diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,60019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.1,
+  "eval_steps": 500,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999000000000001e-05,
+      "loss": 2.1193,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.998e-05,
+      "loss": 3.2013,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.997e-05,
+      "loss": 3.0146,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.996000000000001e-05,
+      "loss": 1.4758,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.995e-05,
+      "loss": 1.1104,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.994e-05,
+      "loss": 1.5601,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.993e-05,
+      "loss": 2.6162,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.992e-05,
+      "loss": 2.182,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.991e-05,
+      "loss": 1.7382,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99e-05,
+      "loss": 1.8109,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.989000000000001e-05,
+      "loss": 1.3968,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.988000000000001e-05,
+      "loss": 2.1022,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.987e-05,
+      "loss": 2.2954,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.986000000000001e-05,
+      "loss": 0.9789,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.985000000000001e-05,
+      "loss": 0.6379,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.984e-05,
+      "loss": 0.7254,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.983e-05,
+      "loss": 0.2922,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.982e-05,
+      "loss": 0.8353,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.981000000000001e-05,
+      "loss": 1.3944,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.98e-05,
+      "loss": 1.487,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.979e-05,
+      "loss": 1.4506,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.978000000000001e-05,
+      "loss": 0.9054,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.977e-05,
+      "loss": 1.3134,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.976000000000001e-05,
+      "loss": 1.6775,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.975000000000001e-05,
+      "loss": 1.6893,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.974e-05,
+      "loss": 1.5867,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.973e-05,
+      "loss": 0.5063,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.972e-05,
+      "loss": 0.1098,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.971e-05,
+      "loss": 0.4943,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.970000000000001e-05,
+      "loss": 0.284,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.969e-05,
+      "loss": 1.3894,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.968000000000001e-05,
+      "loss": 1.7604,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.967000000000001e-05,
+      "loss": 1.4693,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.966e-05,
+      "loss": 1.5,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.965000000000001e-05,
+      "loss": 1.5201,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.964e-05,
+      "loss": 2.0646,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.963e-05,
+      "loss": 1.857,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.962e-05,
+      "loss": 1.4568,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.961e-05,
+      "loss": 0.6173,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.960000000000001e-05,
+      "loss": 0.1817,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.959e-05,
+      "loss": 0.2202,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.958e-05,
+      "loss": 0.1072,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.957000000000001e-05,
+      "loss": 1.3942,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.956e-05,
+      "loss": 1.6437,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.955000000000001e-05,
+      "loss": 1.6175,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.954e-05,
+      "loss": 1.4699,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.953e-05,
+      "loss": 1.2554,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.952e-05,
+      "loss": 1.0615,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.951e-05,
+      "loss": 1.1844,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.95e-05,
+      "loss": 1.875,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.949000000000001e-05,
+      "loss": 1.1984,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.948e-05,
+      "loss": 0.5217,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.947000000000001e-05,
+      "loss": 0.2438,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.946000000000001e-05,
+      "loss": 0.1652,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.945e-05,
+      "loss": 0.237,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.944e-05,
+      "loss": 1.243,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.943e-05,
+      "loss": 1.2991,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.942000000000001e-05,
+      "loss": 1.5517,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.941e-05,
+      "loss": 1.1408,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.94e-05,
+      "loss": 0.9738,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.939000000000001e-05,
+      "loss": 1.3895,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.938e-05,
+      "loss": 2.1746,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.937e-05,
+      "loss": 1.7696,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.936000000000001e-05,
+      "loss": 0.5912,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.935000000000002e-05,
+      "loss": 0.0821,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.934e-05,
+      "loss": 0.6918,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.933e-05,
+      "loss": 0.2113,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.932e-05,
+      "loss": 0.9191,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.931000000000001e-05,
+      "loss": 0.8967,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.93e-05,
+      "loss": 1.2833,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.929e-05,
+      "loss": 1.7788,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.928000000000001e-05,
+      "loss": 1.2299,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.927e-05,
+      "loss": 1.2273,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.926000000000001e-05,
+      "loss": 1.2456,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.925000000000001e-05,
+      "loss": 0.8398,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.924e-05,
+      "loss": 0.9798,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.923e-05,
+      "loss": 0.3513,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.922e-05,
+      "loss": 0.2345,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.921000000000001e-05,
+      "loss": 0.5712,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.92e-05,
+      "loss": 1.4842,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.919e-05,
+      "loss": 1.2857,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.918000000000001e-05,
+      "loss": 1.7423,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.917e-05,
+      "loss": 1.1566,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.916e-05,
+      "loss": 1.3668,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.915000000000001e-05,
+      "loss": 1.42,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.914e-05,
+      "loss": 1.4063,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.913e-05,
+      "loss": 1.4537,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.912e-05,
+      "loss": 1.9227,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.911e-05,
+      "loss": 0.2538,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.910000000000001e-05,
+      "loss": 0.1372,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.909e-05,
+      "loss": 0.2406,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.908000000000001e-05,
+      "loss": 0.6183,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.907000000000001e-05,
+      "loss": 1.091,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.906e-05,
+      "loss": 1.3006,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.905000000000001e-05,
+      "loss": 1.1319,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.904e-05,
+      "loss": 1.3151,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.903e-05,
+      "loss": 1.5663,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.902e-05,
+      "loss": 1.2496,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.901e-05,
+      "loss": 1.1157,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 1.1731,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.899e-05,
+      "loss": 1.0314,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.898e-05,
+      "loss": 0.5713,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.897000000000001e-05,
+      "loss": 0.1841,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.896000000000001e-05,
+      "loss": 0.2561,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.895e-05,
+      "loss": 1.3964,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.894e-05,
+      "loss": 1.298,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.893e-05,
+      "loss": 1.516,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.892e-05,
+      "loss": 1.4896,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.891e-05,
+      "loss": 1.0191,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.89e-05,
+      "loss": 1.4864,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.889000000000001e-05,
