221211474e-06,
+      "loss": 0.0052,
+      "step": 6854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938973552885323e-06,
+      "loss": 0.0329,
+      "step": 6855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9367247483234743e-06,
+      "loss": 0.0611,
+      "step": 6856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9344771087605062e-06,
+      "loss": 0.0477,
+      "step": 6857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.932230634521168e-06,
+      "loss": 0.011,
+      "step": 6858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9299853259300373e-06,
+      "loss": 0.0023,
+      "step": 6859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9277411833115135e-06,
+      "loss": 0.0697,
+      "step": 6860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9254982069898386e-06,
+      "loss": 0.0008,
+      "step": 6861
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.923256397289084e-06,
+      "loss": 0.0457,
+      "step": 6862
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.921015754533149e-06,
+      "loss": 0.026,
+      "step": 6863
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9187762790457667e-06,
+      "loss": 0.0075,
+      "step": 6864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916537971150503e-06,
+      "loss": 0.0758,
+      "step": 6865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914300831170751e-06,
+      "loss": 0.0244,
+      "step": 6866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912064859429742e-06,
+      "loss": 0.0689,
+      "step": 6867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.0025,
+      "step": 6868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9075964219559984e-06,
+      "loss": 0.0102,
+      "step": 6869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9053639568688787e-06,
+      "loss": 0.0014,
+      "step": 6870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.903132661311714e-06,
+      "loss": 0.0664,
+      "step": 6871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9009025356068888e-06,
+      "loss": 0.0141,
+      "step": 6872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8986735800766175e-06,
+      "loss": 0.0028,
+      "step": 6873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8964457950429439e-06,
+      "loss": 0.003,
+      "step": 6874
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8942191808277421e-06,
+      "loss": 0.0083,
+      "step": 6875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.891993737752722e-06,
+      "loss": 0.0083,
+      "step": 6876
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8897694661394128e-06,
+      "loss": 0.0498,
+      "step": 6877
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8875463663091876e-06,
+      "loss": 0.0451,
+      "step": 6878
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8853244385832414e-06,
+      "loss": 0.0948,
+      "step": 6879
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8831036832826055e-06,
+      "loss": 0.0056,
+      "step": 6880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8808841007281375e-06,
+      "loss": 0.0135,
+      "step": 6881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8786656912405288e-06,
+      "loss": 0.0121,
+      "step": 6882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8764484551402984e-06,
+      "loss": 0.0369,
+      "step": 6883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8742323927478013e-06,
+      "loss": 0.0173,
+      "step": 6884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8720175043832123e-06,
+      "loss": 0.0074,
+      "step": 6885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8698037903665479e-06,
+      "loss": 0.025,
+      "step": 6886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8675912510176476e-06,
+      "loss": 0.0033,
+      "step": 6887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8653798866561833e-06,
+      "loss": 0.14,
+      "step": 6888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.86316969760166e-06,
+      "loss": 0.1397,
+      "step": 6889
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8609606841734086e-06,
+      "loss": 0.0276,
+      "step": 6890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8587528466905925e-06,
+      "loss": 0.0038,
+      "step": 6891
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8565461854722055e-06,
+      "loss": 0.0876,
+      "step": 6892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8543407008370695e-06,
+      "loss": 0.009,
+      "step": 6893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8521363931038405e-06,
+      "loss": 0.0636,
+      "step": 6894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.849933262590996e-06,
+      "loss": 0.0006,
+      "step": 6895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8477313096168525e-06,
+      "loss": 0.0406,
+      "step": 6896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8455305344995523e-06,
+      "loss": 0.0034,
+      "step": 6897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8433309375570708e-06,
+      "loss": 0.083,
+      "step": 6898
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8411325191072049e-06,
+      "loss": 0.0171,
+      "step": 6899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8389352794675885e-06,
+      "loss": 0.0224,
+      "step": 6900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8367392189556843e-06,
+      "loss": 0.0081,
+      "step": 6901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.834544337888784e-06,
+      "loss": 0.0398,
+      "step": 6902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8323506365840105e-06,
+      "loss": 0.027,
+      "step": 6903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8301581153583092e-06,
+      "loss": 0.0132,
+      "step": 6904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8279667745284635e-06,
+      "loss": 0.031,
+      "step": 6905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8257766144110823e-06,
+      "loss": 0.0026,
+      "step": 6906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8235876353226045e-06,
+      "loss": 0.0335,
+      "step": 6907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8213998375792985e-06,
+      "loss": 0.0555,
+      "step": 6908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8192132214972614e-06,
+      "loss": 0.0428,
+      "step": 6909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8170277873924213e-06,
+      "loss": 0.0009,
+      "step": 6910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.814843535580536e-06,
+      "loss": 0.065,
+      "step": 6911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8126604663771884e-06,
+      "loss": 0.0096,
+      "step": 6912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8104785800977898e-06,
+      "loss": 0.0038,
+      "step": 6913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8082978770575865e-06,
+      "loss": 0.0438,
+      "step": 6914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8061183575716523e-06,
+      "loss": 0.0027,
+      "step": 6915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8039400219548876e-06,
+      "loss": 0.0534,
+      "step": 6916
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8017628705220224e-06,
+      "loss": 0.0851,
+      "step": 6917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7995869035876167e-06,
+      "loss": 0.0482,
+      "step": 6918
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7974121214660589e-06,
+      "loss": 0.0476,
+      "step": 6919
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7952385244715686e-06,
+      "loss": 0.0043,
+      "step": 6920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.793066112918186e-06,
+      "loss": 0.0761,
+      "step": 6921
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7908948871197885e-06,
+      "loss": 0.0226,
+      "step": 6922
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7887248473900798e-06,
+      "loss": 0.003,
+      "step": 6923
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7865559940425924e-06,
+      "loss": 0.0475,
+      "step": 6924
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.784388327390687e-06,
+      "loss": 0.0046,
+      "step": 6925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7822218477475496e-06,
+      "loss": 0.0085,
+      "step": 6926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7800565554261994e-06,
+      "loss": 0.0354,
+      "step": 6927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7778924507394823e-06,
+      "loss": 0.0251,
+      "step": 6928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7757295340000758e-06,
+      "loss": 0.0119,
+      "step": 6929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7735678055204774e-06,
+      "loss": 0.1532,
+      "step": 6930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7714072656130199e-06,
+      "loss": 0.0116,
+      "step": 6931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7692479145898633e-06,
+      "loss": 0.0355,
+      "step": 6932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.767089752762996e-06,
+      "loss": 0.0013,
+      "step": 6933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7649327804442319e-06,
+      "loss": 0.0026,
+      "step": 6934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.762776997945216e-06,
+      "loss": 0.0394,
+      "step": 6935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7606224055774202e-06,
+      "loss": 0.0134,
+      "step": 6936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7584690036521456e-06,
+      "loss": 0.032,
+      "step": 6937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.756316792480518e-06,
+      "loss": 0.0417,
+      "step": 6938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7541657723734973e-06,
+      "loss": 0.0536,
+      "step": 6939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7520159436418604e-06,
+      "loss": 0.0013,
+      "step": 6940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7498673065962235e-06,
+      "loss": 0.0818,
+      "step": 6941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.747719861547027e-06,
+      "loss": 0.0064,
+      "step": 6942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.745573608804536e-06,
+      "loss": 0.1168,
+      "step": 6943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.743428548678847e-06,
+      "loss": 0.0014,
+      "step": 6944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7412846814798823e-06,
+      "loss": 0.0012,
+      "step": 6945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7391420075173915e-06,
+      "loss": 0.007,
+      "step": 6946
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7370005271009572e-06,
+      "loss": 0.0055,
+      "step": 6947
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7348602405399783e-06,
+      "loss": 0.0082,
+      "step": 6948
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.732721148143691e-06,
+      "loss": 0.0856,
+      "step": 6949
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.730583250221156e-06,
+      "loss": 0.0333,
+      "step": 6950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7284465470812605e-06,
+      "loss": 0.0909,
+      "step": 6951
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.726311039032721e-06,
+      "loss": 0.0393,
+      "step": 6952
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7241767263840824e-06,
+      "loss": 0.1125,
+      "step": 6953
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7220436094437088e-06,
+      "loss": 0.0556,
+      "step": 6954
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7199116885197996e-06,
+      "loss": 0.0008,
+      "step": 6955
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7177809639203834e-06,
+      "loss": 0.0041,
+      "step": 6956
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.715651435953305e-06,
+      "loss": 0.0025,
+      "step": 6957
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7135231049262468e-06,
+      "loss": 0.0008,
+      "step": 6958
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7113959711467143e-06,
+      "loss": 0.0131,
+      "step": 6959
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7092700349220403e-06,
+      "loss": 0.0016,
+      "step": 6960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071452965593839e-06,
+      "loss": 0.0029,
+      "step": 6961
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7050217563657322e-06,
+      "loss": 0.0042,
+      "step": 6962
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7028994146478983e-06,
+      "loss": 0.0118,
+      "step": 6963
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7007782717125266e-06,
+      "loss": 0.0253,
+      "step": 6964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6986583278660773e-06,
+      "loss": 0.0979,
+      "step": 6965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6965395834148468e-06,
+      "loss": 0.0432,
+      "step": 6966
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6944220386649602e-06,
+      "loss": 0.0089,
+      "step": 6967
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6923056939223582e-06,
+      "loss": 0.004,
+      "step": 6968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6901905494928172e-06,
+      "loss": 0.0366,
+      "step": 6969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6880766056819387e-06,
+      "loss": 0.0312,
+      "step": 6970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.685963862795149e-06,
+      "loss": 0.0031,
+      "step": 6971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6838523211377022e-06,
+      "loss": 0.0036,
+      "step": 6972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6817419810146806e-06,
+      "loss": 0.0207,
+      "step": 6973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6796328427309849e-06,
+      "loss": 0.0038,
+      "step": 6974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.67752490659135e-06,
+      "loss": 0.0013,
+      "step": 6975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.675418172900336e-06,
+      "loss": 0.0308,
+      "step": 6976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.673312641962329e-06,
+      "loss": 0.0396,
+      "step": 6977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6712083140815382e-06,
+      "loss": 0.1467,
+      "step": 6978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6691051895620025e-06,
+      "loss": 0.0598,
+      "step": 6979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6670032687075888e-06,
+      "loss": 0.0258,
+      "step": 6980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6649025518219809e-06,
+      "loss": 0.002,
+      "step": 6981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6628030392087001e-06,
+      "loss": 0.1105,
+      "step": 6982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6607047311710845e-06,
+      "loss": 0.0182,
+      "step": 6983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6586076280123032e-06,
+      "loss": 0.1182,
+      "step": 6984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6565117300353496e-06,
+      "loss": 0.0257,
+      "step": 6985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6544170375430458e-06,
+      "loss": 0.064,
+      "step": 6986
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6523235508380342e-06,
+      "loss": 0.002,
+      "step": 6987
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6502312702227885e-06,
+      "loss": 0.0066,
+      "step": 6988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6481401959996058e-06,
+      "loss": 0.0432,
+      "step": 6989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6460503284706098e-06,
+      "loss": 0.001,
+      "step": 6990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6439616679377445e-06,
+      "loss": 0.0422,
+      "step": 6991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6418742147027876e-06,
+      "loss": 0.011,
+      "step": 6992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6397879690673368e-06,
+      "loss": 0.0224,
+      "step": 6993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6377029313328218e-06,
+      "loss": 0.0044,
+      "step": 6994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6356191018004864e-06,
+      "loss": 0.0018,
+      "step": 6995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6335364807714104e-06,
+      "loss": 0.0298,
+      "step": 6996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6314550685464947e-06,
+      "loss": 0.0504,
+      "step": 6997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6293748654264663e-06,
+      "loss": 0.0234,
+      "step": 6998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6272958717118791e-06,
+      "loss": 0.004,
+      "step": 6999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6252180877031066e-06,
+      "loss": 0.0675,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6231415137003536e-06,
+      "loss": 0.0005,
+      "step": 7001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6210661500036474e-06,
+      "loss": 0.0117,
+      "step": 7002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.618991996912841e-06,
+      "loss": 0.0011,
+      "step": 7003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6169190547276115e-06,
+      "loss": 0.0534,
+      "step": 7004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6148473237474638e-06,
+      "loss": 0.0041,
+      "step": 7005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6127768042717252e-06,
+      "loss": 0.0659,
+      "step": 7006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6107074965995506e-06,
+      "loss": 0.0242,
+      "step": 7007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6086394010299134e-06,
+      "loss": 0.0114,
+      "step": 7008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.606572517861621e-06,
+      "loss": 0.0077,
+      "step": 7009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6045068473932967e-06,
+      "loss": 0.001,
+      "step": 7010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6024423899233953e-06,
+      "loss": 0.0453,
+      "step": 7011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6003791457501927e-06,
+      "loss": 0.0048,
+      "step": 7012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5983171151717924e-06,
+      "loss": 0.0798,
+      "step": 7013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5962562984861196e-06,
+      "loss": 0.0104,
+      "step": 7014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5941966959909262e-06,
+      "loss": 0.0412,
+      "step": 7015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5921383079837882e-06,
+      "loss": 0.0333,
+      "step": 7016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.590081134762107e-06,
+      "loss": 0.065,
+      "step": 7017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5880251766231026e-06,
+      "loss": 0.0056,
+      "step": 7018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5859704338638283e-06,
+      "loss": 0.0578,
+      "step": 7019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5839169067811555e-06,
+      "loss": 0.0483,
+      "step": 7020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5818645956717827e-06,
+      "loss": 0.0641,
+      "step": 7021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5798135008322357e-06,
+      "loss": 0.0498,
+      "step": 7022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5777636225588545e-06,
+      "loss": 0.0035,
+      "step": 7023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.575714961147814e-06,
+      "loss": 0.0698,
+      "step": 7024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5736675168951065e-06,
+      "loss": 0.0551,
+      "step": 7025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5716212900965565e-06,
+      "loss": 0.0072,
+      "step": 7026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5695762810478e-06,
+      "loss": 0.0113,
+      "step": 7027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5675324900443078e-06,
+      "loss": 0.0997,
+      "step": 7028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.565489917381371e-06,
+      "loss": 0.0013,
+      "step": 7029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.563448563354103e-06,
+      "loss": 0.0751,
+      "step": 7030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.561408428257446e-06,
+      "loss": 0.0202,
+      "step": 7031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5593695123861608e-06,
+      "loss": 0.0012,
+      "step": 7032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.557331816034835e-06,
+      "loss": 0.0061,
+      "step": 7033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.555295339497881e-06,
+      "loss": 0.0021,
+      "step": 7034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5532600830695289e-06,
+      "loss": 0.0356,
+      "step": 7035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5512260470438422e-06,
+      "loss": 0.0632,
+      "step": 7036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5491932317146964e-06,
+      "loss": 0.0628,
+      "step": 7037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5471616373758002e-06,
+      "loss": 0.0006,
+      "step": 7038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5451312643206828e-06,
+      "loss": 0.0337,
+      "step": 7039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5431021128426971e-06,
+      "loss": 0.0013,
+      "step": 7040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5410741832350173e-06,
+      "loss": 0.0064,
+      "step": 7041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5390474757906449e-06,
+      "loss": 0.0026,
+      "step": 7042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5370219908024043e-06,
+      "loss": 0.0623,
+      "step": 7043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.534997728562938e-06,
+      "loss": 0.0139,
+      "step": 7044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5329746893647156e-06,
+      "loss": 0.0217,
+      "step": 7045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5309528735000335e-06,
+      "loss": 0.0024,
+      "step": 7046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5289322812610051e-06,
+      "loss": 0.0011,
+      "step": 7047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526912912939571e-06,
+      "loss": 0.0849,
+      "step": 7048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5248947688274928e-06,
+      "loss": 0.0013,
+      "step": 7049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5228778492163598e-06,
+      "loss": 0.0117,
+      "step": 7050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5208621543975766e-06,
+      "loss": 0.0074,
+      "step": 7051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5188476846623767e-06,
+      "loss": 0.0013,
+      "step": 7052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5168344403018131e-06,
+      "loss": 0.0026,
+      "step": 7053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5148224216067653e-06,
+      "loss": 0.0845,
+      "step": 7054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5128116288679328e-06,
+      "loss": 0.0083,
+      "step": 7055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5108020623758413e-06,
+      "loss": 0.0054,
+      "step": 7056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.508793722420835e-06,
+      "loss": 0.0936,
+      "step": 7057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.506786609293084e-06,
+      "loss": 0.0383,
+      "step": 7058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5047807232825795e-06,
+      "loss": 0.04,
+      "step": 7059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.502776064679139e-06,
+      "loss": 0.0008,
+      "step": 7060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.500772633772396e-06,
+      "loss": 0.0096,
+      "step": 7061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4987704308518113e-06,
+      "loss": 0.0628,
+      "step": 7062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4967694562066703e-06,
+      "loss": 0.1453,