+      "loss": 1.4024,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.888e-05,
+      "loss": 1.0597,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.887000000000001e-05,
+      "loss": 2.0742,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.886000000000001e-05,
+      "loss": 0.8925,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.885e-05,
+      "loss": 0.2972,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.884e-05,
+      "loss": 0.4913,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.883e-05,
+      "loss": 0.5849,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.882e-05,
+      "loss": 0.9493,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.881e-05,
+      "loss": 1.1613,
+      "step": 119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.88e-05,
+      "loss": 1.0378,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.879000000000001e-05,
+      "loss": 1.5454,
+      "step": 121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.878e-05,
+      "loss": 1.2575,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.877e-05,
+      "loss": 1.4233,
+      "step": 123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.876000000000001e-05,
+      "loss": 1.4563,
+      "step": 124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.875000000000002e-05,
+      "loss": 2.1416,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.874e-05,
+      "loss": 1.306,
+      "step": 126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.873e-05,
+      "loss": 0.4739,
+      "step": 127
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.872e-05,
+      "loss": 0.042,
+      "step": 128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.871000000000001e-05,
+      "loss": 0.1388,
+      "step": 129
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.87e-05,
+      "loss": 0.2936,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.869e-05,
+      "loss": 0.8114,
+      "step": 131
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.868000000000001e-05,
+      "loss": 1.269,
+      "step": 132
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.867e-05,
+      "loss": 1.3061,
+      "step": 133
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.866000000000001e-05,
+      "loss": 1.154,
+      "step": 134
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.865000000000001e-05,
+      "loss": 1.2021,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.864e-05,
+      "loss": 1.6457,
+      "step": 136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.863e-05,
+      "loss": 0.7804,
+      "step": 137
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.862e-05,
+      "loss": 2.0935,
+      "step": 138
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.861e-05,
+      "loss": 0.8456,
+      "step": 139
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.86e-05,
+      "loss": 0.3428,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.859e-05,
+      "loss": 0.7731,
+      "step": 141
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.858000000000001e-05,
+      "loss": 0.0159,
+      "step": 142
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.857000000000001e-05,
+      "loss": 0.1826,
+      "step": 143
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.856e-05,
+      "loss": 0.8708,
+      "step": 144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.855000000000001e-05,
+      "loss": 0.8791,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.854000000000002e-05,
+      "loss": 1.1211,
+      "step": 146
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.853e-05,
+      "loss": 0.9997,
+      "step": 147
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.852e-05,
+      "loss": 1.0586,
+      "step": 148
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.851e-05,
+      "loss": 0.8024,
+      "step": 149
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.850000000000001e-05,
+      "loss": 1.6331,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.849e-05,
+      "loss": 1.3844,
+      "step": 151
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.848e-05,
+      "loss": 0.5534,
+      "step": 152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.847000000000001e-05,
+      "loss": 0.4745,
+      "step": 153
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.846e-05,
+      "loss": 0.0497,
+      "step": 154
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.845000000000001e-05,
+      "loss": 0.1856,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.844000000000001e-05,
+      "loss": 0.911,
+      "step": 156
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.843e-05,
+      "loss": 1.0752,
+      "step": 157
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.842e-05,
+      "loss": 1.3982,
+      "step": 158
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.841e-05,
+      "loss": 1.5982,
+      "step": 159
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.84e-05,
+      "loss": 1.8059,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.839e-05,
+      "loss": 1.023,
+      "step": 161
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.838e-05,
+      "loss": 1.8905,
+      "step": 162
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.837000000000001e-05,
+      "loss": 1.5294,
+      "step": 163
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.836000000000001e-05,
+      "loss": 0.8915,
+      "step": 164
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.835e-05,
+      "loss": 0.1469,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.834000000000001e-05,
+      "loss": 0.0534,
+      "step": 166
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.833e-05,
+      "loss": 0.1833,
+      "step": 167
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.832000000000001e-05,
+      "loss": 0.5727,
+      "step": 168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.831e-05,
+      "loss": 1.3217,
+      "step": 169
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.83e-05,
+      "loss": 1.492,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.829000000000001e-05,
+      "loss": 1.1367,
+      "step": 171
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.828e-05,
+      "loss": 1.4649,
+      "step": 172
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.827e-05,
+      "loss": 0.9041,
+      "step": 173
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.826000000000001e-05,
+      "loss": 1.1555,
+      "step": 174
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.825e-05,
+      "loss": 1.8606,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.824000000000001e-05,
+      "loss": 1.073,
+      "step": 176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.823e-05,
+      "loss": 0.0549,
+      "step": 177
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.822e-05,
+      "loss": 0.1333,
+      "step": 178
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.821e-05,
+      "loss": 0.5884,
+      "step": 179
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.82e-05,
+      "loss": 0.3034,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.819000000000001e-05,
+      "loss": 0.8181,
+      "step": 181
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.818000000000001e-05,
+      "loss": 1.0309,
+      "step": 182
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.817e-05,
+      "loss": 1.5412,
+      "step": 183
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.816000000000001e-05,
+      "loss": 1.0065,
+      "step": 184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.815000000000001e-05,
+      "loss": 1.7842,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.814e-05,
+      "loss": 1.3414,
+      "step": 186
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.813e-05,
+      "loss": 1.812,
+      "step": 187
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.812e-05,
+      "loss": 1.8811,
+      "step": 188
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.811000000000001e-05,
+      "loss": 0.9683,
+      "step": 189
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.81e-05,
+      "loss": 0.2767,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.809e-05,
+      "loss": 0.3057,
+      "step": 191
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.808000000000001e-05,
+      "loss": 0.3095,
+      "step": 192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.807e-05,
+      "loss": 0.7317,
+      "step": 193
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.806e-05,
+      "loss": 1.1017,
+      "step": 194
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.805000000000001e-05,
+      "loss": 1.6427,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.804e-05,
+      "loss": 1.0045,
+      "step": 196
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.803e-05,
+      "loss": 1.4905,
+      "step": 197
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.802e-05,
+      "loss": 1.2139,
+      "step": 198
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.801e-05,
+      "loss": 1.5089,
+      "step": 199
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.8e-05,
+      "loss": 0.5876,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.799e-05,
+      "loss": 1.2493,
+      "step": 201
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.798000000000001e-05,
+      "loss": 0.3105,
+      "step": 202
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.797000000000001e-05,
+      "loss": 0.1724,
+      "step": 203
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.796e-05,
+      "loss": 0.1943,
+      "step": 204
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.795000000000001e-05,
+      "loss": 0.2591,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.794000000000001e-05,
+      "loss": 0.4895,
+      "step": 206