+      "step": 7063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.494769710126074e-06,
+      "loss": 0.208,
+      "step": 7064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4927711928989508e-06,
+      "loss": 0.008,
+      "step": 7065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4907739048140524e-06,
+      "loss": 0.0783,
+      "step": 7066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.488777846159949e-06,
+      "loss": 0.0024,
+      "step": 7067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4867830172250363e-06,
+      "loss": 0.0045,
+      "step": 7068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4847894182975331e-06,
+      "loss": 0.0107,
+      "step": 7069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4827970496654786e-06,
+      "loss": 0.0693,
+      "step": 7070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4808059116167306e-06,
+      "loss": 0.0383,
+      "step": 7071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4788160044389754e-06,
+      "loss": 0.0046,
+      "step": 7072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4768273284197188e-06,
+      "loss": 0.0125,
+      "step": 7073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4748398838462886e-06,
+      "loss": 0.0026,
+      "step": 7074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.472853671005835e-06,
+      "loss": 0.0548,
+      "step": 7075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.470868690185332e-06,
+      "loss": 0.0386,
+      "step": 7076
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4688849416715734e-06,
+      "loss": 0.1009,
+      "step": 7077
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4669024257511722e-06,
+      "loss": 0.0076,
+      "step": 7078
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4649211427105702e-06,
+      "loss": 0.0012,
+      "step": 7079
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4629410928360232e-06,
+      "loss": 0.0043,
+      "step": 7080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4609622764136155e-06,
+      "loss": 0.0979,
+      "step": 7081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4589846937292507e-06,
+      "loss": 0.0061,
+      "step": 7082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4570083450686544e-06,
+      "loss": 0.0034,
+      "step": 7083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.455033230717372e-06,
+      "loss": 0.1012,
+      "step": 7084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4530593509607738e-06,
+      "loss": 0.037,
+      "step": 7085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4510867060840483e-06,
+      "loss": 0.0033,
+      "step": 7086
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4491152963722121e-06,
+      "loss": 0.0694,
+      "step": 7087
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.447145122110092e-06,
+      "loss": 0.0008,
+      "step": 7088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4451761835823464e-06,
+      "loss": 0.0889,
+      "step": 7089
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4432084810734504e-06,
+      "loss": 0.0789,
+      "step": 7090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4412420148677052e-06,
+      "loss": 0.0276,
+      "step": 7091
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4392767852492251e-06,
+      "loss": 0.0274,
+      "step": 7092
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4373127925019536e-06,
+      "loss": 0.0145,
+      "step": 7093
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4353500369096508e-06,
+      "loss": 0.1059,
+      "step": 7094
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4333885187559004e-06,
+      "loss": 0.0044,
+      "step": 7095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4314282383241097e-06,
+      "loss": 0.0049,
+      "step": 7096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4294691958974983e-06,
+      "loss": 0.046,
+      "step": 7097
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4275113917591167e-06,
+      "loss": 0.0254,
+      "step": 7098
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4255548261918307e-06,
+      "loss": 0.0398,
+      "step": 7099
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4235994994783297e-06,
+      "loss": 0.0027,
+      "step": 7100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4216454119011236e-06,
+      "loss": 0.0159,
+      "step": 7101
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.419692563742544e-06,
+      "loss": 0.0051,
+      "step": 7102
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.417740955284741e-06,
+      "loss": 0.0064,
+      "step": 7103
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4157905868096888e-06,
+      "loss": 0.0425,
+      "step": 7104
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4138414585991811e-06,
+      "loss": 0.0076,
+      "step": 7105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.411893570934827e-06,
+      "loss": 0.0066,
+      "step": 7106
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4099469240980657e-06,
+      "loss": 0.0053,
+      "step": 7107
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4080015183701512e-06,
+      "loss": 0.0032,
+      "step": 7108
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4060573540321609e-06,
+      "loss": 0.0467,
+      "step": 7109
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4041144313649913e-06,
+      "loss": 0.0313,
+      "step": 7110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.40217275064936e-06,
+      "loss": 0.0712,
+      "step": 7111
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4002323121658046e-06,
+      "loss": 0.1242,
+      "step": 7112
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3982931161946867e-06,
+      "loss": 0.0845,
+      "step": 7113
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3963551630161798e-06,
+      "loss": 0.047,
+      "step": 7114
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3944184529102878e-06,
+      "loss": 0.0646,
+      "step": 7115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.392482986156829e-06,
+      "loss": 0.0565,
+      "step": 7116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3905487630354441e-06,
+      "loss": 0.1027,
+      "step": 7117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3886157838255932e-06,
+      "loss": 0.0019,
+      "step": 7118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3866840488065602e-06,
+      "loss": 0.073,
+      "step": 7119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3847535582574412e-06,
+      "loss": 0.0451,
+      "step": 7120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3828243124571594e-06,
+      "loss": 0.0591,
+      "step": 7121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3808963116844598e-06,
+      "loss": 0.041,
+      "step": 7122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3789695562178983e-06,
+      "loss": 0.0012,
+      "step": 7123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3770440463358592e-06,
+      "loss": 0.0672,
+      "step": 7124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3751197823165453e-06,
+      "loss": 0.0143,
+      "step": 7125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3731967644379761e-06,
+      "loss": 0.0471,
+      "step": 7126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3712749929779944e-06,
+      "loss": 0.0357,
+      "step": 7127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3693544682142622e-06,
+      "loss": 0.0029,
+      "step": 7128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.367435190424261e-06,
+      "loss": 0.0015,
+      "step": 7129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3655171598852934e-06,
+      "loss": 0.0036,
+      "step": 7130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3636003768744754e-06,
+      "loss": 0.0547,
+      "step": 7131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3616848416687545e-06,
+      "loss": 0.0025,
+      "step": 7132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3597705545448847e-06,
+      "loss": 0.0309,
+      "step": 7133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3578575157794504e-06,
+      "loss": 0.0311,
+      "step": 7134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3559457256488494e-06,
+      "loss": 0.0566,
+      "step": 7135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.354035184429302e-06,
+      "loss": 0.0408,
+      "step": 7136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3521258923968483e-06,
+      "loss": 0.0389,
+      "step": 7137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3502178498273454e-06,
+      "loss": 0.0018,
+      "step": 7138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.348311056996471e-06,
+      "loss": 0.0349,
+      "step": 7139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.346405514179726e-06,
+      "loss": 0.0056,
+      "step": 7140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3445012216524223e-06,
+      "loss": 0.2187,
+      "step": 7141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.342598179689698e-06,
+      "loss": 0.0715,
+      "step": 7142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.340696388566508e-06,
+      "loss": 0.0514,
+      "step": 7143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.338795848557629e-06,
+      "loss": 0.0052,
+      "step": 7144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3368965599376527e-06,
+      "loss": 0.0106,
+      "step": 7145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.334998522980997e-06,
+      "loss": 0.0022,
+      "step": 7146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.333101737961887e-06,
+      "loss": 0.0873,
+      "step": 7147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3312062051543783e-06,
+      "loss": 0.1454,
+      "step": 7148
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3293119248323439e-06,
+      "loss": 0.0442,
+      "step": 7149
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.327418897269468e-06,
+      "loss": 0.0615,
+      "step": 7150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.325527122739262e-06,
+      "loss": 0.0666,
+      "step": 7151
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3236366015150515e-06,
+      "loss": 0.0009,
+      "step": 7152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.321747333869986e-06,
+      "loss": 0.0339,
+      "step": 7153
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3198593200770293e-06,
+      "loss": 0.0079,
+      "step": 7154
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3179725604089654e-06,
+      "loss": 0.0134,
+      "step": 7155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3160870551383975e-06,
+      "loss": 0.0078,
+      "step": 7156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3142028045377497e-06,
+      "loss": 0.013,
+      "step": 7157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3123198088792577e-06,
+      "loss": 0.0041,
+      "step": 7158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3104380684349838e-06,
+      "loss": 0.0017,
+      "step": 7159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3085575834768072e-06,
+      "loss": 0.1429,
+      "step": 7160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3066783542764194e-06,
+      "loss": 0.0343,
+      "step": 7161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.304800381105339e-06,
+      "loss": 0.0018,
+      "step": 7162
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3029236642348987e-06,
+      "loss": 0.0017,
+      "step": 7163
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3010482039362494e-06,
+      "loss": 0.0096,
+      "step": 7164
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2991740004803644e-06,
+      "loss": 0.0213,
+      "step": 7165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2973010541380315e-06,
+      "loss": 0.0827,
+      "step": 7166
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2954293651798556e-06,
+      "loss": 0.0816,
+      "step": 7167
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2935589338762632e-06,
+      "loss": 0.0021,
+      "step": 7168
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2916897604975e-06,
+      "loss": 0.0022,
+      "step": 7169
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2898218453136257e-06,
+      "loss": 0.002,
+      "step": 7170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.287955188594522e-06,
+      "loss": 0.0098,
+      "step": 7171
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2860897906098869e-06,
+      "loss": 0.009,
+      "step": 7172
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2842256516292374e-06,
+      "loss": 0.0086,
+      "step": 7173
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.28236277192191e-06,
+      "loss": 0.0046,
+      "step": 7174
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2805011517570553e-06,
+      "loss": 0.0004,
+      "step": 7175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2786407914036425e-06,
+      "loss": 0.0336,
+      "step": 7176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.276781691130462e-06,
+      "loss": 0.0716,
+      "step": 7177
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2749238512061213e-06,
+      "loss": 0.0017,
+      "step": 7178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2730672718990434e-06,
+      "loss": 0.0441,
+      "step": 7179
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2712119534774715e-06,
+      "loss": 0.0905,
+      "step": 7180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2693578962094666e-06,
+      "loss": 0.0028,
+      "step": 7181
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2675051003629058e-06,
+      "loss": 0.0036,
+      "step": 7182
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.265653566205487e-06,
+      "loss": 0.0145,
+      "step": 7183
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2638032940047207e-06,
+      "loss": 0.1291,
+      "step": 7184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2619542840279397e-06,
+      "loss": 0.0055,
+      "step": 7185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2601065365422915e-06,
+      "loss": 0.0006,
+      "step": 7186
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2582600518147448e-06,
+      "loss": 0.0277,
+      "step": 7187
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2564148301120838e-06,
+      "loss": 0.0329,
+      "step": 7188
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2545708717009063e-06,
+      "loss": 0.0811,
+      "step": 7189
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.252728176847634e-06,
+      "loss": 0.034,
+      "step": 7190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2508867458185037e-06,
+      "loss": 0.0011,
+      "step": 7191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2490465788795669e-06,
+      "loss": 0.0468,
+      "step": 7192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.247207676296699e-06,
+      "loss": 0.0235,
+      "step": 7193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2453700383355838e-06,
+      "loss": 0.1087,
+      "step": 7194
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2435336652617292e-06,
+      "loss": 0.0734,
+      "step": 7195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2416985573404571e-06,
+      "loss": 0.0066,
+      "step": 7196
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.239864714836909e-06,
+      "loss": 0.0357,
+      "step": 7197
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2380321380160432e-06,
+      "loss": 0.0011,
+      "step": 7198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2362008271426317e-06,
+      "loss": 0.0029,
+      "step": 7199
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.23437078248127e-06,
+      "loss": 0.0075,
+      "step": 7200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.232542004296362e-06,
+      "loss": 0.0228,
+      "step": 7201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2307144928521376e-06,
+      "loss": 0.0145,
+      "step": 7202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2288882484126352e-06,
+      "loss": 0.0161,
+      "step": 7203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2270632712417175e-06,
+      "loss": 0.004,
+      "step": 7204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.22523956160306e-06,
+      "loss": 0.0606,
+      "step": 7205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2234171197601563e-06,
+      "loss": 0.0004,
+      "step": 7206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.221595945976316e-06,
+      "loss": 0.1039,
+      "step": 7207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2197760405146675e-06,
+      "loss": 0.0466,
+      "step": 7208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2179574036381535e-06,
+      "loss": 0.0466,
+      "step": 7209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2161400356095376e-06,
+      "loss": 0.0048,
+      "step": 7210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2143239366913906e-06,
+      "loss": 0.0027,
+      "step": 7211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2125091071461114e-06,
+      "loss": 0.0548,
+      "step": 7212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2106955472359073e-06,
+      "loss": 0.0231,
+      "step": 7213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.208883257222807e-06,
+      "loss": 0.0088,
+      "step": 7214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.207072237368656e-06,
+      "loss": 0.0076,
+      "step": 7215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2052624879351105e-06,
+      "loss": 0.0156,
+      "step": 7216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2034540091836477e-06,
+      "loss": 0.044,
+      "step": 7217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.20164680137556e-06,
+      "loss": 0.0107,
+      "step": 7218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1998408647719606e-06,
+      "loss": 0.0012,
+      "step": 7219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1980361996337697e-06,
+      "loss": 0.0012,
+      "step": 7220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.196232806221731e-06,
+      "loss": 0.0009,
+      "step": 7221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1944306847964027e-06,
+      "loss": 0.0209,
+      "step": 7222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1926298356181597e-06,
+      "loss": 0.0458,
+      "step": 7223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.190830258947192e-06,
+      "loss": 0.0054,
+      "step": 7224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1890319550435059e-06,
+      "loss": 0.0027,
+      "step": 7225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1872349241669245e-06,
+      "loss": 0.0011,
+      "step": 7226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1854391665770892e-06,
+      "loss": 0.0347,
+      "step": 7227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1836446825334491e-06,
+      "loss": 0.12,
+      "step": 7228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1818514722952812e-06,
+      "loss": 0.0471,
+      "step": 7229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.180059536121666e-06,
+      "loss": 0.0338,
+      "step": 7230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1782688742715098e-06,
+      "loss": 0.024,
+      "step": 7231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1764794870035301e-06,
+      "loss": 0.0027,
+      "step": 7232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1746913745762623e-06,
+      "loss": 0.003,
+      "step": 7233
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1729045372480562e-06,
+      "loss": 0.0462,
+      "step": 7234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.171118975277078e-06,
+      "loss": 0.0372,
+      "step": 7235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1693346889213108e-06,
+      "loss": 0.2304,
+      "step": 7236
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1675516784385489e-06,
+      "loss": 0.0021,
+      "step": 7237
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1657699440864068e-06,
+      "loss": 0.002,
+      "step": 7238
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1639894861223122e-06,
+      "loss": 0.0012,
+      "step": 7239
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1622103048035115e-06,
+      "loss": 0.0352,
+      "step": 7240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1604324003870637e-06,
+      "loss": 0.006,
+      "step": 7241
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1586557731298442e-06,
+      "loss": 0.0073,
+      "step": 7242
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1568804232885455e-06,
+      "loss": 0.0716,
+      "step": 7243
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1551063511196703e-06,
+      "loss": 0.0958,
+      "step": 7244
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1533335568795412e-06,
+      "loss": 0.0027,
+      "step": 7245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1515620408242989e-06,
+      "loss": 0.0806,
+      "step": 7246
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.14979180320989e-06,
+      "loss": 0.0022,
+      "step": 7247
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1480228442920848e-06,
+      "loss": 0.0489,
+      "step": 7248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1462551643264674e-06,
+      "loss": 0.0031,
+      "step": 7249
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1444887635684343e-06,
+      "loss": 0.0251,
+      "step": 7250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1427236422731991e-06,
+      "loss": 0.0965,
+      "step": 7251
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1409598006957912e-06,
+      "loss": 0.0418,
+      "step": 7252
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1391972390910544e-06,
+      "loss": 0.0215,
+      "step": 7253
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1374359577136452e-06,
+      "loss": 0.0283,
+      "step": 7254
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1356759568180375e-06,
+      "loss": 0.0156,
+      "step": 7255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1339172366585216e-06,
+      "loss": 0.0277,
+      "step": 7256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1321597974892019e-06,
+      "loss": 0.0031,
+      "step": 7257
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1304036395639928e-06,
+      "loss": 0.0046,
+      "step": 7258
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1286487631366305e-06,
+      "loss": 0.043,
+      "step": 7259
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1268951684606623e-06,
+      "loss": 0.0105,
+      "step": 7260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1251428557894516e-06,
+      "loss": 0.0037,
+      "step": 7261
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1233918253761767e-06,
+      "loss": 0.001,
+      "step": 7262
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1216420774738324e-06,
+      "loss": 0.0011,
+      "step": 7263
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.11989361233522e-06,
+      "loss": 0.0026,
+      "step": 7264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1181464302129653e-06,
+      "loss": 0.0248,
+      "step": 7265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1164005313595038e-06,
+      "loss": 0.0081,
+      "step": 7266
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1146559160270875e-06,
+      "loss": 0.0347,
+      "step": 7267
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1129125844677812e-06,
+      "loss": 0.018,
+      "step": 7268
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1111705369334658e-06,
+      "loss": 0.007,
+      "step": 7269
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1094297736758375e-06,
+      "loss": 0.0047,
+      "step": 7270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.107690294946402e-06,
+      "loss": 0.1129,
+      "step": 7271