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.793e-05,
+      "loss": 1.026,
+      "step": 207
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.792e-05,
+      "loss": 1.3034,
+      "step": 208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.791e-05,
+      "loss": 1.4546,
+      "step": 209
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.790000000000001e-05,
+      "loss": 1.5558,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.789e-05,
+      "loss": 1.2231,
+      "step": 211
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.788e-05,
+      "loss": 1.1751,
+      "step": 212
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.787000000000001e-05,
+      "loss": 1.4498,
+      "step": 213
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.786e-05,
+      "loss": 0.704,
+      "step": 214
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.785e-05,
+      "loss": 0.2319,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.784000000000001e-05,
+      "loss": 0.3372,
+      "step": 216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.783e-05,
+      "loss": 0.2702,
+      "step": 217
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.782e-05,
+      "loss": 0.4243,
+      "step": 218
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.781e-05,
+      "loss": 0.9223,
+      "step": 219
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.78e-05,
+      "loss": 1.3385,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.779e-05,
+      "loss": 1.0706,
+      "step": 221
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.778e-05,
+      "loss": 1.57,
+      "step": 222
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.777000000000001e-05,
+      "loss": 1.5087,
+      "step": 223
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.776000000000001e-05,
+      "loss": 1.867,
+      "step": 224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.775e-05,
+      "loss": 1.3028,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.774000000000001e-05,
+      "loss": 1.5561,
+      "step": 226
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.773e-05,
+      "loss": 0.5387,
+      "step": 227
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.772e-05,
+      "loss": 0.0324,
+      "step": 228
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.771e-05,
+      "loss": 0.4853,
+      "step": 229
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.77e-05,
+      "loss": 0.5322,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.769000000000001e-05,
+      "loss": 0.9069,
+      "step": 231
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.768e-05,
+      "loss": 1.6229,
+      "step": 232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.767e-05,
+      "loss": 1.7144,
+      "step": 233
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.766000000000001e-05,
+      "loss": 1.3527,
+      "step": 234
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.765e-05,
+      "loss": 1.2389,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.764000000000001e-05,
+      "loss": 1.3891,
+      "step": 236
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.763e-05,
+      "loss": 0.7929,
+      "step": 237
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.762e-05,
+      "loss": 1.3078,
+      "step": 238
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.761e-05,
+      "loss": 1.2669,
+      "step": 239
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.76e-05,
+      "loss": 0.4716,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.759e-05,
+      "loss": 0.2557,
+      "step": 241
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.758000000000001e-05,
+      "loss": 0.1365,
+      "step": 242
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.757e-05,
+      "loss": 0.6165,
+      "step": 243
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.756000000000001e-05,
+      "loss": 0.6399,
+      "step": 244
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.755000000000001e-05,
+      "loss": 1.3119,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.754e-05,
+      "loss": 1.1163,
+      "step": 246
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.753e-05,
+      "loss": 1.0424,
+      "step": 247
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.752e-05,
+      "loss": 1.3402,
+      "step": 248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.751e-05,
+      "loss": 1.2477,
+      "step": 249
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.75e-05,
+      "loss": 1.591,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.749e-05,
+      "loss": 1.2483,
+      "step": 251
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.748000000000001e-05,
+      "loss": 0.0763,
+      "step": 252
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.747e-05,
+      "loss": 0.265,
+      "step": 253
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.746e-05,
+      "loss": 0.1752,
+      "step": 254
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.745000000000001e-05,
+      "loss": 0.546,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.744000000000002e-05,
+      "loss": 0.7983,
+      "step": 256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.743000000000001e-05,
+      "loss": 1.1896,
+      "step": 257
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.742e-05,
+      "loss": 1.2037,
+      "step": 258
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.741e-05,
+      "loss": 1.3669,
+      "step": 259
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.74e-05,
+      "loss": 1.2724,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.739e-05,
+      "loss": 1.3849,
+      "step": 261
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.738e-05,
+      "loss": 1.4517,
+      "step": 262
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.737000000000001e-05,
+      "loss": 1.3578,
+      "step": 263
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.736e-05,
+      "loss": 0.549,
+      "step": 264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.735000000000001e-05,
+      "loss": 0.437,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.734000000000001e-05,
+      "loss": 0.051,
+      "step": 266
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.733e-05,
+      "loss": 0.363,
+      "step": 267
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.732e-05,
+      "loss": 0.8294,
+      "step": 268
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.731e-05,
+      "loss": 0.7568,
+      "step": 269
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.730000000000001e-05,
+      "loss": 1.1569,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.729e-05,
+      "loss": 1.3048,
+      "step": 271
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.728e-05,
+      "loss": 1.3609,
+      "step": 272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.727000000000001e-05,
+      "loss": 0.8898,
+      "step": 273
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.726e-05,
+      "loss": 1.273,
+      "step": 274
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.725e-05,
+      "loss": 1.0708,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.724000000000001e-05,
+      "loss": 1.6123,
+      "step": 276
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.723000000000002e-05,
+      "loss": 0.4367,
+      "step": 277
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.722e-05,
+      "loss": 0.1648,
+      "step": 278
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.721e-05,
+      "loss": 0.2279,
+      "step": 279
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.72e-05,
+      "loss": 0.5328,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.719000000000001e-05,
+      "loss": 1.3129,
+      "step": 281
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.718e-05,
+      "loss": 1.3602,
+      "step": 282
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.717e-05,
+      "loss": 1.1465,
+      "step": 283
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.716000000000001e-05,
+      "loss": 1.0027,
+      "step": 284
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.715e-05,
+      "loss": 1.3141,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.714000000000001e-05,
+      "loss": 1.08,
+      "step": 286
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.713000000000001e-05,
+      "loss": 1.3428,
+      "step": 287
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.712e-05,
+      "loss": 1.8211,
+      "step": 288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.711e-05,
+      "loss": 0.7112,
+      "step": 289
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.71e-05,
+      "loss": 0.2153,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.709000000000001e-05,
+      "loss": 0.2229,
+      "step": 291
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.708e-05,
+      "loss": 0.3821,
+      "step": 292
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.707e-05,
+      "loss": 0.9344,
+      "step": 293
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.706000000000001e-05,
+      "loss": 1.2463,
+      "step": 294
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.705e-05,
+      "loss": 1.5315,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.704e-05,
+      "loss": 1.7242,
+      "step": 296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.703000000000001e-05,
+      "loss": 1.4012,
+      "step": 297
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.702e-05,
+      "loss": 1.0716,
+      "step": 298
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.701e-05,
+      "loss": 1.2233,
+      "step": 299
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.7e-05,
+      "loss": 1.4147,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.699e-05,
+      "loss": 1.6318,
+      "step": 301