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1059521009964857e-06,
+      "loss": 0.0005,
+      "step": 7272
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1042151920772238e-06,
+      "loss": 0.0016,
+      "step": 7273
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1024795684395695e-06,
+      "loss": 0.0056,
+      "step": 7274
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1007452303342903e-06,
+      "loss": 0.0022,
+      "step": 7275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0990121780119668e-06,
+      "loss": 0.0076,
+      "step": 7276
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0972804117229929e-06,
+      "loss": 0.0034,
+      "step": 7277
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0955499317175789e-06,
+      "loss": 0.055,
+      "step": 7278
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0938207382457478e-06,
+      "loss": 0.0914,
+      "step": 7279
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0920928315573386e-06,
+      "loss": 0.0014,
+      "step": 7280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0903662119020008e-06,
+      "loss": 0.0679,
+      "step": 7281
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0886408795292003e-06,
+      "loss": 0.0133,
+      "step": 7282
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0869168346882164e-06,
+      "loss": 0.0058,
+      "step": 7283
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0851940776281444e-06,
+      "loss": 0.0457,
+      "step": 7284
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.083472608597892e-06,
+      "loss": 0.0008,
+      "step": 7285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0817524278461777e-06,
+      "loss": 0.0439,
+      "step": 7286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0800335356215386e-06,
+      "loss": 0.0599,
+      "step": 7287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0783159321723247e-06,
+      "loss": 0.0064,
+      "step": 7288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0765996177466998e-06,
+      "loss": 0.027,
+      "step": 7289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0748845925926366e-06,
+      "loss": 0.003,
+      "step": 7290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.073170856957928e-06,
+      "loss": 0.0115,
+      "step": 7291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.071458411090177e-06,
+      "loss": 0.1104,
+      "step": 7292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0697472552368038e-06,
+      "loss": 0.041,
+      "step": 7293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.068037389645038e-06,
+      "loss": 0.0078,
+      "step": 7294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0663288145619244e-06,
+      "loss": 0.0023,
+      "step": 7295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.064621530234322e-06,
+      "loss": 0.068,
+      "step": 7296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.062915536908905e-06,
+      "loss": 0.0477,
+      "step": 7297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.061210834832157e-06,
+      "loss": 0.0031,
+      "step": 7298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0595074242503745e-06,
+      "loss": 0.1441,
+      "step": 7299
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0578053054096714e-06,
+      "loss": 0.0049,
+      "step": 7300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0561044785559737e-06,
+      "loss": 0.028,
+      "step": 7301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0544049439350213e-06,
+      "loss": 0.0011,
+      "step": 7302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0527067017923654e-06,
+      "loss": 0.0322,
+      "step": 7303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0510097523733732e-06,
+      "loss": 0.0523,
+      "step": 7304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0493140959232206e-06,
+      "loss": 0.0243,
+      "step": 7305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0476197326869043e-06,
+      "loss": 0.0027,
+      "step": 7306
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0459266629092246e-06,
+      "loss": 0.0588,
+      "step": 7307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0442348868348018e-06,
+      "loss": 0.0168,
+      "step": 7308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0425444047080657e-06,
+      "loss": 0.0246,
+      "step": 7309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0408552167732634e-06,
+      "loss": 0.0294,
+      "step": 7310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0391673232744504e-06,
+      "loss": 0.0032,
+      "step": 7311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0374807244555008e-06,
+      "loss": 0.1251,
+      "step": 7312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0357954205600916e-06,
+      "loss": 0.0805,
+      "step": 7313
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.034111411831722e-06,
+      "loss": 0.0141,
+      "step": 7314
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0324286985137022e-06,
+      "loss": 0.0278,
+      "step": 7315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0307472808491547e-06,
+      "loss": 0.0266,
+      "step": 7316
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.029067159081012e-06,
+      "loss": 0.0045,
+      "step": 7317
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.027388333452022e-06,
+      "loss": 0.0023,
+      "step": 7318
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0257108042047448e-06,
+      "loss": 0.009,
+      "step": 7319
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0240345715815536e-06,
+      "loss": 0.0625,
+      "step": 7320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0223596358246346e-06,
+      "loss": 0.0058,
+      "step": 7321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0206859971759863e-06,
+      "loss": 0.0168,
+      "step": 7322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.019013655877421e-06,
+      "loss": 0.0046,
+      "step": 7323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0173426121705577e-06,
+      "loss": 0.0103,
+      "step": 7324
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0156728662968341e-06,
+      "loss": 0.0053,
+      "step": 7325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0140044184975017e-06,
+      "loss": 0.0021,
+      "step": 7326
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.012337269013617e-06,
+      "loss": 0.0037,
+      "step": 7327
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0106714180860543e-06,
+      "loss": 0.1078,
+      "step": 7328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0090068659554998e-06,
+      "loss": 0.0128,
+      "step": 7329
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0073436128624524e-06,
+      "loss": 0.0301,
+      "step": 7330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0056816590472206e-06,
+      "loss": 0.005,
+      "step": 7331
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0040210047499289e-06,
+      "loss": 0.0032,
+      "step": 7332
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0023616502105126e-06,
+      "loss": 0.0112,
+      "step": 7333
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.000703595668716e-06,
+      "loss": 0.0725,
+      "step": 7334
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.990468413640987e-07,
+      "loss": 0.002,
+      "step": 7335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.973913875360331e-07,
+      "loss": 0.0248,
+      "step": 7336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.957372344237026e-07,
+      "loss": 0.0214,
+      "step": 7337
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.940843822661018e-07,
+      "loss": 0.0038,
+      "step": 7338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.924328313020403e-07,
+      "loss": 0.0071,
+      "step": 7339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.907825817701367e-07,
+      "loss": 0.0045,
+      "step": 7340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.89133633908821e-07,
+      "loss": 0.0334,
+      "step": 7341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.874859879563404e-07,
+      "loss": 0.003,
+      "step": 7342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.858396441507445e-07,
+      "loss": 0.0169,
+      "step": 7343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.841946027299043e-07,
+      "loss": 0.0178,
+      "step": 7344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.825508639314972e-07,
+      "loss": 0.0009,
+      "step": 7345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.809084279930159e-07,
+      "loss": 0.0312,
+      "step": 7346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.792672951517611e-07,
+      "loss": 0.0202,
+      "step": 7347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.77627465644848e-07,
+      "loss": 0.0023,
+      "step": 7348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.759889397092038e-07,
+      "loss": 0.0464,
+      "step": 7349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.743517175815665e-07,
+      "loss": 0.0268,
+      "step": 7350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.727157994984815e-07,
+      "loss": 0.0456,
+      "step": 7351
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.71081185696312e-07,
+      "loss": 0.0011,
+      "step": 7352
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.694478764112303e-07,
+      "loss": 0.0415,
+      "step": 7353
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.67815871879223e-07,
+      "loss": 0.0491,
+      "step": 7354
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.661851723360804e-07,
+      "loss": 0.0488,
+      "step": 7355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.64555778017413e-07,
+      "loss": 0.01,
+      "step": 7356
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.62927689158637e-07,
+      "loss": 0.0053,
+      "step": 7357
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.61300905994984e-07,
+      "loss": 0.002,
+      "step": 7358
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59675428761495e-07,
+      "loss": 0.0479,
+      "step": 7359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.580512576930213e-07,
+      "loss": 0.0202,
+      "step": 7360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.564283930242258e-07,
+      "loss": 0.0731,
+      "step": 7361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.548068349895845e-07,
+      "loss": 0.0761,
+      "step": 7362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.531865838233833e-07,
+      "loss": 0.0364,
+      "step": 7363
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.515676397597206e-07,
+      "loss": 0.0354,
+      "step": 7364
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.499500030325026e-07,
+      "loss": 0.0005,
+      "step": 7365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.4833367387545e-07,
+      "loss": 0.0027,
+      "step": 7366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.467186525220962e-07,
+      "loss": 0.0131,
+      "step": 7367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.451049392057776e-07,
+      "loss": 0.0154,
+      "step": 7368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.434925341596513e-07,
+      "loss": 0.0551,
+      "step": 7369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.418814376166774e-07,
+      "loss": 0.0064,
+      "step": 7370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.402716498096331e-07,
+      "loss": 0.0197,
+      "step": 7371
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.386631709711014e-07,
+      "loss": 0.0007,
+      "step": 7372
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.370560013334806e-07,
+      "loss": 0.0037,
+      "step": 7373
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.354501411289785e-07,
+      "loss": 0.0016,
+      "step": 7374
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.338455905896115e-07,
+      "loss": 0.0787,
+      "step": 7375
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.32242349947211e-07,
+      "loss": 0.0523,
+      "step": 7376
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.306404194334129e-07,
+      "loss": 0.0125,
+      "step": 7377
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.290397992796696e-07,
+      "loss": 0.0013,
+      "step": 7378
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.274404897172418e-07,
+      "loss": 0.0163,
+      "step": 7379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.258424909772002e-07,
+      "loss": 0.017,
+      "step": 7380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.242458032904311e-07,
+      "loss": 0.0641,
+      "step": 7381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.226504268876213e-07,
+      "loss": 0.003,
+      "step": 7382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.210563619992763e-07,
+      "loss": 0.0023,
+      "step": 7383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.194636088557118e-07,
+      "loss": 0.0543,
+      "step": 7384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.178721676870495e-07,
+      "loss": 0.038,
+      "step": 7385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.162820387232285e-07,
+      "loss": 0.0689,
+      "step": 7386
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.146932221939875e-07,
+      "loss": 0.0032,
+      "step": 7387
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.131057183288872e-07,
+      "loss": 0.0316,
+      "step": 7388
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.115195273572908e-07,
+      "loss": 0.055,
+      "step": 7389
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.09934649508375e-07,
+      "loss": 0.0218,
+      "step": 7390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.083510850111277e-07,
+      "loss": 0.048,
+      "step": 7391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.067688340943437e-07,
+      "loss": 0.028,
+      "step": 7392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.051878969866323e-07,
+      "loss": 0.0281,
+      "step": 7393
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.036082739164087e-07,
+      "loss": 0.0456,
+      "step": 7394
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.020299651119014e-07,
+      "loss": 0.0022,
+      "step": 7395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.004529708011455e-07,
+      "loss": 0.0459,
+      "step": 7396
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.988772912119891e-07,
+      "loss": 0.0042,
+      "step": 7397
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.973029265720911e-07,
+      "loss": 0.0064,
+      "step": 7398
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.957298771089185e-07,
+      "loss": 0.0012,
+      "step": 7399
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.941581430497481e-07,
+      "loss": 0.0493,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.925877246216686e-07,
+      "loss": 0.0792,
+      "step": 7401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.91018622051576e-07,
+      "loss": 0.1617,
+      "step": 7402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.894508355661802e-07,
+      "loss": 0.0122,
+      "step": 7403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.878843653919944e-07,
+      "loss": 0.1027,
+      "step": 7404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.863192117553465e-07,
+      "loss": 0.0367,
+      "step": 7405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.847553748823745e-07,
+      "loss": 0.0798,
+      "step": 7406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.831928549990243e-07,
+      "loss": 0.0883,
+      "step": 7407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.816316523310519e-07,
+      "loss": 0.0146,
+      "step": 7408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.800717671040249e-07,
+      "loss": 0.074,
+      "step": 7409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.78513199543316e-07,
+      "loss": 0.0047,
+      "step": 7410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.769559498741109e-07,
+      "loss": 0.0799,
+      "step": 7411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.754000183214073e-07,
+      "loss": 0.0054,
+      "step": 7412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.738454051100042e-07,
+      "loss": 0.0122,
+      "step": 7413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.722921104645188e-07,
+      "loss": 0.0529,
+      "step": 7414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.707401346093736e-07,
+      "loss": 0.0272,
+      "step": 7415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.691894777688004e-07,
+      "loss": 0.0064,
+      "step": 7416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.676401401668432e-07,
+      "loss": 0.0444,
+      "step": 7417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.660921220273533e-07,
+      "loss": 0.0194,
+      "step": 7418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 0.0016,
+      "step": 7419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.630000450302267e-07,
+      "loss": 0.0756,
+      "step": 7420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.614559866193395e-07,
+      "loss": 0.0018,
+      "step": 7421
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.59913248564418e-07,
+      "loss": 0.0657,
+      "step": 7422
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.583718310883627e-07,
+      "loss": 0.0088,
+      "step": 7423
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.568317344138765e-07,
+      "loss": 0.0024,
+      "step": 7424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.552929587634784e-07,
+      "loss": 0.0028,
+      "step": 7425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.537555043594936e-07,
+      "loss": 0.0242,
+      "step": 7426
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.522193714240567e-07,
+      "loss": 0.0477,
+      "step": 7427
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.506845601791114e-07,
+      "loss": 0.0015,
+      "step": 7428
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.491510708464123e-07,
+      "loss": 0.0307,
+      "step": 7429
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.476189036475179e-07,
+      "loss": 0.1067,
+      "step": 7430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.460880588038001e-07,
+      "loss": 0.0158,
+      "step": 7431
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.445585365364384e-07,
+      "loss": 0.016,
+      "step": 7432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.430303370664228e-07,
+      "loss": 0.0064,
+      "step": 7433
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.415034606145489e-07,
+      "loss": 0.028,
+      "step": 7434
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.399779074014236e-07,
+      "loss": 0.0048,
+      "step": 7435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.384536776474639e-07,
+      "loss": 0.1295,
+      "step": 7436
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.369307715728924e-07,
+      "loss": 0.0268,
+      "step": 7437
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.354091893977401e-07,
+      "loss": 0.0155,
+      "step": 7438
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.338889313418519e-07,
+      "loss": 0.0018,
+      "step": 7439
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.323699976248733e-07,
+      "loss": 0.005,
+      "step": 7440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.308523884662656e-07,
+      "loss": 0.0147,
+      "step": 7441
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.293361040852954e-07,
+      "loss": 0.0647,
+      "step": 7442
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.278211447010398e-07,
+      "loss": 0.0782,
+      "step": 7443
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.263075105323837e-07,
+      "loss": 0.0534,
+      "step": 7444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.247952017980177e-07,
+      "loss": 0.0082,
+      "step": 7445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.232842187164469e-07,
+      "loss": 0.0565,
+      "step": 7446
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.21774561505978e-07,
+      "loss": 0.0019,
+      "step": 7447
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.202662303847298e-07,
+      "loss": 0.0214,
+      "step": 7448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.1875922557063e-07,
+      "loss": 0.016,
+      "step": 7449
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.172535472814158e-07,
+      "loss": 0.0035,
+      "step": 7450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.157491957346275e-07,
+      "loss": 0.0043,
+      "step": 7451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.142461711476168e-07,
+      "loss": 0.005,
+      "step": 7452
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.127444737375468e-07,
+      "loss": 0.0056,
+      "step": 7453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.112441037213836e-07,
+      "loss": 0.0053,
+      "step": 7454
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.097450613159042e-07,
+      "loss": 0.0069,
+      "step": 7455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.08247346737695e-07,
+      "loss": 0.0733,
+      "step": 7456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.067509602031465e-07,
+      "loss": 0.1454,
+      "step": 7457
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.052559019284611e-07,
+      "loss": 0.0531,
+      "step": 7458
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.037621721296473e-07,
+      "loss": 0.0021,
+      "step": 7459
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.022697710225235e-07,
+      "loss": 0.0068,
+      "step": 7460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.007786988227139e-07,
+      "loss": 0.0224,
+      "step": 7461
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.992889557456507e-07,
+      "loss": 0.0619,
+      "step": 7462
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.978005420065793e-07,
+      "loss": 0.0037,
+      "step": 7463
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.963134578205444e-07,
+      "loss": 0.0548,
+      "step": 7464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.948277034024055e-07,
+      "loss": 0.0519,
+      "step": 7465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.93343278966825e-07,
+      "loss": 0.021,
+      "step": 7466
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.918601847282781e-07,
+      "loss": 0.0446,
+      "step": 7467
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.903784209010434e-07,
+      "loss": 0.0185,
+      "step": 7468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.888979876992109e-07,
+      "loss": 0.0221,
+      "step": 7469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.87418885336676e-07,
+      "loss": 0.0039,
+      "step": 7470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.859411140271422e-07,
+      "loss": 0.0441,
+      "step": 7471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.844646739841222e-07,
+      "loss": 0.0358,
+      "step": 7472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.829895654209351e-07,
+      "loss": 0.0072,
+      "step": 7473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.815157885507064e-07,
+      "loss": 0.0345,
+      "step": 7474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.800433435863697e-07,
+      "loss": 0.002,
+      "step": 7475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.785722307406685e-07,
+      "loss": 0.0034,
+      "step": 7476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.771024502261526e-07,
+      "loss": 0.0734,
+      "step": 7477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.7563400225518e-07,
+      "loss": 0.0256,
+      "step": 7478
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.74166887039911e-07,
+      "loss": 0.0255,
+      "step": 7479
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.727011047923206e-07,
+      "loss": 0.0043,
+      "step": 7480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.71236655724188e-07,