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.698000000000001e-05,
+      "loss": 0.2226,
+      "step": 302
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.697e-05,
+      "loss": 0.1635,
+      "step": 303
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.696000000000001e-05,
+      "loss": 0.4435,
+      "step": 304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.695000000000001e-05,
+      "loss": 1.3365,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.694e-05,
+      "loss": 1.3978,
+      "step": 306
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.693000000000001e-05,
+      "loss": 1.3402,
+      "step": 307
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.692e-05,
+      "loss": 1.0531,
+      "step": 308
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.691e-05,
+      "loss": 1.5927,
+      "step": 309
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.69e-05,
+      "loss": 1.1876,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.689e-05,
+      "loss": 1.2744,
+      "step": 311
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.688000000000001e-05,
+      "loss": 1.4194,
+      "step": 312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.687e-05,
+      "loss": 1.1266,
+      "step": 313
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.686e-05,
+      "loss": 0.8382,
+      "step": 314
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.685000000000001e-05,
+      "loss": 0.2948,
+      "step": 315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.684000000000001e-05,
+      "loss": 0.09,
+      "step": 316
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.683e-05,
+      "loss": 0.4139,
+      "step": 317
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.682e-05,
+      "loss": 0.6556,
+      "step": 318
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.681e-05,
+      "loss": 0.7877,
+      "step": 319
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.680000000000001e-05,
+      "loss": 1.2062,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.679e-05,
+      "loss": 1.0222,
+      "step": 321
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.678e-05,
+      "loss": 1.726,
+      "step": 322
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.677000000000001e-05,
+      "loss": 1.0411,
+      "step": 323
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.676e-05,
+      "loss": 1.7161,
+      "step": 324
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.675000000000001e-05,
+      "loss": 2.4922,
+      "step": 325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.674000000000001e-05,
+      "loss": 1.7673,
+      "step": 326
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.673e-05,
+      "loss": 0.8039,
+      "step": 327
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.672e-05,
+      "loss": 0.0319,
+      "step": 328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.671e-05,
+      "loss": 0.0031,
+      "step": 329
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.67e-05,
+      "loss": 0.5966,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.669e-05,
+      "loss": 1.1531,
+      "step": 331
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.668e-05,
+      "loss": 0.9185,
+      "step": 332
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.667000000000001e-05,
+      "loss": 1.4574,
+      "step": 333
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.666e-05,
+      "loss": 1.2606,
+      "step": 334
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.665e-05,
+      "loss": 1.1825,
+      "step": 335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.664000000000001e-05,
+      "loss": 1.3139,
+      "step": 336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.663000000000002e-05,
+      "loss": 1.4452,
+      "step": 337
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.661999999999999e-05,
+      "loss": 2.3935,
+      "step": 338
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.661e-05,
+      "loss": 0.665,
+      "step": 339
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.66e-05,
+      "loss": 0.3476,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.659000000000001e-05,
+      "loss": 0.1796,
+      "step": 341
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.658e-05,
+      "loss": 0.3593,
+      "step": 342
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.657e-05,
+      "loss": 0.3148,
+      "step": 343
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.656000000000001e-05,
+      "loss": 1.1924,
+      "step": 344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.655e-05,
+      "loss": 0.9721,
+      "step": 345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.654000000000001e-05,
+      "loss": 1.8102,
+      "step": 346
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.653000000000001e-05,
+      "loss": 1.7172,
+      "step": 347
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.652e-05,
+      "loss": 1.3179,
+      "step": 348
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.651e-05,
+      "loss": 1.4838,
+      "step": 349
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.65e-05,
+      "loss": 1.1989,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.649e-05,
+      "loss": 1.3023,
+      "step": 351
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.648e-05,
+      "loss": 0.3855,
+      "step": 352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.647e-05,
+      "loss": 0.0664,
+      "step": 353
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.646000000000001e-05,
+      "loss": 0.0027,
+      "step": 354
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.645000000000001e-05,
+      "loss": 0.7807,
+      "step": 355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.644e-05,
+      "loss": 0.7082,
+      "step": 356
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.643000000000001e-05,
+      "loss": 1.309,
+      "step": 357
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.642e-05,
+      "loss": 1.1741,
+      "step": 358
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.641000000000001e-05,
+      "loss": 1.2552,
+      "step": 359
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.64e-05,
+      "loss": 1.7219,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.639e-05,
+      "loss": 1.337,
+      "step": 361
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.638000000000001e-05,
+      "loss": 0.7772,
+      "step": 362
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.637e-05,
+      "loss": 1.3173,
+      "step": 363
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.636e-05,
+      "loss": 0.8785,
+      "step": 364
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.635000000000001e-05,
+      "loss": 0.4206,
+      "step": 365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.634e-05,
+      "loss": 0.3733,
+      "step": 366
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.633000000000001e-05,
+      "loss": 0.2189,
+      "step": 367
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.632e-05,
+      "loss": 1.1153,
+      "step": 368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.631e-05,
+      "loss": 1.209,
+      "step": 369
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.63e-05,
+      "loss": 2.3862,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.629e-05,
+      "loss": 1.1535,
+      "step": 371
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.628e-05,
+      "loss": 1.1624,
+      "step": 372
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.627e-05,
+      "loss": 1.5244,
+      "step": 373
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.626e-05,
+      "loss": 1.3052,
+      "step": 374
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.625000000000001e-05,
+      "loss": 1.5366,
+      "step": 375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.624000000000001e-05,
+      "loss": 1.4342,
+      "step": 376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.623e-05,
+      "loss": 0.5158,
+      "step": 377
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.622000000000001e-05,
+      "loss": 0.0287,
+      "step": 378
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.621e-05,
+      "loss": 0.1761,
+      "step": 379
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.620000000000001e-05,
+      "loss": 0.086,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.619e-05,
+      "loss": 1.5867,
+      "step": 381
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.618e-05,
+      "loss": 1.6299,
+      "step": 382
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.617000000000001e-05,
+      "loss": 0.9489,
+      "step": 383
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.616e-05,
+      "loss": 1.224,
+      "step": 384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.615e-05,
+      "loss": 1.4334,
+      "step": 385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.614000000000001e-05,
+      "loss": 1.0389,
+      "step": 386
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.613e-05,
+      "loss": 1.3584,
+      "step": 387
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.612000000000001e-05,
+      "loss": 1.4874,
+      "step": 388
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.611e-05,
+      "loss": 0.5826,
+      "step": 389
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.61e-05,
+      "loss": 0.365,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.609e-05,
+      "loss": 0.5111,
+      "step": 391
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.608e-05,
+      "loss": 0.0315,
+      "step": 392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.607000000000001e-05,
+      "loss": 0.3675,
+      "step": 393
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.606000000000001e-05,
+      "loss": 0.7859,
+      "step": 394
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.605e-05,
+      "loss": 1.2154,
+      "step": 395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.604000000000001e-05,
+      "loss": 1.526,
+      "step": 396
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.603000000000001e-05,
+      "loss": 1.1852,
+      "step": 397
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.602e-05,
+      "loss": 1.4545,
+      "step": 398
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.601e-05,
+      "loss": 1.0362,
+      "step": 399
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.6e-05,
+      "loss": 1.2564,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.599000000000001e-05,
+      "loss": 1.4337,
+      "step": 401
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.598e-05,
+      "loss": 0.4902,
+      "step": 402
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.597e-05,
+      "loss": 0.2083,
+      "step": 403
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.596000000000001e-05,
+      "loss": 0.0964,
+      "step": 404
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.595e-05,
+      "loss": 0.3849,
+      "step": 405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.594e-05,
+      "loss": 1.7248,
+      "step": 406
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.593000000000001e-05,
+      "loss": 0.8921,
+      "step": 407
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.592e-05,
+      "loss": 1.0784,
+      "step": 408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.591e-05,
+      "loss": 1.278,
+      "step": 409
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.59e-05,
+      "loss": 1.7597,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.589e-05,
+      "loss": 1.0343,
+      "step": 411
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.588e-05,
+      "loss": 1.1177,
+      "step": 412
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.587e-05,
+      "loss": 1.4455,
+      "step": 413
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.586000000000001e-05,
+      "loss": 0.6078,
+      "step": 414
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.585000000000001e-05,
+      "loss": 0.1522,
+      "step": 415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.584e-05,
+      "loss": 0.2429,
+      "step": 416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.583000000000001e-05,
+      "loss": 0.1491,
+      "step": 417
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.582000000000001e-05,
+      "loss": 0.6509,
+      "step": 418
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.581e-05,
+      "loss": 0.7102,
+      "step": 419
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.58e-05,
+      "loss": 0.7805,
+      "step": 420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.579e-05,
+      "loss": 1.0778,
+      "step": 421
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.578000000000001e-05,
+      "loss": 1.359,
+      "step": 422
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.577e-05,
+      "loss": 1.3736,
+      "step": 423
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.576e-05,
+      "loss": 1.2362,
+      "step": 424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.575000000000001e-05,
+      "loss": 1.1764,
+      "step": 425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.574e-05,
+      "loss": 1.0224,
+      "step": 426
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.573e-05,
+      "loss": 0.2773,
+      "step": 427
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.572000000000001e-05,
+      "loss": 0.2168,
+      "step": 428
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.571e-05,
+      "loss": 0.3687,
+      "step": 429
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.57e-05,
+      "loss": 0.6725,
+      "step": 430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.569e-05,
+      "loss": 0.9386,
+      "step": 431
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.568e-05,
+      "loss": 1.4838,
+      "step": 432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.567000000000001e-05,
+      "loss": 1.1499,
+      "step": 433
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.566e-05,
+      "loss": 1.2,
+      "step": 434
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.565000000000001e-05,
+      "loss": 1.2652,
+      "step": 435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.564000000000001e-05,
+      "loss": 1.4311,
+      "step": 436
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.563e-05,
+      "loss": 1.3548,
+      "step": 437
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.562000000000001e-05,
+      "loss": 1.2621,
+      "step": 438
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.561e-05,
+      "loss": 0.1905,
+      "step": 439
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.56e-05,
+      "loss": 0.4325,
+      "step": 440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.559e-05,
+      "loss": 0.0396,
+      "step": 441
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.558e-05,
+      "loss": 0.5174,
+      "step": 442
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.557000000000001e-05,
+      "loss": 1.2584,
+      "step": 443
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.556e-05,
+      "loss": 1.0272,
+      "step": 444
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.555e-05,
+      "loss": 1.5793,
+      "step": 445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.554000000000001e-05,
+      "loss": 1.1685,
+      "step": 446
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.553e-05,
+      "loss": 1.1388,
+      "step": 447
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.552000000000001e-05,
+      "loss": 0.9716,
+      "step": 448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.551e-05,
+      "loss": 0.9506,
+      "step": 449
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.55e-05,
+      "loss": 1.6895,
+      "step": 450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.549e-05,
+      "loss": 1.0962,
+      "step": 451
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.548e-05,
+      "loss": 0.913,
+      "step": 452
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.547e-05,
+      "loss": 0.3286,
+      "step": 453
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.546000000000001e-05,
+      "loss": 0.059,
+      "step": 454
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.545e-05,
+      "loss": 0.1163,
+      "step": 455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.544000000000001e-05,
+      "loss": 1.1911,
+      "step": 456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.543000000000001e-05,
+      "loss": 1.4272,
+      "step": 457
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.542e-05,
+      "loss": 1.2541,
+      "step": 458
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.541e-05,
+      "loss": 1.3129,
+      "step": 459
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.54e-05,
+      "loss": 1.0522,
+      "step": 460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.539e-05,
+      "loss": 1.0866,
+      "step": 461
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.538e-05,
+      "loss": 1.3613,
+      "step": 462
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.537e-05,
+      "loss": 1.6409,
+      "step": 463
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.536000000000001e-05,
+      "loss": 0.9909,
+      "step": 464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.535e-05,
+      "loss": 0.265,
+      "step": 465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.534e-05,
+      "loss": 0.0987,
+      "step": 466
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.533000000000001e-05,
+      "loss": 0.093,
+      "step": 467
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.532000000000002e-05,
+      "loss": 0.4013,
+      "step": 468
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.531e-05,
+      "loss": 1.1758,
+      "step": 469
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.53e-05,
+      "loss": 1.1598,
+      "step": 470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.529e-05,
+      "loss": 1.2626,
+      "step": 471
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.528000000000001e-05,
+      "loss": 1.0471,
+      "step": 472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.527e-05,
+      "loss": 1.1345,
+      "step": 473
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.526e-05,
+      "loss": 1.4549,
+      "step": 474
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.525000000000001e-05,
+      "loss": 1.532,
+      "step": 475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.524e-05,
+      "loss": 1.2065,
+      "step": 476
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.523000000000001e-05,
+      "loss": 0.3352,
+      "step": 477
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.522000000000001e-05,
+      "loss": 0.521,
+      "step": 478
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.521e-05,
+      "loss": 0.4547,
+      "step": 479
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.52e-05,
+      "loss": 0.637,
+      "step": 480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.519e-05,
+      "loss": 0.9398,
+      "step": 481
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.518000000000001e-05,
+      "loss": 1.2614,
+      "step": 482
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.517e-05,
+      "loss": 0.9308,
+      "step": 483
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.516e-05,
+      "loss": 1.2233,
+      "step": 484
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.515000000000001e-05,
+      "loss": 0.9142,
+      "step": 485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.514e-05,