+      "loss": 0.0024,
+      "step": 7481
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.697735400470996e-07,
+      "loss": 0.0563,
+      "step": 7482
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.683117579724487e-07,
+      "loss": 0.0616,
+      "step": 7483
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.668513097114361e-07,
+      "loss": 0.0143,
+      "step": 7484
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.653921954750721e-07,
+      "loss": 0.019,
+      "step": 7485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.639344154741713e-07,
+      "loss": 0.0043,
+      "step": 7486
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.624779699193574e-07,
+      "loss": 0.1173,
+      "step": 7487
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.610228590210611e-07,
+      "loss": 0.0131,
+      "step": 7488
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.595690829895186e-07,
+      "loss": 0.0109,
+      "step": 7489
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.581166420347785e-07,
+      "loss": 0.0621,
+      "step": 7490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.566655363666864e-07,
+      "loss": 0.0011,
+      "step": 7491
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.552157661949066e-07,
+      "loss": 0.0033,
+      "step": 7492
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.537673317288996e-07,
+      "loss": 0.0038,
+      "step": 7493
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.523202331779422e-07,
+      "loss": 0.0642,
+      "step": 7494
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.508744707511118e-07,
+      "loss": 0.0156,
+      "step": 7495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.494300446572978e-07,
+      "loss": 0.0007,
+      "step": 7496
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.47986955105191e-07,
+      "loss": 0.0016,
+      "step": 7497
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.465452023032949e-07,
+      "loss": 0.0032,
+      "step": 7498
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.451047864599159e-07,
+      "loss": 0.0026,
+      "step": 7499
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.436657077831677e-07,
+      "loss": 0.0348,
+      "step": 7500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.422279664809706e-07,
+      "loss": 0.007,
+      "step": 7501
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.40791562761054e-07,
+      "loss": 0.005,
+      "step": 7502
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.393564968309519e-07,
+      "loss": 0.0459,
+      "step": 7503
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.379227688980062e-07,
+      "loss": 0.0502,
+      "step": 7504
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.364903791693644e-07,
+      "loss": 0.0109,
+      "step": 7505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.350593278519824e-07,
+      "loss": 0.0649,
+      "step": 7506
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.336296151526201e-07,
+      "loss": 0.0011,
+      "step": 7507
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.322012412778445e-07,
+      "loss": 0.0074,
+      "step": 7508
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.307742064340329e-07,
+      "loss": 0.0403,
+      "step": 7509
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.293485108273624e-07,
+      "loss": 0.0072,
+      "step": 7510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.279241546638239e-07,
+      "loss": 0.0012,
+      "step": 7511
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.265011381492094e-07,
+      "loss": 0.0406,
+      "step": 7512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.250794614891199e-07,
+      "loss": 0.0069,
+      "step": 7513
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.236591248889624e-07,
+      "loss": 0.0203,
+      "step": 7514
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.222401285539505e-07,
+      "loss": 0.0287,
+      "step": 7515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.208224726891044e-07,
+      "loss": 0.0156,
+      "step": 7516
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.194061574992473e-07,
+      "loss": 0.0615,
+      "step": 7517
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.179911831890141e-07,
+      "loss": 0.0133,
+      "step": 7518
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.165775499628425e-07,
+      "loss": 0.0036,
+      "step": 7519
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.151652580249757e-07,
+      "loss": 0.0494,
+      "step": 7520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.137543075794651e-07,
+      "loss": 0.0551,
+      "step": 7521
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.123446988301685e-07,
+      "loss": 0.0245,
+      "step": 7522
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.109364319807499e-07,
+      "loss": 0.1005,
+      "step": 7523
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.095295072346775e-07,
+      "loss": 0.0085,
+      "step": 7524
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.081239247952265e-07,
+      "loss": 0.0804,
+      "step": 7525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0671968486548e-07,
+      "loss": 0.0201,
+      "step": 7526
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.053167876483235e-07,
+      "loss": 0.004,
+      "step": 7527
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.039152333464517e-07,
+      "loss": 0.0355,
+      "step": 7528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.025150221623633e-07,
+      "loss": 0.1169,
+      "step": 7529
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.011161542983647e-07,
+      "loss": 0.006,
+      "step": 7530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.997186299565661e-07,
+      "loss": 0.0067,
+      "step": 7531
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.983224493388863e-07,
+      "loss": 0.0063,
+      "step": 7532
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.969276126470493e-07,
+      "loss": 0.0474,
+      "step": 7533
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.955341200825794e-07,
+      "loss": 0.0263,
+      "step": 7534
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.941419718468168e-07,
+      "loss": 0.0692,
+      "step": 7535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.927511681408971e-07,
+      "loss": 0.0706,
+      "step": 7536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.913617091657698e-07,
+      "loss": 0.0031,
+      "step": 7537
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.899735951221853e-07,
+      "loss": 0.0066,
+      "step": 7538
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.88586826210702e-07,
+      "loss": 0.0064,
+      "step": 7539
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.872014026316831e-07,
+      "loss": 0.0031,
+      "step": 7540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.858173245852962e-07,
+      "loss": 0.0016,
+      "step": 7541
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.844345922715179e-07,
+      "loss": 0.0015,
+      "step": 7542
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.830532058901296e-07,
+      "loss": 0.0009,
+      "step": 7543
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.816731656407116e-07,
+      "loss": 0.0555,
+      "step": 7544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.802944717226589e-07,
+      "loss": 0.0128,
+      "step": 7545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.789171243351678e-07,
+      "loss": 0.0918,
+      "step": 7546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7754112367724e-07,
+      "loss": 0.0946,
+      "step": 7547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.761664699476811e-07,
+      "loss": 0.0093,
+      "step": 7548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.747931633451066e-07,
+      "loss": 0.0448,
+      "step": 7549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.734212040679333e-07,
+      "loss": 0.0414,
+      "step": 7550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.72050592314385e-07,
+      "loss": 0.0014,
+      "step": 7551
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.70681328282492e-07,
+      "loss": 0.0712,
+      "step": 7552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.69313412170085e-07,
+      "loss": 0.022,
+      "step": 7553
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.679468441748061e-07,
+      "loss": 0.0046,
+      "step": 7554
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.665816244940992e-07,
+      "loss": 0.0018,
+      "step": 7555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.652177533252146e-07,
+      "loss": 0.0858,
+      "step": 7556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.638552308652069e-07,
+      "loss": 0.0358,
+      "step": 7557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.62494057310935e-07,
+      "loss": 0.0054,
+      "step": 7558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.611342328590664e-07,
+      "loss": 0.0034,
+      "step": 7559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.597757577060715e-07,
+      "loss": 0.0528,
+      "step": 7560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.584186320482222e-07,
+      "loss": 0.025,
+      "step": 7561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.570628560816028e-07,
+      "loss": 0.0214,
+      "step": 7562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.557084300020955e-07,
+      "loss": 0.0034,
+      "step": 7563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.543553540053926e-07,
+      "loss": 0.0589,
+      "step": 7564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.530036282869879e-07,
+      "loss": 0.0044,
+      "step": 7565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.516532530421826e-07,
+      "loss": 0.1055,
+      "step": 7566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.50304228466081e-07,
+      "loss": 0.0058,
+      "step": 7567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.489565547535947e-07,
+      "loss": 0.0021,
+      "step": 7568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.476102320994371e-07,
+      "loss": 0.0306,
+      "step": 7569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.462652606981279e-07,
+      "loss": 0.0047,
+      "step": 7570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.449216407439906e-07,
+      "loss": 0.0062,
+      "step": 7571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.435793724311546e-07,
+      "loss": 0.0062,
+      "step": 7572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.422384559535543e-07,
+      "loss": 0.0366,
+      "step": 7573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.408988915049286e-07,
+      "loss": 0.0908,
+      "step": 7574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3956067927882e-07,
+      "loss": 0.0005,
+      "step": 7575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.382238194685752e-07,
+      "loss": 0.0028,
+      "step": 7576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.368883122673475e-07,
+      "loss": 0.017,
+      "step": 7577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.355541578680935e-07,
+      "loss": 0.0566,
+      "step": 7578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.342213564635757e-07,
+      "loss": 0.0469,
+      "step": 7579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.32889908246358e-07,
+      "loss": 0.0514,
+      "step": 7580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.315598134088119e-07,
+      "loss": 0.0077,
+      "step": 7581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.302310721431115e-07,
+      "loss": 0.0162,
+      "step": 7582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.289036846412378e-07,
+      "loss": 0.0013,
+      "step": 7583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.275776510949727e-07,
+      "loss": 0.0363,
+      "step": 7584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.262529716959053e-07,
+      "loss": 0.0023,
+      "step": 7585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.249296466354305e-07,
+      "loss": 0.0228,
+      "step": 7586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.236076761047394e-07,
+      "loss": 0.0349,
+      "step": 7587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.222870602948372e-07,
+      "loss": 0.0551,
+      "step": 7588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.209677993965302e-07,
+      "loss": 0.1051,
+      "step": 7589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.196498936004236e-07,
+      "loss": 0.042,
+      "step": 7590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.18333343096933e-07,
+      "loss": 0.0022,
+      "step": 7591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.170181480762782e-07,
+      "loss": 0.0072,
+      "step": 7592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.157043087284797e-07,
+      "loss": 0.0009,
+      "step": 7593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.143918252433645e-07,
+      "loss": 0.0043,
+      "step": 7594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.130806978105619e-07,
+      "loss": 0.0046,
+      "step": 7595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.117709266195083e-07,
+      "loss": 0.0012,
+      "step": 7596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.104625118594398e-07,
+      "loss": 0.0235,
+      "step": 7597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.091554537194e-07,
+      "loss": 0.0245,
+      "step": 7598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.078497523882354e-07,
+      "loss": 0.0105,
+      "step": 7599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.065454080545974e-07,
+      "loss": 0.0203,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.052424209069385e-07,
+      "loss": 0.0288,
+      "step": 7601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.039407911335193e-07,
+      "loss": 0.0446,
+      "step": 7602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.026405189223994e-07,
+      "loss": 0.0132,
+      "step": 7603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.013416044614462e-07,
+      "loss": 0.0351,
+      "step": 7604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.000440479383319e-07,
+      "loss": 0.001,
+      "step": 7605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.987478495405252e-07,
+      "loss": 0.0048,
+      "step": 7606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.974530094553066e-07,
+      "loss": 0.0024,
+      "step": 7607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.961595278697562e-07,
+      "loss": 0.0222,
+      "step": 7608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.948674049707604e-07,
+      "loss": 0.0095,
+      "step": 7609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.935766409450072e-07,
+      "loss": 0.0855,
+      "step": 7610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.922872359789878e-07,
+      "loss": 0.0463,
+      "step": 7611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.909991902589996e-07,
+      "loss": 0.0259,
+      "step": 7612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.897125039711426e-07,
+      "loss": 0.0018,
+      "step": 7613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.884271773013172e-07,
+      "loss": 0.0031,
+      "step": 7614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.871432104352326e-07,
+      "loss": 0.0339,
+      "step": 7615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.858606035583992e-07,
+      "loss": 0.0308,
+      "step": 7616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.845793568561287e-07,
+      "loss": 0.0006,
+      "step": 7617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.832994705135387e-07,
+      "loss": 0.0671,
+      "step": 7618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.820209447155511e-07,
+      "loss": 0.0342,
+      "step": 7619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.807437796468885e-07,
+      "loss": 0.0053,
+      "step": 7620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.794679754920796e-07,
+      "loss": 0.0104,
+      "step": 7621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.781935324354571e-07,
+      "loss": 0.0931,
+      "step": 7622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.769204506611503e-07,
+      "loss": 0.139,
+      "step": 7623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.756487303531011e-07,
+      "loss": 0.0016,
+      "step": 7624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.743783716950468e-07,
+      "loss": 0.0069,
+      "step": 7625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.731093748705341e-07,
+      "loss": 0.0092,
+      "step": 7626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.718417400629106e-07,
+      "loss": 0.0031,
+      "step": 7627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.705754674553243e-07,
+      "loss": 0.0012,
+      "step": 7628
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.693105572307311e-07,
+      "loss": 0.0054,
+      "step": 7629
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.680470095718882e-07,
+      "loss": 0.0024,
+      "step": 7630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.667848246613528e-07,
+      "loss": 0.0115,
+      "step": 7631
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.655240026814923e-07,
+      "loss": 0.003,
+      "step": 7632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.642645438144678e-07,
+      "loss": 0.024,
+      "step": 7633
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.630064482422526e-07,
+      "loss": 0.0042,
+      "step": 7634
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.617497161466157e-07,
+      "loss": 0.0332,
+      "step": 7635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.604943477091351e-07,
+      "loss": 0.0321,
+      "step": 7636
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.592403431111881e-07,
+      "loss": 0.0041,
+      "step": 7637
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.579877025339564e-07,
+      "loss": 0.0044,
+      "step": 7638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.567364261584251e-07,
+      "loss": 0.0251,
+      "step": 7639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.554865141653776e-07,
+      "loss": 0.0176,
+      "step": 7640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.54237966735407e-07,
+      "loss": 0.0018,
+      "step": 7641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.529907840489046e-07,
+      "loss": 0.0077,
+      "step": 7642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.517449662860663e-07,
+      "loss": 0.0027,
+      "step": 7643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.505005136268915e-07,
+      "loss": 0.0356,
+      "step": 7644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.492574262511796e-07,
+      "loss": 0.0512,
+      "step": 7645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.480157043385359e-07,
+      "loss": 0.0198,
+      "step": 7646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.467753480683658e-07,
+      "loss": 0.0074,
+      "step": 7647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.455363576198791e-07,
+      "loss": 0.0565,
+      "step": 7648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.442987331720894e-07,
+      "loss": 0.0071,
+      "step": 7649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.430624749038083e-07,
+      "loss": 0.0466,
+      "step": 7650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.418275829936537e-07,
+      "loss": 0.0043,
+      "step": 7651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.405940576200463e-07,
+      "loss": 0.0042,
+      "step": 7652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.393618989612082e-07,
+      "loss": 0.0788,
+      "step": 7653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.381311071951645e-07,
+      "loss": 0.0034,
+      "step": 7654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.36901682499742e-07,
+      "loss": 0.0125,
+      "step": 7655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.356736250525718e-07,
+      "loss": 0.0519,
+      "step": 7656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.344469350310843e-07,
+      "loss": 0.0025,
+      "step": 7657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.332216126125167e-07,
+      "loss": 0.0036,
+      "step": 7658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.319976579739028e-07,
+      "loss": 0.0281,
+      "step": 7659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.307750712920845e-07,
+      "loss": 0.029,
+      "step": 7660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.295538527437027e-07,
+      "loss": 0.0779,
+      "step": 7661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28334002505202e-07,
+      "loss": 0.0435,
+      "step": 7662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.271155207528289e-07,
+      "loss": 0.0345,
+      "step": 7663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.258984076626328e-07,
+      "loss": 0.0009,
+      "step": 7664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.246826634104629e-07,
+      "loss": 0.0083,
+      "step": 7665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.234682881719766e-07,
+      "loss": 0.1025,
+      "step": 7666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.222552821226246e-07,
+      "loss": 0.0135,
+      "step": 7667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.210436454376666e-07,
+      "loss": 0.0386,
+      "step": 7668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.198333782921616e-07,
+      "loss": 0.0973,
+      "step": 7669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.186244808609731e-07,
+      "loss": 0.022,
+      "step": 7670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.174169533187634e-07,
+      "loss": 0.0223,
+      "step": 7671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.16210795840002e-07,
+      "loss": 0.0036,
+      "step": 7672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.150060085989516e-07,
+      "loss": 0.0521,
+      "step": 7673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.138025917696865e-07,
+      "loss": 0.0067,
+      "step": 7674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.126005455260785e-07,
+      "loss": 0.0408,
+      "step": 7675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.113998700418e-07,
+      "loss": 0.0063,
+      "step": 7676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.102005654903286e-07,
+      "loss": 0.0543,
+      "step": 7677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.090026320449404e-07,
+      "loss": 0.0536,
+      "step": 7678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.078060698787179e-07,
+      "loss": 0.0217,
+      "step": 7679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.066108791645407e-07,
+      "loss": 0.0011,
+      "step": 7680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.05417060075094e-07,
+      "loss": 0.0657,
+      "step": 7681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.042246127828632e-07,
+      "loss": 0.0072,
+      "step": 7682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.030335374601358e-07,
+      "loss": 0.02,
+      "step": 7683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.018438342789989e-07,
+      "loss": 0.007,
+      "step": 7684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.006555034113459e-07,
+      "loss": 0.0178,
+      "step": 7685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.994685450288672e-07,
+      "loss": 0.1373,
+      "step": 7686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.98282959303058e-07,
+      "loss": 0.0033,
+      "step": 7687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.970987464052134e-07,
+      "loss": 0.1302,
+      "step": 7688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.959159065064311e-07,
+      "loss": 0.0273,
+      "step": 7689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.947344397776121e-07,
+      "loss": 0.0229,
+      "step": 7690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.935543463894554e-07,
+      "loss": 0.0227,
+      "step": 7691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.923756265124646e-07,
+      "loss": 0.0032,
+      "step": 7692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.911982803169424e-07,
+      "loss": 0.0005,
+      "step": 7693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.900223079729949e-07,
+      "loss": 0.007,
+      "step": 7694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888477096505284e-07,
+      "loss": 0.1346,
+      "step": 7695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.876744855192516e-07,
+      "loss": 0.0328,
+      "step": 7696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.865026357486746e-07,
+      "loss": 0.0008,
+      "step": 7697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853321605081085e-07,
+      "loss": 0.0484,
+      "step": 7698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.84163059966668e-07,
+      "loss": 0.05,
+      "step": 7699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.829953342932637e-07,
+      "loss": 0.0419,
+      "step": 7700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.818289836566126e-07,
+      "loss": 0.0389,
+      "step": 7701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.806640082252334e-07,
+      "loss": 0.0334,
+      "step": 7702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.795004081674403e-07,
+      "loss": 0.0004,
+      "step": 7703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.783381836513557e-07,
+      "loss": 0.0005,
+      "step": 7704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.771773348448994e-07,
+      "loss": 0.0281,
+      "step": 7705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7601786191579293e-07,
+      "loss": 0.0101,
+      "step": 7706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7485976503155896e-07,
+      "loss": 0.0609,
+      "step": 7707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7370304435952254e-07,
+      "loss": 0.0304,
+      "step": 7708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7254770006681105e-07,
+      "loss": 0.0408,
+      "step": 7709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7139373232034656e-07,
+      "loss": 0.0088,
+      "step": 7710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7024114128686017e-07,
+      "loss": 0.0081,
+      "step": 7711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.690899271328786e-07,
+      "loss": 0.0692,
+      "step": 7712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6794009002473437e-07,
+      "loss": 0.024,
+      "step": 7713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.667916301285558e-07,
+      "loss": 0.1287,
+      "step": 7714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6564454761027555e-07,
+      "loss": 0.057,
+      "step": 7715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6449884263562564e-07,
+      "loss": 0.0717,
+      "step": 7716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6335451537014244e-07,
+      "loss": 0.0015,
+      "step": 7717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.622115659791582e-07,
+      "loss": 0.0079,
+      "step": 7718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.610699946278108e-07,
+      "loss": 0.0155,
+      "step": 7719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.599298014810349e-07,
+      "loss": 0.0028,
+      "step": 7720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.587909867035678e-07,
+      "loss": 0.0069,
+      "step": 7721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5765355045995e-07,
+      "loss": 0.0646,
+      "step": 7722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5651749291451884e-07,
+      "loss": 0.0295,
+      "step": 7723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553828142314154e-07,
+      "loss": 0.0031,
+      "step": 7724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.542495145745796e-07,
+      "loss": 0.007,
+      "step": 7725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.531175941077537e-07,
+      "loss": 0.0466,
+      "step": 7726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.519870529944803e-07,
+      "loss": 0.0288,
+      "step": 7727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.508578913981021e-07,
+      "loss": 0.0053,
+      "step": 7728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4973010948176075e-07,
+      "loss": 0.0037,
+      "step": 7729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4860370740840263e-07,
+      "loss": 0.0214,
+      "step": 7730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4747868534077197e-07,
+      "loss": 0.0014,
+      "step": 7731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.463550434414132e-07,
+      "loss": 0.1427,
+      "step": 7732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.452327818726743e-07,
+      "loss": 0.0919,
+      "step": 7733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4411190079670207e-07,
+      "loss": 0.0028,
+      "step": 7734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.429924003754427e-07,
+      "loss": 0.1108,
+      "step": 7735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4187428077064443e-07,
+      "loss": 0.096,
+      "step": 7736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.407575421438548e-07,
+      "loss": 0.0137,
+      "step": 7737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.396421846564236e-07,
+      "loss": 0.0634,
+      "step": 7738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3852820846949973e-07,
+      "loss": 0.1036,
+      "step": 7739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3741561374403107e-07,
+      "loss": 0.0015,
+      "step": 7740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.363044006407724e-07,
+      "loss": 0.0038,
+      "step": 7741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.351945693202686e-07,
+      "loss": 0.0522,
+      "step": 7742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3408611994287255e-07,
+      "loss": 0.0155,
+      "step": 7743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3297905266873495e-07,
+      "loss": 0.0126,
+      "step": 7744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.318733676578102e-07,
+      "loss": 0.0026,
+      "step": 7745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3076906506984595e-07,
+      "loss": 0.095,
+      "step": 7746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.296661450643946e-07,
+      "loss": 0.0339,
+      "step": 7747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.28564607800811e-07,
+      "loss": 0.0082,
+      "step": 7748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.274644534382455e-07,
+      "loss": 0.0872,
+      "step": 7749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.26365682135651e-07,
+      "loss": 0.0033,
+      "step": 7750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.252682940517805e-07,
+      "loss": 0.0074,
+      "step": 7751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2417228934518717e-07,
+      "loss": 0.0498,
+      "step": 7752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2307766817422437e-07,
+      "loss": 0.0086,
+      "step": 7753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2198443069704443e-07,
+      "loss": 0.0017,
+      "step": 7754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.20892577071601e-07,
+      "loss": 0.0685,
+      "step": 7755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.198021074556469e-07,
+      "loss": 0.0092,
+      "step": 7756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.187130220067359e-07,
+      "loss": 0.0064,
+      "step": 7757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1762532088222116e-07,
+      "loss": 0.0264,
+      "step": 7758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.165390042392559e-07,
+      "loss": 0.0042,
+      "step": 7759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.154540722347933e-07,
+      "loss": 0.0703,
+      "step": 7760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1437052502558693e-07,
+      "loss": 0.0037,
+      "step": 7761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.132883627681916e-07,
+      "loss": 0.0024,
+      "step": 7762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1220758561895777e-07,
+      "loss": 0.0039,
+      "step": 7763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1112819373403834e-07,
+      "loss": 0.0005,
+      "step": 7764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.100501872693874e-07,
+      "loss": 0.0009,
+      "step": 7765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0897356638075815e-07,
+      "loss": 0.075,
+      "step": 7766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.078983312237017e-07,
+      "loss": 0.0047,
+      "step": 7767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0682448195357163e-07,
+      "loss": 0.0023,
+      "step": 7768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.057520187255182e-07,
+      "loss": 0.0237,
+      "step": 7769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0468094169449414e-07,
+      "loss": 0.0108,
+      "step": 7770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0361125101525124e-07,
+      "loss": 0.0898,
+      "step": 7771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.025429468423403e-07,
+      "loss": 0.0062,
+      "step": 7772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.014760293301112e-07,
+      "loss": 0.0008,
+      "step": 7773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0041049863271506e-07,
+      "loss": 0.0018,
+      "step": 7774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.993463549041021e-07,
+      "loss": 0.0392,
+      "step": 7775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9828359829802263e-07,
+      "loss": 0.0761,
+      "step": 7776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.97222228968025e-07,
+      "loss": 0.0059,
+      "step": 7777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961622470674586e-07,
+      "loss": 0.0221,
+      "step": 7778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9510365274947227e-07,
+      "loss": 0.0057,
+      "step": 7779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.940464461670135e-07,
+      "loss": 0.1014,
+      "step": 7780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.92990627472829e-07,
+      "loss": 0.0435,
+      "step": 7781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.919361968194668e-07,
+      "loss": 0.1051,
+      "step": 7782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.908831543592728e-07,
+      "loss": 0.0067,
+      "step": 7783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8983150024439196e-07,
+      "loss": 0.064,
+      "step": 7784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8878123462676944e-07,
+      "loss": 0.0087,
+      "step": 7785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8773235765815155e-07,
+      "loss": 0.133,
+      "step": 7786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8668486949008045e-07,
+      "loss": 0.0049,
+      "step": 7787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8563877027390175e-07,
+      "loss": 0.0297,
+      "step": 7788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.845940601607556e-07,
+      "loss": 0.0127,
+      "step": 7789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.835507393015847e-07,
+      "loss": 0.0078,
+      "step": 7790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8250880784713064e-07,
+      "loss": 0.0209,
+      "step": 7791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8146826594793293e-07,
+      "loss": 0.077,
+      "step": 7792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8042911375433253e-07,
+      "loss": 0.0109,
+      "step": 7793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.793913514164671e-07,
+      "loss": 0.0056,
+      "step": 7794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7835497908427555e-07,
+      "loss": 0.0019,
+      "step": 7795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.773199969074959e-07,
+      "loss": 0.006,
+      "step": 7796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.762864050356629e-07,
+      "loss": 0.0208,
+      "step": 7797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7525420361811396e-07,
+      "loss": 0.0649,
+      "step": 7798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.742233928039796e-07,
+      "loss": 0.0226,
+      "step": 7799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7319397274219735e-07,
+      "loss": 0.0975,
+      "step": 7800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.721659435814995e-07,
+      "loss": 0.004,
+      "step": 7801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.711393054704171e-07,
+      "loss": 0.048,
+      "step": 7802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.701140585572804e-07,
+      "loss": 0.0044,
+      "step": 7803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6909020299022103e-07,
+      "loss": 0.0062,
+      "step": 7804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.680677389171661e-07,
+      "loss": 0.0033,
+      "step": 7805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.670466664858452e-07,
+      "loss": 0.0026,
+      "step": 7806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.660269858437826e-07,
+      "loss": 0.0149,
+      "step": 7807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.650086971383049e-07,
+      "loss": 0.0173,
+      "step": 7808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.639918005165377e-07,
+      "loss": 0.0027,
+      "step": 7809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6297629612540465e-07,
+      "loss": 0.0047,
+      "step": 7810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.61962184111625e-07,
+      "loss": 0.0346,
+      "step": 7811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.609494646217215e-07,
+      "loss": 0.0153,
+      "step": 7812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5993813780201394e-07,
+      "loss": 0.005,
+      "step": 7813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.589282037986208e-07,
+      "loss": 0.004,
+      "step": 7814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5791966275746213e-07,
+      "loss": 0.011,
+      "step": 7815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5691251482424914e-07,
+      "loss": 0.005,
+      "step": 7816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5590676014449987e-07,
+      "loss": 0.0191,
+      "step": 7817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5490239886352584e-07,
+      "loss": 0.0143,
+      "step": 7818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.538994311264421e-07,
+      "loss": 0.0117,
+      "step": 7819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5289785707815716e-07,
+      "loss": 0.0074,
+      "step": 7820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5189767686338196e-07,
+      "loss": 0.0043,
+      "step": 7821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5089889062662306e-07,
+      "loss": 0.0487,
+      "step": 7822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4990149851219067e-07,
+      "loss": 0.0051,
+      "step": 7823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4890550066418506e-07,
+      "loss": 0.0345,
+      "step": 7824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4791089722651437e-07,
+      "loss": 0.0037,
+      "step": 7825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469176883428782e-07,
+      "loss": 0.1458,
+      "step": 7826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.459258741567784e-07,
+      "loss": 0.0034,
+      "step": 7827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4493545481151356e-07,
+      "loss": 0.0174,
+      "step": 7828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4394643045018385e-07,
+      "loss": 0.045,
+      "step": 7829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.429588012156826e-07,
+      "loss": 0.0041,
+      "step": 7830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.419725672507068e-07,
+      "loss": 0.0482,
+      "step": 7831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4098772869775033e-07,
+      "loss": 0.0231,
+      "step": 7832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.400042856991015e-07,
+      "loss": 0.0265,
+      "step": 7833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3902223839685224e-07,
+      "loss": 0.0087,
+      "step": 7834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3804158693289124e-07,
+      "loss": 0.0008,
+      "step": 7835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3706233144890296e-07,
+      "loss": 0.0041,
+      "step": 7836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.360844720863765e-07,
+      "loss": 0.0083,
+      "step": 7837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.351080089865899e-07,
+      "loss": 0.0134,
+      "step": 7838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.34132942290627e-07,
+      "loss": 0.0025,
+      "step": 7839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3315927213936726e-07,
+      "loss": 0.0272,
+      "step": 7840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3218699867348937e-07,
+      "loss": 0.0031,
+      "step": 7841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3121612203346975e-07,
+      "loss": 0.0341,
+      "step": 7842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.302466423595807e-07,
+      "loss": 0.0014,
+      "step": 7843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.292785597918957e-07,
+      "loss": 0.0353,
+      "step": 7844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.283118744702851e-07,
+      "loss": 0.0292,
+      "step": 7845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.273465865344172e-07,
+      "loss": 0.0726,
+      "step": 7846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.263826961237604e-07,
+      "loss": 0.0041,
+      "step": 7847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2542020337757775e-07,
+      "loss": 0.0343,
+      "step": 7848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2445910843493466e-07,
+      "loss": 0.0366,
+      "step": 7849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.234994114346901e-07,
+      "loss": 0.0139,
+      "step": 7850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2254111251550424e-07,
+      "loss": 0.0443,
+      "step": 7851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.21584211815833e-07,
+      "loss": 0.0023,
+      "step": 7852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.206287094739313e-07,
+      "loss": 0.0106,
+      "step": 7853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1967460562785325e-07,
+      "loss": 0.0218,
+      "step": 7854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.187219004154485e-07,
+      "loss": 0.1001,
+      "step": 7855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1777059397436693e-07,
+      "loss": 0.002,
+      "step": 7856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.168206864420553e-07,
+      "loss": 0.0006,
+      "step": 7857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.158721779557572e-07,
+      "loss": 0.0034,
+      "step": 7858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1492506865251737e-07,
+      "loss": 0.0048,
+      "step": 7859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.139793586691742e-07,
+      "loss": 0.0733,
+      "step": 7860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1303504814236494e-07,
+      "loss": 0.0141,
+      "step": 7861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.120921372085284e-07,
+      "loss": 0.0524,
+      "step": 7862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.111506260038955e-07,
+      "loss": 0.0323,
+      "step": 7863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.102105146644996e-07,
+      "loss": 0.0007,
+      "step": 7864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.092718033261699e-07,
+      "loss": 0.0011,
+      "step": 7865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0833449212453237e-07,
+      "loss": 0.003,
+      "step": 7866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0739858119501197e-07,
+      "loss": 0.025,
+      "step": 7867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.064640706728328e-07,
+      "loss": 0.0057,
+      "step": 7868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0553096069301234e-07,
+      "loss": 0.0371,
+      "step": 7869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.045992513903695e-07,
+      "loss": 0.0191,
+      "step": 7870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0366894289951875e-07,
+      "loss": 0.0056,
+      "step": 7871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.027400353548737e-07,
+      "loss": 0.0082,
+      "step": 7872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.018125288906448e-07,
+      "loss": 0.0306,
+      "step": 7873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0088642364084044e-07,
+      "loss": 0.1216,
+      "step": 7874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9996171973926567e-07,
+      "loss": 0.0099,
+      "step": 7875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.99038417319526e-07,
+      "loss": 0.0021,
+      "step": 7876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9811651651501794e-07,
+      "loss": 0.0344,
+      "step": 7877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9719601745894277e-07,
+      "loss": 0.0053,
+      "step": 7878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9627692028429523e-07,
+      "loss": 0.0509,
+      "step": 7879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9535922512386794e-07,
+      "loss": 0.0206,
+      "step": 7880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.944429321102504e-07,
+      "loss": 0.079,
+      "step": 7881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9352804137583344e-07,
+      "loss": 0.0262,
+      "step": 7882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.926145530528002e-07,
+      "loss": 0.0061,
+      "step": 7883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.917024672731339e-07,
+      "loss": 0.0728,
+      "step": 7884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9079178416861587e-07,
+      "loss": 0.0997,
+      "step": 7885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8988250387082083e-07,
+      "loss": 0.0035,
+      "step": 7886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8897462651112596e-07,
+      "loss": 0.0089,
+      "step": 7887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.880681522207018e-07,
+      "loss": 0.0019,
+      "step": 7888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.871630811305182e-07,
+      "loss": 0.0222,
+      "step": 7889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8625941337134254e-07,
+      "loss": 0.0019,
+      "step": 7890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.853571490737372e-07,
+      "loss": 0.0065,
+      "step": 7891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.844562883680635e-07,
+      "loss": 0.0188,
+      "step": 7892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.835568313844816e-07,
+      "loss": 0.0852,
+      "step": 7893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.826587782529444e-07,
+      "loss": 0.0606,
+      "step": 7894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8176212910320577e-07,
+      "loss": 0.0513,
+      "step": 7895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8086688406481546e-07,
+      "loss": 0.0488,
+      "step": 7896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7997304326711884e-07,
+      "loss": 0.0032,
+      "step": 7897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7908060683926154e-07,
+      "loss": 0.0191,
+      "step": 7898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.781895749101837e-07,