+      "loss": 1.4805,
+      "step": 486
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.513e-05,
+      "loss": 0.889,
+      "step": 487
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.512000000000001e-05,
+      "loss": 1.5211,
+      "step": 488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.511e-05,
+      "loss": 0.8276,
+      "step": 489
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.51e-05,
+      "loss": 0.5433,
+      "step": 490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.509e-05,
+      "loss": 0.3888,
+      "step": 491
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.508e-05,
+      "loss": 0.253,
+      "step": 492
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.507000000000001e-05,
+      "loss": 0.3107,
+      "step": 493
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.506e-05,
+      "loss": 0.7182,
+      "step": 494
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.505e-05,
+      "loss": 1.5798,
+      "step": 495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.504000000000001e-05,
+      "loss": 1.3765,
+      "step": 496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.503e-05,
+      "loss": 1.394,
+      "step": 497
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.502000000000001e-05,
+      "loss": 1.89,
+      "step": 498
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.501e-05,
+      "loss": 1.1796,
+      "step": 499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.5e-05,
+      "loss": 1.6569,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.499e-05,
+      "loss": 2.4708,
+      "step": 501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.498e-05,
+      "loss": 0.3504,
+      "step": 502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.497000000000001e-05,
+      "loss": 0.2038,
+      "step": 503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.496e-05,
+      "loss": 0.326,
+      "step": 504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.495e-05,
+      "loss": 0.6632,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.494000000000001e-05,
+      "loss": 1.1833,
+      "step": 506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.493000000000001e-05,
+      "loss": 0.9039,
+      "step": 507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.492e-05,
+      "loss": 1.191,
+      "step": 508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.491000000000001e-05,
+      "loss": 1.5764,
+      "step": 509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.49e-05,
+      "loss": 1.3158,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.489e-05,
+      "loss": 1.3365,
+      "step": 511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.488e-05,
+      "loss": 1.5832,
+      "step": 512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.487e-05,
+      "loss": 1.2481,
+      "step": 513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.486000000000001e-05,
+      "loss": 0.8569,
+      "step": 514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.485e-05,
+      "loss": 0.2305,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.484e-05,
+      "loss": 0.2454,
+      "step": 516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.483000000000001e-05,
+      "loss": 0.1122,
+      "step": 517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.482e-05,
+      "loss": 1.2373,
+      "step": 518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.481000000000001e-05,
+      "loss": 0.9594,
+      "step": 519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.48e-05,
+      "loss": 1.0952,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.479e-05,
+      "loss": 1.061,
+      "step": 521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.478e-05,
+      "loss": 1.1364,
+      "step": 522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.477e-05,
+      "loss": 1.2291,
+      "step": 523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.476000000000001e-05,
+      "loss": 1.3026,
+      "step": 524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.475e-05,
+      "loss": 1.1565,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.474e-05,
+      "loss": 0.867,
+      "step": 526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.473000000000001e-05,
+      "loss": 0.2744,
+      "step": 527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.472000000000001e-05,
+      "loss": 0.2364,
+      "step": 528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.471e-05,
+      "loss": 0.1303,
+      "step": 529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.47e-05,
+      "loss": 0.2366,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.469e-05,
+      "loss": 0.8382,
+      "step": 531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.468000000000001e-05,
+      "loss": 1.6347,
+      "step": 532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.467e-05,
+      "loss": 0.9925,
+      "step": 533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.466e-05,
+      "loss": 1.1191,
+      "step": 534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.465000000000001e-05,
+      "loss": 1.2145,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.464e-05,
+      "loss": 0.7468,
+      "step": 536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.463000000000001e-05,
+      "loss": 1.6879,
+      "step": 537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.462000000000001e-05,
+      "loss": 1.3572,
+      "step": 538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.461e-05,
+      "loss": 0.6752,
+      "step": 539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.46e-05,
+      "loss": 0.1666,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.459e-05,
+      "loss": 0.3675,
+      "step": 541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.458e-05,
+      "loss": 0.89,
+      "step": 542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.457e-05,
+      "loss": 1.2539,
+      "step": 543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.456e-05,
+      "loss": 1.1738,
+      "step": 544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.455000000000001e-05,
+      "loss": 1.1053,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.454000000000001e-05,
+      "loss": 0.9746,
+      "step": 546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.453e-05,
+      "loss": 1.3563,
+      "step": 547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.452000000000001e-05,
+      "loss": 0.9374,
+      "step": 548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.451000000000002e-05,
+      "loss": 1.3496,
+      "step": 549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.449999999999999e-05,
+      "loss": 1.235,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.449e-05,
+      "loss": 0.886,
+      "step": 551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.448e-05,
+      "loss": 0.3953,
+      "step": 552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.447000000000001e-05,
+      "loss": 0.3649,
+      "step": 553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.446e-05,
+      "loss": 0.2361,
+      "step": 554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.445e-05,
+      "loss": 0.4571,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.444000000000001e-05,
+      "loss": 1.153,
+      "step": 556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.443e-05,
+      "loss": 1.3535,
+      "step": 557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.442000000000001e-05,
+      "loss": 1.1155,
+      "step": 558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.441000000000001e-05,
+      "loss": 1.1134,
+      "step": 559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.44e-05,
+      "loss": 1.5228,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.439e-05,
+      "loss": 1.126,
+      "step": 561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.438e-05,
+      "loss": 1.2701,
+      "step": 562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.437e-05,
+      "loss": 1.6568,
+      "step": 563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.436e-05,
+      "loss": 0.6307,
+      "step": 564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.435e-05,
+      "loss": 0.1789,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.434000000000001e-05,
+      "loss": 0.0215,
+      "step": 566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.433000000000001e-05,
+      "loss": 0.1956,
+      "step": 567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.432e-05,
+      "loss": 0.4574,
+      "step": 568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.431000000000001e-05,
+      "loss": 1.0574,
+      "step": 569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.43e-05,
+      "loss": 1.2892,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.429000000000001e-05,
+      "loss": 0.7791,
+      "step": 571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.428e-05,
+      "loss": 1.4321,
+      "step": 572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.427e-05,
+      "loss": 1.4091,
+      "step": 573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.426000000000001e-05,
+      "loss": 1.278,
+      "step": 574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.425e-05,
+      "loss": 1.4202,
+      "step": 575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.424e-05,
+      "loss": 1.4371,
+      "step": 576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.423000000000001e-05,
+      "loss": 0.704,
+      "step": 577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.422e-05,
+      "loss": 0.26,
+      "step": 578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.421000000000001e-05,
+      "loss": 0.2493,
+      "step": 579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.42e-05,
+      "loss": 0.0441,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.419e-05,