+      "loss": 0.0016,
+      "step": 7899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7729994760862467e-07,
+      "loss": 0.0167,
+      "step": 7900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.764117250631182e-07,
+      "loss": 0.0012,
+      "step": 7901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.755249074019972e-07,
+      "loss": 0.0104,
+      "step": 7902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.746394947533926e-07,
+      "loss": 0.043,
+      "step": 7903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.737554872452286e-07,
+      "loss": 0.0016,
+      "step": 7904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.728728850052287e-07,
+      "loss": 0.0219,
+      "step": 7905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7199168816091416e-07,
+      "loss": 0.0016,
+      "step": 7906
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.711118968395998e-07,
+      "loss": 0.0153,
+      "step": 7907
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.702335111684018e-07,
+      "loss": 0.0269,
+      "step": 7908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6935653127423077e-07,
+      "loss": 0.0729,
+      "step": 7909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.684809572837943e-07,
+      "loss": 0.0024,
+      "step": 7910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6760678932359787e-07,
+      "loss": 0.051,
+      "step": 7911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.667340275199426e-07,
+      "loss": 0.0155,
+      "step": 7912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.658626719989266e-07,
+      "loss": 0.0011,
+      "step": 7913
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.649927228864446e-07,
+      "loss": 0.0069,
+      "step": 7914
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6412418030819064e-07,
+      "loss": 0.004,
+      "step": 7915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6325704438965206e-07,
+      "loss": 0.011,
+      "step": 7916
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.623913152561142e-07,
+      "loss": 0.009,
+      "step": 7917
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6152699303266025e-07,
+      "loss": 0.1248,
+      "step": 7918
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6066407784416937e-07,
+      "loss": 0.0402,
+      "step": 7919
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.598025698153184e-07,
+      "loss": 0.0169,
+      "step": 7920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.589424690705777e-07,
+      "loss": 0.0007,
+      "step": 7921
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.580837757342169e-07,
+      "loss": 0.0634,
+      "step": 7922
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.572264899303023e-07,
+      "loss": 0.0912,
+      "step": 7923
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.563706117826958e-07,
+      "loss": 0.1363,
+      "step": 7924
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5551614141505756e-07,
+      "loss": 0.0535,
+      "step": 7925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5466307895084194e-07,
+      "loss": 0.0665,
+      "step": 7926
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5381142451330165e-07,
+      "loss": 0.0827,
+      "step": 7927
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5296117822548594e-07,
+      "loss": 0.0534,
+      "step": 7928
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.521123402102399e-07,
+      "loss": 0.0642,
+      "step": 7929
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.512649105902054e-07,
+      "loss": 0.0034,
+      "step": 7930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5041888948782124e-07,
+      "loss": 0.0052,
+      "step": 7931
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4957427702532177e-07,
+      "loss": 0.0209,
+      "step": 7932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4873107332473724e-07,
+      "loss": 0.0059,
+      "step": 7933
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.47889278507899e-07,
+      "loss": 0.0647,
+      "step": 7934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.470488926964276e-07,
+      "loss": 0.0086,
+      "step": 7935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4620991601174596e-07,
+      "loss": 0.0178,
+      "step": 7936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.453723485750703e-07,
+      "loss": 0.0604,
+      "step": 7937
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.445361905074151e-07,
+      "loss": 0.0054,
+      "step": 7938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.43701441929588e-07,
+      "loss": 0.0513,
+      "step": 7939
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.428681029621971e-07,
+      "loss": 0.0208,
+      "step": 7940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.420361737256438e-07,
+      "loss": 0.0065,
+      "step": 7941
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4120565434012866e-07,
+      "loss": 0.0365,
+      "step": 7942
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4037654492564565e-07,
+      "loss": 0.0311,
+      "step": 7943
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3954884560198567e-07,
+      "loss": 0.006,
+      "step": 7944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3872255648873856e-07,
+      "loss": 0.0313,
+      "step": 7945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.378976777052866e-07,
+      "loss": 0.0055,
+      "step": 7946
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3707420937081005e-07,
+      "loss": 0.0003,
+      "step": 7947
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3625215160428705e-07,
+      "loss": 0.0277,
+      "step": 7948
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3543150452448705e-07,
+      "loss": 0.0046,
+      "step": 7949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3461226824998073e-07,
+      "loss": 0.0485,
+      "step": 7950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3379444289913344e-07,
+      "loss": 0.0351,
+      "step": 7951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.329780285901062e-07,
+      "loss": 0.0172,
+      "step": 7952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3216302544085468e-07,
+      "loss": 0.0475,
+      "step": 7953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3134943356913464e-07,
+      "loss": 0.0054,
+      "step": 7954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3053725309249432e-07,
+      "loss": 0.0695,
+      "step": 7955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2972648412827982e-07,
+      "loss": 0.0221,
+      "step": 7956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2891712679363077e-07,
+      "loss": 0.0005,
+      "step": 7957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2810918120548698e-07,
+      "loss": 0.0534,
+      "step": 7958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2730264748058063e-07,
+      "loss": 0.0007,
+      "step": 7959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.264975257354429e-07,
+      "loss": 0.0996,
+      "step": 7960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.256938160863986e-07,
+      "loss": 0.0005,
+      "step": 7961
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2489151864957037e-07,
+      "loss": 0.0476,
+      "step": 7962
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2409063354087435e-07,
+      "loss": 0.0282,
+      "step": 7963
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2329116087602464e-07,
+      "loss": 0.0613,
+      "step": 7964
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.224931007705311e-07,
+      "loss": 0.0158,
+      "step": 7965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2169645333969815e-07,
+      "loss": 0.0537,
+      "step": 7966
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2090121869862814e-07,
+      "loss": 0.0031,
+      "step": 7967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2010739696221805e-07,
+      "loss": 0.0007,
+      "step": 7968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.193149882451595e-07,
+      "loss": 0.1917,
+      "step": 7969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 0.01,
+      "step": 7970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1773441032685417e-07,
+      "loss": 0.0069,
+      "step": 7971
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1694624135397024e-07,
+      "loss": 0.0574,
+      "step": 7972
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1615948585717138e-07,
+      "loss": 0.0007,
+      "step": 7973
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1537414395012667e-07,
+      "loss": 0.0148,
+      "step": 7974
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1459021574630535e-07,
+      "loss": 0.0015,
+      "step": 7975
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1380770135897122e-07,
+      "loss": 0.0491,
+      "step": 7976
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1302660090118167e-07,
+      "loss": 0.0081,
+      "step": 7977
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1224691448579305e-07,
+      "loss": 0.0658,
+      "step": 7978
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1146864222545748e-07,
+      "loss": 0.0028,
+      "step": 7979
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1069178423261838e-07,
+      "loss": 0.0226,
+      "step": 7980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0991634061952038e-07,
+      "loss": 0.0328,
+      "step": 7981
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0914231149820164e-07,
+      "loss": 0.004,
+      "step": 7982
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0836969698049158e-07,
+      "loss": 0.0604,
+      "step": 7983
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0759849717802317e-07,
+      "loss": 0.0118,
+      "step": 7984
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0682871220221945e-07,
+      "loss": 0.0409,
+      "step": 7985
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.060603421643004e-07,
+      "loss": 0.1193,
+      "step": 7986
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0529338717528157e-07,
+      "loss": 0.0062,
+      "step": 7987
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.045278473459744e-07,
+      "loss": 0.0015,
+      "step": 7988
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.037637227869871e-07,
+      "loss": 0.0027,
+      "step": 7989
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0300101360872015e-07,
+      "loss": 0.0436,
+      "step": 7990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0223971992137325e-07,
+      "loss": 0.0788,
+      "step": 7991
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.014798418349373e-07,
+      "loss": 0.001,
+      "step": 7992
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0072137945920222e-07,
+      "loss": 0.0785,
+      "step": 7993
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999643329037515e-07,
+      "loss": 0.017,
+      "step": 7994
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.992087022779665e-07,
+      "loss": 0.1695,
+      "step": 7995
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9845448769102105e-07,
+      "loss": 0.0737,
+      "step": 7996
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.977016892518857e-07,
+      "loss": 0.002,
+      "step": 7997
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9695030706932682e-07,
+      "loss": 0.0021,
+      "step": 7998
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9620034125190645e-07,
+      "loss": 0.0421,
+      "step": 7999
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9545179190798015e-07,
+      "loss": 0.0289,
+      "step": 8000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9470465914569915e-07,
+      "loss": 0.009,
+      "step": 8001
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9395894307301154e-07,
+      "loss": 0.0188,
+      "step": 8002
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.932146437976612e-07,
+      "loss": 0.0016,
+      "step": 8003
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9247176142718427e-07,
+      "loss": 0.0175,
+      "step": 8004
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.917302960689138e-07,
+      "loss": 0.0077,
+      "step": 8005
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9099024782997965e-07,
+      "loss": 0.0503,
+      "step": 8006
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.902516168173041e-07,
+      "loss": 0.044,
+      "step": 8007
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8951440313760838e-07,
+      "loss": 0.0029,
+      "step": 8008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8877860689740512e-07,
+      "loss": 0.0465,
+      "step": 8009
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.880442282030026e-07,
+      "loss": 0.0129,
+      "step": 8010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8731126716050707e-07,
+      "loss": 0.0035,
+      "step": 8011
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8657972387581713e-07,
+      "loss": 0.0108,
+      "step": 8012
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8584959845462824e-07,
+      "loss": 0.0087,
+      "step": 8013
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8512089100243158e-07,
+      "loss": 0.0053,
+      "step": 8014
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.843936016245107e-07,
+      "loss": 0.1039,
+      "step": 8015
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.836677304259471e-07,
+      "loss": 0.0242,
+      "step": 8016
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8294327751161467e-07,
+      "loss": 0.0086,
+      "step": 8017
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8222024298618524e-07,
+      "loss": 0.0057,
+      "step": 8018
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.814986269541241e-07,
+      "loss": 0.001,
+      "step": 8019
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8077842951969127e-07,
+      "loss": 0.0026,
+      "step": 8020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8005965078694233e-07,
+      "loss": 0.0015,
+      "step": 8021
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7934229085972866e-07,
+      "loss": 0.0243,
+      "step": 8022
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7862634984169513e-07,
+      "loss": 0.0037,
+      "step": 8023
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7791182783628346e-07,
+      "loss": 0.1101,
+      "step": 8024
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7719872494672775e-07,
+      "loss": 0.0491,
+      "step": 8025
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.764870412760611e-07,
+      "loss": 0.0262,
+      "step": 8026
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7577677692710682e-07,
+      "loss": 0.0752,
+      "step": 8027
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7506793200248507e-07,
+      "loss": 0.0088,
+      "step": 8028
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.743605066046128e-07,
+      "loss": 0.0076,
+      "step": 8029
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7365450083570046e-07,
+      "loss": 0.0108,
+      "step": 8030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7294991479775202e-07,
+      "loss": 0.0403,
+      "step": 8031
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.722467485925683e-07,
+      "loss": 0.0427,
+      "step": 8032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7154500232174353e-07,
+      "loss": 0.0003,
+      "step": 8033
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7084467608666778e-07,
+      "loss": 0.0262,
+      "step": 8034
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7014576998852673e-07,
+      "loss": 0.0188,
+      "step": 8035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.694482841282974e-07,
+      "loss": 0.1116,
+      "step": 8036
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6875221860675694e-07,
+      "loss": 0.0081,
+      "step": 8037
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.680575735244716e-07,
+      "loss": 0.0023,
+      "step": 8038
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6736434898180664e-07,
+      "loss": 0.0397,
+      "step": 8039
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6667254507892083e-07,
+      "loss": 0.0035,
+      "step": 8040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6598216191576643e-07,
+      "loss": 0.0698,
+      "step": 8041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6529319959209255e-07,
+      "loss": 0.0013,
+      "step": 8042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.646056582074418e-07,
+      "loss": 0.022,
+      "step": 8043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6391953786115133e-07,
+      "loss": 0.0602,
+      "step": 8044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.63234838652353e-07,
+      "loss": 0.0037,
+      "step": 8045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6255156067997325e-07,
+      "loss": 0.0388,
+      "step": 8046
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6186970404273307e-07,
+      "loss": 0.0395,
+      "step": 8047
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6118926883915032e-07,
+      "loss": 0.0384,
+      "step": 8048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6051025516753417e-07,
+      "loss": 0.065,
+      "step": 8049
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5983266312599056e-07,
+      "loss": 0.0323,
+      "step": 8050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5915649281242007e-07,
+      "loss": 0.1204,
+      "step": 8051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.584817443245168e-07,
+      "loss": 0.0043,
+      "step": 8052
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.578084177597694e-07,
+      "loss": 0.0074,
+      "step": 8053
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.571365132154612e-07,
+      "loss": 0.0762,
+      "step": 8054
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5646603078867007e-07,
+      "loss": 0.0755,
+      "step": 8055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5579697057627074e-07,
+      "loss": 0.0006,
+      "step": 8056
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5512933267492813e-07,
+      "loss": 0.0034,
+      "step": 8057
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5446311718110618e-07,
+      "loss": 0.0091,
+      "step": 8058
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.53798324191059e-07,
+      "loss": 0.0516,
+      "step": 8059
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5313495380083864e-07,
+      "loss": 0.0012,
+      "step": 8060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5247300610628957e-07,
+      "loss": 0.0058,
+      "step": 8061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.518124812030519e-07,
+      "loss": 0.0142,
+      "step": 8062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5115337918655827e-07,
+      "loss": 0.0069,
+      "step": 8063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5049570015203908e-07,
+      "loss": 0.0878,
+      "step": 8064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4983944419451613e-07,
+      "loss": 0.0257,
+      "step": 8065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4918461140880691e-07,
+      "loss": 0.0562,
+      "step": 8066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.485312018895224e-07,
+      "loss": 0.0256,
+      "step": 8067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4787921573106933e-07,
+      "loss": 0.0074,
+      "step": 8068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.47228653027649e-07,
+      "loss": 0.0922,
+      "step": 8069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4657951387325398e-07,
+      "loss": 0.0029,
+      "step": 8070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.459317983616737e-07,
+      "loss": 0.0102,
+      "step": 8071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4528550658649332e-07,
+      "loss": 0.0404,
+      "step": 8072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.44640638641087e-07,
+      "loss": 0.0035,
+      "step": 8073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4399719461863025e-07,
+      "loss": 0.0043,
+      "step": 8074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4335517461208647e-07,
+      "loss": 0.0018,
+      "step": 8075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.42714578714217e-07,
+      "loss": 0.0927,
+      "step": 8076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4207540701757781e-07,
+      "loss": 0.013,
+      "step": 8077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4143765961451505e-07,
+      "loss": 0.0379,
+      "step": 8078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4080133659717498e-07,
+      "loss": 0.0015,
+      "step": 8079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4016643805749298e-07,
+      "loss": 0.0297,
+      "step": 8080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3953296408720007e-07,
+      "loss": 0.0251,
+      "step": 8081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3890091477782198e-07,
+      "loss": 0.1237,
+      "step": 8082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3827029022068006e-07,
+      "loss": 0.033,
+      "step": 8083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.37641090506887e-07,
+      "loss": 0.025,
+      "step": 8084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3701331572735232e-07,
+      "loss": 0.0727,
+      "step": 8085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3638696597277678e-07,
+      "loss": 0.0033,
+      "step": 8086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.357620413336558e-07,
+      "loss": 0.0462,
+      "step": 8087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3513854190028263e-07,
+      "loss": 0.0969,
+      "step": 8088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3451646776273973e-07,
+      "loss": 0.0014,
+      "step": 8089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3389581901090632e-07,
+      "loss": 0.0093,
+      "step": 8090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3327659573445395e-07,
+      "loss": 0.0089,
+      "step": 8091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3265879802285107e-07,
+      "loss": 0.0037,
+      "step": 8092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3204242596535632e-07,
+      "loss": 0.1573,
+      "step": 8093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3142747965102732e-07,
+      "loss": 0.0057,
+      "step": 8094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3081395916870965e-07,
+      "loss": 0.0767,
+      "step": 8095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3020186460704908e-07,
+      "loss": 0.0025,
+      "step": 8096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2959119605447934e-07,
+      "loss": 0.0391,
+      "step": 8097
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2898195359923317e-07,
+      "loss": 0.0008,
+      "step": 8098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2837413732933457e-07,
+      "loss": 0.0208,
+      "step": 8099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2776774733260222e-07,
+      "loss": 0.0013,
+      "step": 8100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2716278369664825e-07,
+      "loss": 0.0295,
+      "step": 8101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.265592465088783e-07,
+      "loss": 0.0489,
+      "step": 8102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2595713585649482e-07,
+      "loss": 0.0324,
+      "step": 8103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.253564518264905e-07,
+      "loss": 0.0111,
+      "step": 8104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2475719450565471e-07,
+      "loss": 0.0007,
+      "step": 8105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2415936398056715e-07,
+      "loss": 0.0034,
+      "step": 8106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.235629603376054e-07,
+      "loss": 0.035,