+      "loss": 0.9571,
+      "step": 581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.418e-05,
+      "loss": 1.0837,
+      "step": 582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.417e-05,
+      "loss": 1.2545,
+      "step": 583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.416e-05,
+      "loss": 1.4749,
+      "step": 584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.415e-05,
+      "loss": 1.6649,
+      "step": 585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.414e-05,
+      "loss": 1.694,
+      "step": 586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.413000000000001e-05,
+      "loss": 1.0465,
+      "step": 587
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.412000000000001e-05,
+      "loss": 1.5054,
+      "step": 588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.411e-05,
+      "loss": 0.3327,
+      "step": 589
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.41e-05,
+      "loss": 0.1664,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.409e-05,
+      "loss": 0.1055,
+      "step": 591
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.408000000000001e-05,
+      "loss": 0.4805,
+      "step": 592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.407e-05,
+      "loss": 1.9282,
+      "step": 593
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.406e-05,
+      "loss": 1.1295,
+      "step": 594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.405000000000001e-05,
+      "loss": 1.2192,
+      "step": 595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.404e-05,
+      "loss": 1.5333,
+      "step": 596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.403e-05,
+      "loss": 1.2271,
+      "step": 597
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.402000000000001e-05,
+      "loss": 0.898,
+      "step": 598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.401e-05,
+      "loss": 1.197,
+      "step": 599
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.4e-05,
+      "loss": 1.0845,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.399e-05,
+      "loss": 1.3517,
+      "step": 601
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.398e-05,
+      "loss": 0.6654,
+      "step": 602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.397e-05,
+      "loss": 0.2364,
+      "step": 603
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.396e-05,
+      "loss": 0.4093,
+      "step": 604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.395000000000001e-05,
+      "loss": 0.383,
+      "step": 605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.394000000000001e-05,
+      "loss": 0.7704,
+      "step": 606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.393e-05,
+      "loss": 1.0373,
+      "step": 607
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.392000000000001e-05,
+      "loss": 1.3647,
+      "step": 608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.391000000000001e-05,
+      "loss": 1.4541,
+      "step": 609
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.39e-05,
+      "loss": 1.355,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.389e-05,
+      "loss": 1.027,
+      "step": 611
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.388e-05,
+      "loss": 1.2059,
+      "step": 612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.387000000000001e-05,
+      "loss": 1.4198,
+      "step": 613
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.386e-05,
+      "loss": 1.0881,
+      "step": 614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.385e-05,
+      "loss": 0.3812,
+      "step": 615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.384000000000001e-05,
+      "loss": 0.2511,
+      "step": 616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.383e-05,
+      "loss": 0.2056,
+      "step": 617
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.382e-05,
+      "loss": 0.3509,
+      "step": 618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.381000000000001e-05,
+      "loss": 0.6825,
+      "step": 619
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.38e-05,
+      "loss": 1.0566,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.379e-05,
+      "loss": 1.2205,
+      "step": 621
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.378e-05,
+      "loss": 1.1354,
+      "step": 622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.377e-05,
+      "loss": 1.0536,
+      "step": 623
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.376e-05,
+      "loss": 0.7985,
+      "step": 624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.375e-05,
+      "loss": 1.7129,
+      "step": 625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.374000000000001e-05,
+      "loss": 1.5255,
+      "step": 626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.373000000000001e-05,
+      "loss": 0.6464,
+      "step": 627
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.372e-05,
+      "loss": 0.0407,
+      "step": 628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.371000000000001e-05,
+      "loss": 0.0379,
+      "step": 629
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.370000000000001e-05,
+      "loss": 0.0059,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.369e-05,
+      "loss": 0.2609,
+      "step": 631
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.368e-05,
+      "loss": 0.8964,
+      "step": 632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.367e-05,
+      "loss": 1.4672,
+      "step": 633
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.366000000000001e-05,
+      "loss": 1.2947,
+      "step": 634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.365e-05,
+      "loss": 1.319,
+      "step": 635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.364e-05,
+      "loss": 1.181,
+      "step": 636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.363000000000001e-05,
+      "loss": 0.9135,
+      "step": 637
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.362e-05,
+      "loss": 1.4152,
+      "step": 638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.361e-05,
+      "loss": 0.9077,
+      "step": 639
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.360000000000001e-05,
+      "loss": 0.2298,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.359e-05,
+      "loss": 0.032,
+      "step": 641
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.358e-05,
+      "loss": 0.2461,
+      "step": 642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.357e-05,
+      "loss": 0.0517,
+      "step": 643
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.356e-05,
+      "loss": 0.9401,
+      "step": 644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.355000000000001e-05,
+      "loss": 0.8198,
+      "step": 645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.354e-05,
+      "loss": 1.3786,
+      "step": 646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.353000000000001e-05,
+      "loss": 1.086,
+      "step": 647
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.352000000000001e-05,
+      "loss": 1.3785,
+      "step": 648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.351e-05,
+      "loss": 1.1653,
+      "step": 649
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.350000000000001e-05,
+      "loss": 1.4604,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.349e-05,
+      "loss": 1.4185,
+      "step": 651
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.348e-05,
+      "loss": 0.32,
+      "step": 652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.347e-05,
+      "loss": 0.0162,
+      "step": 653
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.346e-05,
+      "loss": 0.6172,
+      "step": 654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.345000000000001e-05,
+      "loss": 0.1026,
+      "step": 655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.344e-05,
+      "loss": 1.172,
+      "step": 656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.343e-05,
+      "loss": 1.5152,
+      "step": 657
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.342000000000001e-05,
+      "loss": 1.1055,
+      "step": 658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.341000000000002e-05,
+      "loss": 1.413,
+      "step": 659
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.340000000000001e-05,
+      "loss": 1.2869,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.339e-05,
+      "loss": 0.9953,
+      "step": 661
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.338e-05,
+      "loss": 1.5404,
+      "step": 662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.337e-05,
+      "loss": 1.1724,
+      "step": 663
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.336e-05,
+      "loss": 1.0243,
+      "step": 664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.335e-05,
+      "loss": 0.2395,
+      "step": 665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.334000000000001e-05,
+      "loss": 0.1125,
+      "step": 666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.333e-05,
+      "loss": 0.0036,
+      "step": 667
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.332000000000001e-05,
+      "loss": 0.6662,
+      "step": 668
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.331000000000001e-05,
+      "loss": 1.1542,
+      "step": 669
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.33e-05,
+      "loss": 1.8092,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.329e-05,
+      "loss": 1.0888,
+      "step": 671
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.328e-05,
+      "loss": 0.8886,
+      "step": 672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.327e-05,
+      "loss": 1.16,
+      "step": 673
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.326e-05,