+      "step": 8107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2296798366293826e-07,
+      "loss": 0.003,
+      "step": 8108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2237443404253037e-07,
+      "loss": 0.0051,
+      "step": 8109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2178231156213748e-07,
+      "loss": 0.0007,
+      "step": 8110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2119161630731234e-07,
+      "loss": 0.0081,
+      "step": 8111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2060234836339778e-07,
+      "loss": 0.0075,
+      "step": 8112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.200145078155346e-07,
+      "loss": 0.0728,
+      "step": 8113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.194280947486548e-07,
+      "loss": 0.0093,
+      "step": 8114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.18843109247484e-07,
+      "loss": 0.0699,
+      "step": 8115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1825955139654121e-07,
+      "loss": 0.0412,
+      "step": 8116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1767742128014125e-07,
+      "loss": 0.003,
+      "step": 8117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1709671898239128e-07,
+      "loss": 0.0464,
+      "step": 8118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1651744458719194e-07,
+      "loss": 0.0319,
+      "step": 8119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1593959817823852e-07,
+      "loss": 0.0105,
+      "step": 8120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.153631798390209e-07,
+      "loss": 0.0016,
+      "step": 8121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1478818965281912e-07,
+      "loss": 0.0077,
+      "step": 8122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1421462770270897e-07,
+      "loss": 0.0027,
+      "step": 8123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1364249407156081e-07,
+      "loss": 0.0303,
+      "step": 8124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.130717888420374e-07,
+      "loss": 0.0013,
+      "step": 8125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1250251209659502e-07,
+      "loss": 0.0004,
+      "step": 8126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1193466391748453e-07,
+      "loss": 0.0618,
+      "step": 8127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1136824438675031e-07,
+      "loss": 0.018,
+      "step": 8128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1080325358622911e-07,
+      "loss": 0.0033,
+      "step": 8129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1023969159755122e-07,
+      "loss": 0.0029,
+      "step": 8130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0967755850214257e-07,
+      "loss": 0.0654,
+      "step": 8131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0911685438122155e-07,
+      "loss": 0.113,
+      "step": 8132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.085575793157978e-07,
+      "loss": 0.013,
+      "step": 8133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.079997333866789e-07,
+      "loss": 0.0093,
+      "step": 8134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0744331667446151e-07,
+      "loss": 0.0556,
+      "step": 8135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0688832925954019e-07,
+      "loss": 0.1258,
+      "step": 8136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0633477122209856e-07,
+      "loss": 0.0017,
+      "step": 8137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0578264264211712e-07,
+      "loss": 0.0368,
+      "step": 8138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0523194359936872e-07,
+      "loss": 0.0088,
+      "step": 8139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.046826741734186e-07,
+      "loss": 0.1058,
+      "step": 8140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.041348344436277e-07,
+      "loss": 0.0071,
+      "step": 8141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0358842448914718e-07,
+      "loss": 0.012,
+      "step": 8142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0304344438892389e-07,
+      "loss": 0.0083,
+      "step": 8143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0249989422169926e-07,
+      "loss": 0.0018,
+      "step": 8144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0195777406600493e-07,
+      "loss": 0.0238,
+      "step": 8145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0141708400016714e-07,
+      "loss": 0.0554,
+      "step": 8146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0087782410230784e-07,
+      "loss": 0.005,
+      "step": 8147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0033999445034026e-07,
+      "loss": 0.0033,
+      "step": 8148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.980359512197002e-08,
+      "loss": 0.0017,
+      "step": 8149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.926862619469735e-08,
+      "loss": 0.045,
+      "step": 8150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.873508774581597e-08,
+      "loss": 0.1083,
+      "step": 8151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.820297985241311e-08,
+      "loss": 0.0025,
+      "step": 8152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.767230259136729e-08,
+      "loss": 0.002,
+      "step": 8153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.714305603935382e-08,
+      "loss": 0.0538,
+      "step": 8154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.66152402728382e-08,
+      "loss": 0.0506,
+      "step": 8155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.608885536808055e-08,
+      "loss": 0.0008,
+      "step": 8156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.556390140113337e-08,
+      "loss": 0.0285,
+      "step": 8157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.504037844784487e-08,
+      "loss": 0.0019,
+      "step": 8158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.451828658385342e-08,
+      "loss": 0.0043,
+      "step": 8159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.399762588459205e-08,
+      "loss": 0.0347,
+      "step": 8160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.347839642528721e-08,
+      "loss": 0.0016,
+      "step": 8161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.296059828095783e-08,
+      "loss": 0.0652,
+      "step": 8162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.244423152641735e-08,
+      "loss": 0.0159,
+      "step": 8163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.192929623627055e-08,
+      "loss": 0.0019,
+      "step": 8164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.141579248491684e-08,
+      "loss": 0.0642,
+      "step": 8165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.090372034654793e-08,
+      "loss": 0.078,
+      "step": 8166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.039307989515023e-08,
+      "loss": 0.0341,
+      "step": 8167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.988387120450026e-08,
+      "loss": 0.0464,
+      "step": 8168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.93760943481714e-08,
+      "loss": 0.0327,
+      "step": 8169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.88697493995272e-08,
+      "loss": 0.0318,
+      "step": 8170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.836483643172577e-08,
+      "loss": 0.0368,
+      "step": 8171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.786135551771768e-08,
+      "loss": 0.0665,
+      "step": 8172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.735930673024806e-08,
+      "loss": 0.0901,
+      "step": 8173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.685869014185222e-08,
+      "loss": 0.0071,
+      "step": 8174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.63595058248623e-08,
+      "loss": 0.1209,
+      "step": 8175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.58617538513995e-08,
+      "loss": 0.0008,
+      "step": 8176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.536543429338074e-08,
+      "loss": 0.0244,
+      "step": 8177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.487054722251531e-08,
+      "loss": 0.0059,
+      "step": 8178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.437709271030603e-08,
+      "loss": 0.0247,
+      "step": 8179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.388507082804808e-08,
+      "loss": 0.0398,
+      "step": 8180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.339448164682906e-08,
+      "loss": 0.0355,
+      "step": 8181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.290532523753115e-08,
+      "loss": 0.0096,
+      "step": 8182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.241760167083002e-08,
+      "loss": 0.033,
+      "step": 8183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.193131101719043e-08,
+      "loss": 0.0015,
+      "step": 8184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.144645334687396e-08,
+      "loss": 0.0425,
+      "step": 8185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.096302872993234e-08,
+      "loss": 0.0212,
+      "step": 8186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.048103723621415e-08,
+      "loss": 0.0338,
+      "step": 8187
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.0000478935357e-08,
+      "loss": 0.0321,
+      "step": 8188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.952135389679317e-08,
+      "loss": 0.0596,
+      "step": 8189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.904366218974723e-08,
+      "loss": 0.0065,
+      "step": 8190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.856740388323847e-08,
+      "loss": 0.1331,
+      "step": 8191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.809257904607737e-08,
+      "loss": 0.0497,
+      "step": 8192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.761918774686683e-08,
+      "loss": 0.0017,
+      "step": 8193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.714723005400327e-08,
+      "loss": 0.0017,
+      "step": 8194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.667670603567656e-08,
+      "loss": 0.0045,
+      "step": 8195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.620761575987013e-08,
+      "loss": 0.0045,
+      "step": 8196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.573995929435751e-08,
+      "loss": 0.0056,
+      "step": 8197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.52737367067069e-08,
+      "loss": 0.022,
+      "step": 8198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.480894806427886e-08,
+      "loss": 0.0093,
+      "step": 8199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.434559343422742e-08,
+      "loss": 0.0833,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.388367288350018e-08,
+      "loss": 0.0117,
+      "step": 8201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.342318647883595e-08,
+      "loss": 0.0009,
+      "step": 8202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.296413428676486e-08,
+      "loss": 0.0059,
+      "step": 8203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.250651637361384e-08,
+      "loss": 0.0095,
+      "step": 8204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.205033280549889e-08,
+      "loss": 0.0026,
+      "step": 8205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.159558364833175e-08,
+      "loss": 0.0071,
+      "step": 8206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.114226896781428e-08,
+      "loss": 0.0835,
+      "step": 8207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.069038882944412e-08,
+      "loss": 0.0862,
+      "step": 8208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.023994329850903e-08,
+      "loss": 0.0374,
+      "step": 8209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.979093244008916e-08,
+      "loss": 0.0105,
+      "step": 8210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.934335631905931e-08,
+      "loss": 0.0174,
+      "step": 8211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.889721500008662e-08,
+      "loss": 0.0003,
+      "step": 8212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.845250854763064e-08,
+      "loss": 0.0024,
+      "step": 8213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.80092370259422e-08,
+      "loss": 0.0417,
+      "step": 8214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.756740049906785e-08,
+      "loss": 0.0016,
+      "step": 8215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.71269990308443e-08,
+      "loss": 0.0446,
+      "step": 8216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.668803268490066e-08,
+      "loss": 0.0742,
+      "step": 8217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.625050152466062e-08,
+      "loss": 0.0236,
+      "step": 8218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.58144056133403e-08,
+      "loss": 0.0394,
+      "step": 8219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.537974501394705e-08,
+      "loss": 0.004,
+      "step": 8220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.494651978928179e-08,
+      "loss": 0.0352,
+      "step": 8221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.451473000193664e-08,
+      "loss": 0.003,
+      "step": 8222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.408437571429838e-08,
+      "loss": 0.0006,
+      "step": 8223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.365545698854725e-08,
+      "loss": 0.0035,
+      "step": 8224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.322797388665036e-08,
+      "loss": 0.0413,
+      "step": 8225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.280192647037498e-08,
+      "loss": 0.0134,
+      "step": 8226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.237731480127519e-08,
+      "loss": 0.0016,
+      "step": 8227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.19541389407019e-08,
+      "loss": 0.0044,
+      "step": 8228
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.153239894979402e-08,
+      "loss": 0.0056,
+      "step": 8229
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.11120948894861e-08,
+      "loss": 0.0098,
+      "step": 8230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.069322682050516e-08,
+      "loss": 0.0018,
+      "step": 8231
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.027579480337054e-08,
+      "loss": 0.0053,
+      "step": 8232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.985979889839289e-08,
+      "loss": 0.0878,
+      "step": 8233
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.944523916567524e-08,
+      "loss": 0.009,
+      "step": 8234
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.903211566511635e-08,
+      "loss": 0.004,
+      "step": 8235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.862042845640403e-08,
+      "loss": 0.03,
+      "step": 8236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.821017759901959e-08,
+      "loss": 0.0447,
+      "step": 8237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7801363152236725e-08,
+      "loss": 0.1035,
+      "step": 8238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.739398517512262e-08,
+      "loss": 0.0099,
+      "step": 8239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.698804372653466e-08,
+      "loss": 0.0972,
+      "step": 8240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.65835388651248e-08,
+      "loss": 0.0132,
+      "step": 8241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6180470649337405e-08,
+      "loss": 0.0055,
+      "step": 8242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.577883913740811e-08,
+      "loss": 0.1136,
+      "step": 8243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.537864438736606e-08,
+      "loss": 0.0027,
+      "step": 8244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.497988645703167e-08,
+      "loss": 0.0115,
+      "step": 8245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.458256540401774e-08,
+      "loss": 0.0025,
+      "step": 8246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.418668128573057e-08,
+      "loss": 0.0297,
+      "step": 8247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.379223415936885e-08,
+      "loss": 0.0024,
+      "step": 8248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.339922408192255e-08,
+      "loss": 0.0199,
+      "step": 8249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.300765111017514e-08,
+      "loss": 0.0341,
+      "step": 8250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.261751530070136e-08,
+      "loss": 0.0054,
+      "step": 8251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.222881670986835e-08,
+      "loss": 0.0281,
+      "step": 8252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.184155539383784e-08,
+      "loss": 0.0016,
+      "step": 8253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1455731408560636e-08,
+      "loss": 0.0096,
+      "step": 8254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1071344809781045e-08,
+      "loss": 0.0409,
+      "step": 8255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0688395653036846e-08,
+      "loss": 0.0142,
+      "step": 8256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.030688399365824e-08,
+      "loss": 0.0064,
+      "step": 8257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.992680988676557e-08,
+      "loss": 0.0053,
+      "step": 8258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.954817338727269e-08,
+      "loss": 0.0331,
+      "step": 8259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9170974549885844e-08,
+      "loss": 0.0007,
+      "step": 8260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.879521342910476e-08,
+      "loss": 0.0391,
+      "step": 8261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.842089007921824e-08,
+      "loss": 0.0035,
+      "step": 8262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8048004554310804e-08,
+      "loss": 0.1145,
+      "step": 8263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.767655690825712e-08,
+      "loss": 0.0044,
+      "step": 8264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.730654719472538e-08,
+      "loss": 0.0512,
+      "step": 8265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.693797546717505e-08,
+      "loss": 0.0026,
+      "step": 8266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6570841778856846e-08,
+      "loss": 0.0788,
+      "step": 8267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.620514618281613e-08,
+      "loss": 0.0069,
+      "step": 8268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5840888731890634e-08,
+      "loss": 0.0127,
+      "step": 8269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5478069478706034e-08,
+      "loss": 0.0017,
+      "step": 8270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5116688475685954e-08,
+      "loss": 0.088,
+      "step": 8271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.475674577504308e-08,
+      "loss": 0.0074,
+      "step": 8272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.439824142878024e-08,
+      "loss": 0.0014,
+      "step": 8273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.404117548869824e-08,
+      "loss": 0.0188,
+      "step": 8274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3685548006384696e-08,
+      "loss": 0.2092,
+      "step": 8275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.333135903322072e-08,
+      "loss": 0.0544,
+      "step": 8276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.297860862038317e-08,
+      "loss": 0.0318,
+      "step": 8277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.262729681883571e-08,
+      "loss": 0.0348,
+      "step": 8278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2277423679337736e-08,
+      "loss": 0.0272,
+      "step": 8279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1928989252438825e-08,
+      "loss": 0.0002,
+      "step": 8280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1581993588482025e-08,
+      "loss": 0.0018,
+      "step": 8281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.123643673760169e-08,
+      "loss": 0.047,
+      "step": 8282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.089231874972455e-08,
+      "loss": 0.0002,
+      "step": 8283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.054963967456971e-08,
+      "loss": 0.0118,
+      "step": 8284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0208399561648684e-08,
+      "loss": 0.0034,
+      "step": 8285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.986859846026425e-08,
+      "loss": 0.0648,
+      "step": 8286
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953023641951159e-08,
+      "loss": 0.0823,
+      "step": 8287
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9193313488277155e-08,
+      "loss": 0.0124,
+      "step": 8288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.885782971524088e-08,
+      "loss": 0.0172,
+      "step": 8289
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8523785148875115e-08,
+      "loss": 0.007,
+      "step": 8290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.819117983744125e-08,
+      "loss": 0.0725,
+      "step": 8291
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7860013828996403e-08,
+      "loss": 0.051,
+      "step": 8292
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.753028717138785e-08,
+      "loss": 0.007,
+      "step": 8293
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.720199991225415e-08,
+      "loss": 0.0054,
+      "step": 8294
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6875152099028475e-08,
+      "loss": 0.0034,
+      "step": 8295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6549743778934166e-08,
+      "loss": 0.0004,
+      "step": 8296
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.622577499898583e-08,
+      "loss": 0.043,
+      "step": 8297
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.590324580599158e-08,
+      "loss": 0.0016,
+      "step": 8298
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5582156246551926e-08,
+      "loss": 0.0372,
+      "step": 8299
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5262506367058635e-08,
+      "loss": 0.1214,
+      "step": 8300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.494429621369366e-08,
+      "loss": 0.0089,
+      "step": 8301
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.462752583243356e-08,
+      "loss": 0.0727,
+      "step": 8302
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.431219526904728e-08,
+      "loss": 0.0014,
+      "step": 8303
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.399830456909281e-08,
+      "loss": 0.0939,
+      "step": 8304
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.368585377792166e-08,
+      "loss": 0.014,
+      "step": 8305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3374842940678834e-08,
+      "loss": 0.0873,
+      "step": 8306
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3065272102299484e-08,
+      "loss": 0.0059,
+      "step": 8307
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2757141307510066e-08,
+      "loss": 0.0066,
+      "step": 8308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.245045060083052e-08,
+      "loss": 0.0466,
+      "step": 8309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.214520002657095e-08,
+      "loss": 0.0083,
+      "step": 8310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.18413896288372e-08,
+      "loss": 0.0022,
+      "step": 8311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1539019451521935e-08,
+      "loss": 0.0063,
+      "step": 8312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.123808953831353e-08,
+      "loss": 0.0102,
+      "step": 8313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.093859993269055e-08,
+      "loss": 0.0064,
+      "step": 8314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.064055067792504e-08,
+      "loss": 0.0144,
+      "step": 8315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.034394181707812e-08,
+      "loss": 0.0485,
+      "step": 8316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.004877339300549e-08,
+      "loss": 0.0105,
+      "step": 8317
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9755045448351948e-08,
+      "loss": 0.049,
+      "step": 8318
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.946275802555909e-08,
+      "loss": 0.0089,
+      "step": 8319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.917191116685425e-08,
+      "loss": 0.0114,
+      "step": 8320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.888250491426159e-08,
+      "loss": 0.0022,
+      "step": 8321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8594539309593217e-08,
+      "loss": 0.0393,
+      "step": 8322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8308014394456963e-08,
+      "loss": 0.0982,
+      "step": 8323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.802293021024971e-08,
+      "loss": 0.0423,
+      "step": 8324
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7739286798160737e-08,
+      "loss": 0.0017,
+      "step": 8325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.74570841991717e-08,
+      "loss": 0.0074,
+      "step": 8326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.717632245405555e-08,
+      "loss": 0.0811,
+      "step": 8327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6897001603377603e-08,
+      "loss": 0.0014,
+      "step": 8328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6619121687495586e-08,
+      "loss": 0.0187,
+      "step": 8329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.634268274655627e-08,
+      "loss": 0.0491,
+      "step": 8330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.606768482050215e-08,
+      "loss": 0.0024,
+      "step": 8331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5794127949063664e-08,
+      "loss": 0.0005,
+      "step": 8332
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5522012171765863e-08,
+      "loss": 0.0008,
+      "step": 8333
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.525133752792508e-08,
+      "loss": 0.0227,
+      "step": 8334
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4982104056647805e-08,
+      "loss": 0.0101,
+      "step": 8335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4714311796834035e-08,
+      "loss": 0.0344,
+      "step": 8336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.444796078717504e-08,
+      "loss": 0.1411,
+      "step": 8337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4183051066153374e-08,
+      "loss": 0.0182,
+      "step": 8338
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.391958267204397e-08,
+      "loss": 0.0327,
+      "step": 8339
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3657555642913055e-08,
+      "loss": 0.0049,
+      "step": 8340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3396970016619225e-08,
+      "loss": 0.0793,
+      "step": 8341
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3137825830813476e-08,
+      "loss": 0.0151,
+      "step": 8342
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.288012312293586e-08,
+      "loss": 0.0069,
+      "step": 8343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2623861930219927e-08,
+      "loss": 0.0039,
+      "step": 8344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2369042289690503e-08,
+      "loss": 0.032,
+      "step": 8345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.211566423816591e-08,
+      "loss": 0.0186,
+      "step": 8346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1863727812254653e-08,
+      "loss": 0.0792,
+      "step": 8347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1613233048355385e-08,
+      "loss": 0.0115,
+      "step": 8348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.136417998266138e-08,
+      "loss": 0.0036,
+      "step": 8349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1116568651156076e-08,
+      "loss": 0.0087,
+      "step": 8350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.08703990896153e-08,
+      "loss": 0.0514,
+      "step": 8351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0625671333606156e-08,
+      "loss": 0.006,
+      "step": 8352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0382385418485916e-08,
+      "loss": 0.0243,
+      "step": 8353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0140541379406463e-08,
+      "loss": 0.0126,
+      "step": 8354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9900139251309846e-08,
+      "loss": 0.018,
+      "step": 8355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9661179068928283e-08,
+      "loss": 0.1083,
+      "step": 8356
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9423660866789706e-08,
+      "loss": 0.011,
+      "step": 8357
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9187584679210004e-08,
+      "loss": 0.0138,
+      "step": 8358
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8952950540298552e-08,
+      "loss": 0.0464,
+      "step": 8359
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.871975848395491e-08,
+      "loss": 0.0023,
+      "step": 8360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8488008543871007e-08,
+      "loss": 0.1314,
+      "step": 8361
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.825770075353117e-08,
+      "loss": 0.0855,
+      "step": 8362
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.802883514621101e-08,
+      "loss": 0.0116,
+      "step": 8363
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7801411754977404e-08,
+      "loss": 0.0026,
+      "step": 8364
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7575430612688516e-08,
+      "loss": 0.0945,
+      "step": 8365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7350891751993782e-08,
+      "loss": 0.0061,
+      "step": 8366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.712779520533725e-08,
+      "loss": 0.0038,
+      "step": 8367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6906141004950915e-08,
+      "loss": 0.0837,
+      "step": 8368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6685929182860274e-08,
+      "loss": 0.0033,
+      "step": 8369
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6467159770880982e-08,
+      "loss": 0.1012,
+      "step": 8370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.624983280062331e-08,
+      "loss": 0.1117,
+      "step": 8371
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.603394830348437e-08,
+      "loss": 0.0064,
+      "step": 8372
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5819506310658095e-08,
+      "loss": 0.0012,
+      "step": 8373
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.560650685312637e-08,
+      "loss": 0.0249,
+      "step": 8374
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5394949961664573e-08,
+      "loss": 0.0144,
+      "step": 8375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.518483566683826e-08,
+      "loss": 0.0103,
+      "step": 8376
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4976163999004257e-08,
+      "loss": 0.0211,
+      "step": 8377
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4768934988313998e-08,
+      "loss": 0.0173,
+      "step": 8378
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4563148664706872e-08,
+      "loss": 0.0037,
+      "step": 8379
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4358805057914649e-08,
+      "loss": 0.1226,
+      "step": 8380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.415590419746371e-08,
+      "loss": 0.0125,
+      "step": 8381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3954446112667275e-08,
+      "loss": 0.0006,
+      "step": 8382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.375443083263428e-08,
+      "loss": 0.0255,
+      "step": 8383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.355585838626161e-08,
+      "loss": 0.0293,
+      "step": 8384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3358728802241872e-08,
+      "loss": 0.1353,
+      "step": 8385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3163042109053393e-08,
+      "loss": 0.0099,
+      "step": 8386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2968798334972444e-08,
+      "loss": 0.0247,
+      "step": 8387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2775997508061023e-08,
+      "loss": 0.1435,
+      "step": 8388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2584639656177956e-08,
+      "loss": 0.0058,
+      "step": 8389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2394724806970015e-08,
+      "loss": 0.0018,
+      "step": 8390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2206252987876366e-08,
+      "loss": 0.0031,
+      "step": 8391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2019224226127446e-08,
+      "loss": 0.1587,
+      "step": 8392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1833638548746085e-08,
+      "loss": 0.0544,
+      "step": 8393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1649495982546388e-08,
+      "loss": 0.0011,
+      "step": 8394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.146679655413374e-08,
+      "loss": 0.0148,
+      "step": 8395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1285540289903696e-08,
+      "loss": 0.0014,
+      "step": 8396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1105727216045303e-08,
+      "loss": 0.0191,
+      "step": 8397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0927357358537783e-08,
+      "loss": 0.048,
+      "step": 8398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.075043074315385e-08,
+      "loss": 0.0186,
+      "step": 8399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0574947395455281e-08,
+      "loss": 0.0129,
+      "step": 8400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.040090734079624e-08,
+      "loss": 0.001,
+      "step": 8401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0228310604322166e-08,
+      "loss": 0.0051,
+      "step": 8402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0057157210970892e-08,
+      "loss": 0.0006,
+      "step": 8403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.887447185470412e-09,
+      "loss": 0.0126,
+      "step": 8404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.719180552341113e-09,
+      "loss": 0.0028,
+      "step": 8405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.552357335893547e-09,
+      "loss": 0.0719,
+      "step": 8406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.38697756023288e-09,
+      "loss": 0.0068,
+      "step": 8407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.223041249251107e-09,
+      "loss": 0.0878,
+      "step": 8408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.060548426635951e-09,
+      "loss": 0.0104,
+      "step": 8409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.899499115864185e-09,
+      "loss": 0.0135,
+      "step": 8410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.739893340202754e-09,
+      "loss": 0.0101,
+      "step": 8411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.58173112271543e-09,
+      "loss": 0.062,
+      "step": 8412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.425012486250606e-09,
+      "loss": 0.0951,
+      "step": 8413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.269737453452386e-09,
+      "loss": 0.0093,
+      "step": 8414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.11590604675505e-09,
+      "loss": 0.0006,
+      "step": 8415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.963518288385264e-09,
+      "loss": 0.0652,
+      "step": 8416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.812574200359857e-09,
+      "loss": 0.0004,
+      "step": 8417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.663073804488053e-09,
+      "loss": 0.0049,
+      "step": 8418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.515017122369239e-09,
+      "loss": 0.1067,
+      "step": 8419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.368404175394084e-09,
+      "loss": 0.0009,
+      "step": 8420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.223234984747862e-09,
+      "loss": 0.0752,
+      "step": 8421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.0795095714026874e-09,
+      "loss": 0.0748,
+      "step": 8422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.937227956126391e-09,
+      "loss": 0.0035,
+      "step": 8423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.796390159473643e-09,
+      "loss": 0.0014,
+      "step": 8424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.65699620179594e-09,
+      "loss": 0.0145,
+      "step": 8425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.5190461032305085e-09,
+      "loss": 0.0124,
+      "step": 8426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.382539883710292e-09,
+      "loss": 0.1122,
+      "step": 8427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.2474775629572934e-09,
+      "loss": 0.0104,
+      "step": 8428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.113859160485902e-09,
+      "loss": 0.0007,
+      "step": 8429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.981684695600676e-09,
+      "loss": 0.0276,
+      "step": 8430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.850954187399671e-09,
+      "loss": 0.0372,
+      "step": 8431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.721667654771112e-09,
+      "loss": 0.033,
+      "step": 8432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.593825116394502e-09,
+      "loss": 0.001,
+      "step": 8433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.467426590739511e-09,
+      "loss": 0.0043,
+      "step": 8434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.342472096070417e-09,
+      "loss": 0.0014,
+      "step": 8435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.218961650439447e-09,
+      "loss": 0.0012,
+      "step": 8436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.096895271692326e-09,
+      "loss": 0.0775,
+      "step": 8437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.976272977464947e-09,
+      "loss": 0.0189,
+      "step": 8438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8570947851855895e-09,
+      "loss": 0.0381,
+      "step": 8439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.739360712072705e-09,
+      "loss": 0.0234,
+      "step": 8440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6230707751371284e-09,
+      "loss": 0.0207,
+      "step": 8441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.508224991180976e-09,
+      "loss": 0.0063,
+      "step": 8442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3948233767976415e-09,
+      "loss": 0.0789,
+      "step": 8443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.282865948370685e-09,
+      "loss": 0.0035,
+      "step": 8444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.172352722077167e-09,
+      "loss": 0.0032,
+      "step": 8445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.063283713883203e-09,
+      "loss": 0.0577,
+      "step": 8446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.955658939547302e-09,
+      "loss": 0.0149,
+      "step": 8447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.849478414619246e-09,
+      "loss": 0.0032,
+      "step": 8448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.744742154441206e-09,
+      "loss": 0.0086,
+      "step": 8449
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.641450174145522e-09,
+      "loss": 0.0006,
+      "step": 8450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.539602488654703e-09,
+      "loss": 0.0438,
+      "step": 8451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.439199112685865e-09,
+      "loss": 0.0029,
+      "step": 8452
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3402400607429607e-09,
+      "loss": 0.0063,
+      "step": 8453
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.242725347126774e-09,
+      "loss": 0.1,
+      "step": 8454
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.146654985924924e-09,
+      "loss": 0.0839,
+      "step": 8455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0520289910174194e-09,
+      "loss": 0.0018,
+      "step": 8456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9588473760766566e-09,
+      "loss": 0.004,
+      "step": 8457
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8671101545652004e-09,
+      "loss": 0.0139,
+      "step": 8458
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7768173397380026e-09,
+      "loss": 0.0362,
+      "step": 8459
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6879689446412947e-09,
+      "loss": 0.0017,
+      "step": 8460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.600564982110365e-09,
+      "loss": 0.1322,
+      "step": 8461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5146054647751105e-09,
+      "loss": 0.0119,
+      "step": 8462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.430090405054486e-09,
+      "loss": 0.0046,
+      "step": 8463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.347019815158724e-09,
+      "loss": 0.008,
+      "step": 8464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2653937070915564e-09,
+      "loss": 0.0152,
+      "step": 8465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.185212092645772e-09,
+      "loss": 0.001,
+      "step": 8466
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1064749834054376e-09,
+      "loss": 0.0261,
+      "step": 8467
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.029182390748119e-09,
+      "loss": 0.0138,
+      "step": 8468
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.953334325839329e-09,
+      "loss": 0.0109,
+      "step": 8469
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8789307996391893e-09,
+      "loss": 0.0023,
+      "step": 8470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8059718228979894e-09,
+      "loss": 0.0014,
+      "step": 8471
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7344574061561869e-09,
+      "loss": 0.0042,
+      "step": 8472
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6643875597466275e-09,
+      "loss": 0.0178,
+      "step": 8473
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5957622937923244e-09,
+      "loss": 0.0009,
+      "step": 8474
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5285816182097902e-09,
+      "loss": 0.0541,
+      "step": 8475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.462845542704594e-09,
+      "loss": 0.0034,
+      "step": 8476
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.398554076773584e-09,
+      "loss": 0.0021,
+      "step": 8477
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3357072297082164e-09,
+      "loss": 0.0035,
+      "step": 8478
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2743050105856747e-09,
+      "loss": 0.0016,
+      "step": 8479
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2143474282799716e-09,
+      "loss": 0.0083,
+      "step": 8480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1558344914530673e-09,
+      "loss": 0.0025,
+      "step": 8481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0987662085581996e-09,
+      "loss": 0.0008,
+      "step": 8482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0431425878421053e-09,
+      "loss": 0.0007,
+      "step": 8483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.889636373405787e-10,
+      "loss": 0.0272,
+      "step": 8484
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.362293648818022e-10,
+      "loss": 0.0179,
+      "step": 8485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.849397780841262e-10,
+      "loss": 0.018,
+      "step": 8486
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.350948843593998e-10,
+      "loss": 0.004,
+      "step": 8487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.866946909085294e-10,
+      "loss": 0.0024,
+      "step": 8488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.397392047236995e-10,
+      "loss": 0.0828,
+      "step": 8489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.942284325905934e-10,
+      "loss": 0.0035,
+      "step": 8490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.50162381083952e-10,
+      "loss": 0.0475,
+      "step": 8491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.075410565697937e-10,
+      "loss": 0.0184,
+      "step": 8492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.663644652087463e-10,
+      "loss": 0.0419,
+      "step": 8493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.266326129471644e-10,
+      "loss": 0.0272,
+      "step": 8494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.883455055260111e-10,
+      "loss": 0.0049,
+      "step": 8495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5150314847863806e-10,
+      "loss": 0.0284,
+      "step": 8496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1610554712745444e-10,
+      "loss": 0.0755,
+      "step": 8497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8215270658725767e-10,
+      "loss": 0.0403,
+      "step": 8498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4964463176190286e-10,
+      "loss": 0.088,
+      "step": 8499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.185813273498539e-10,
+      "loss": 0.0012,
+      "step": 8500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8896279783863224e-10,
+      "loss": 0.0007,
+      "step": 8501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.607890475070374e-10,
+      "loss": 0.1101,
+      "step": 8502
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3406008042625714e-10,
+      "loss": 0.0009,
+      "step": 8503
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.087759004587575e-10,
+      "loss": 0.0593,
+      "step": 8504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8493651125606195e-10,
+      "loss": 0.0047,
+      "step": 8505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6254191626430272e-10,
+      "loss": 0.0096,
+      "step": 8506
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4159211871866974e-10,
+      "loss": 0.0347,
+      "step": 8507
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2208712164452076e-10,
+      "loss": 0.0059,
+      "step": 8508
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0402692786182223e-10,
+      "loss": 0.0785,
+      "step": 8509
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.741153997959828e-11,
+      "loss": 0.0566,
+      "step": 8510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.224096039704088e-11,
+      "loss": 0.0242,
+      "step": 8511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.851519130795069e-11,
+      "loss": 0.0684,
+      "step": 8512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.623423469407584e-11,
+      "loss": 0.0624,
+      "step": 8513
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5398092330662935e-11,
+      "loss": 0.0256,
+      "step": 8514
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6006765783126443e-11,
+      "loss": 0.0105,
+      "step": 8515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8060256408158892e-11,
+      "loss": 0.0116,
+      "step": 8516
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.15585653537309e-11,
+      "loss": 0.0035,
+      "step": 8517
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.501693559091138e-12,
+      "loss": 0.0148,
+      "step": 8518
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8896417558765823e-12,
+      "loss": 0.053,
+      "step": 8519
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.224104647818308e-13,
+      "loss": 0.0132,
+      "step": 8520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.0087,
+      "step": 8521
+    },
+    {
+      "epoch": 1.0,
+      "step": 8521,
+      "total_flos": 8.13332388007379e+17,
+      "train_loss": 0.05783833727231607,
+      "train_runtime": 73514.0326,
+      "train_samples_per_second": 1.855,
+      "train_steps_per_second": 0.116
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 8521,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 40000,
+  "total_flos": 8.13332388007379e+17,
+  "trial_name": null,
+  "trial_params": null
+}