+      "loss": 1.5759,
+      "step": 674
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.325e-05,
+      "loss": 1.4589,
+      "step": 675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.324000000000001e-05,
+      "loss": 1.3546,
+      "step": 676
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.323e-05,
+      "loss": 0.3145,
+      "step": 677
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.322e-05,
+      "loss": 0.1953,
+      "step": 678
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.321000000000001e-05,
+      "loss": 0.1691,
+      "step": 679
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.320000000000002e-05,
+      "loss": 0.3196,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.319e-05,
+      "loss": 0.3086,
+      "step": 681
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.318e-05,
+      "loss": 1.32,
+      "step": 682
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.317e-05,
+      "loss": 1.4657,
+      "step": 683
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.316000000000001e-05,
+      "loss": 1.4749,
+      "step": 684
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.315e-05,
+      "loss": 1.2188,
+      "step": 685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.314e-05,
+      "loss": 1.1331,
+      "step": 686
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.313000000000001e-05,
+      "loss": 1.8643,
+      "step": 687
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.312e-05,
+      "loss": 1.8186,
+      "step": 688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.311000000000001e-05,
+      "loss": 1.3654,
+      "step": 689
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.310000000000001e-05,
+      "loss": 0.4859,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.309e-05,
+      "loss": 0.407,
+      "step": 691
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.308e-05,
+      "loss": 0.5704,
+      "step": 692
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.307e-05,
+      "loss": 0.6246,
+      "step": 693
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.306000000000001e-05,
+      "loss": 0.8035,
+      "step": 694
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.305e-05,
+      "loss": 1.4227,
+      "step": 695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.304e-05,
+      "loss": 1.1268,
+      "step": 696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.303000000000001e-05,
+      "loss": 1.0631,
+      "step": 697
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.302e-05,
+      "loss": 1.1822,
+      "step": 698
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.301e-05,
+      "loss": 0.9263,
+      "step": 699
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 2.2675,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.299e-05,
+      "loss": 1.0923,
+      "step": 701
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.298e-05,
+      "loss": 0.1158,
+      "step": 702
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.297e-05,
+      "loss": 0.3168,
+      "step": 703
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.296e-05,
+      "loss": 0.1001,
+      "step": 704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.295000000000001e-05,
+      "loss": 0.833,
+      "step": 705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.294e-05,
+      "loss": 1.2877,
+      "step": 706
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.293e-05,
+      "loss": 1.2462,
+      "step": 707
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.292000000000001e-05,
+      "loss": 0.87,
+      "step": 708
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.291e-05,
+      "loss": 1.3379,
+      "step": 709
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.290000000000001e-05,
+      "loss": 0.9283,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.289e-05,
+      "loss": 1.1484,
+      "step": 711
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.288e-05,
+      "loss": 1.1065,
+      "step": 712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.287e-05,
+      "loss": 1.2207,
+      "step": 713
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.286e-05,
+      "loss": 0.5335,
+      "step": 714
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.285000000000001e-05,
+      "loss": 0.0574,
+      "step": 715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.284e-05,
+      "loss": 0.5301,
+      "step": 716
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.283e-05,
+      "loss": 0.7394,
+      "step": 717
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.282000000000001e-05,
+      "loss": 0.9677,
+      "step": 718
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.281000000000001e-05,
+      "loss": 1.232,
+      "step": 719
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.28e-05,
+      "loss": 1.4723,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.279e-05,
+      "loss": 1.0329,
+      "step": 721
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.278e-05,
+      "loss": 1.1648,
+      "step": 722
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.277000000000001e-05,
+      "loss": 1.1432,
+      "step": 723
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.276e-05,
+      "loss": 1.1713,
+      "step": 724
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.275e-05,
+      "loss": 1.4125,
+      "step": 725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.274000000000001e-05,
+      "loss": 1.8072,
+      "step": 726
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.273e-05,
+      "loss": 0.3401,
+      "step": 727
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.272e-05,
+      "loss": 0.2792,
+      "step": 728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.271000000000001e-05,
+      "loss": 0.0228,
+      "step": 729
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.27e-05,
+      "loss": 0.1775,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.269e-05,
+      "loss": 0.9601,
+      "step": 731
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.268e-05,
+      "loss": 1.4092,
+      "step": 732
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.267e-05,
+      "loss": 1.3338,
+      "step": 733
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.266e-05,
+      "loss": 0.7999,
+      "step": 734
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.265e-05,
+      "loss": 1.2458,
+      "step": 735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.264000000000001e-05,
+      "loss": 1.196,
+      "step": 736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.263e-05,
+      "loss": 1.225,
+      "step": 737
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.262e-05,
+      "loss": 1.2512,
+      "step": 738
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.261000000000001e-05,
+      "loss": 0.3602,
+      "step": 739
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.260000000000001e-05,
+      "loss": 0.0754,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.258999999999999e-05,
+      "loss": 0.0952,
+      "step": 741
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.258e-05,
+      "loss": 0.5211,
+      "step": 742
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.257e-05,
+      "loss": 1.1148,
+      "step": 743
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.256000000000001e-05,
+      "loss": 1.1191,
+      "step": 744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.255e-05,
+      "loss": 1.1697,
+      "step": 745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.254e-05,
+      "loss": 1.5152,
+      "step": 746
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.253000000000001e-05,
+      "loss": 1.3406,
+      "step": 747
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.252e-05,
+      "loss": 1.1525,
+      "step": 748
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.251000000000001e-05,
+      "loss": 0.8296,
+      "step": 749
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.250000000000001e-05,
+      "loss": 1.2499,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.249e-05,
+      "loss": 1.2428,
+      "step": 751
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.248e-05,
+      "loss": 0.3554,
+      "step": 752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.247e-05,
+      "loss": 0.3077,
+      "step": 753
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.246e-05,
+      "loss": 0.1377,
+      "step": 754
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.245e-05,
+      "loss": 0.8009,
+      "step": 755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.244e-05,
+      "loss": 1.1617,
+      "step": 756
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.243000000000001e-05,
+      "loss": 1.0752,
+      "step": 757
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.242000000000001e-05,
+      "loss": 1.2984,
+      "step": 758
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.241e-05,
+      "loss": 1.2314,
+      "step": 759
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.240000000000001e-05,
+      "loss": 1.1578,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.239000000000001e-05,
+      "loss": 0.844,
+      "step": 761
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.238e-05,
+      "loss": 0.8398,
+      "step": 762
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.237e-05,
+      "loss": 1.4431,
+      "step": 763
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.236e-05,
+      "loss": 1.2718,
+      "step": 764
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.235000000000001e-05,
+      "loss": 0.2484,
+      "step": 765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.234e-05,
+      "loss": 0.3292,
+      "step": 766
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.233e-05,
+      "loss": 0.2173,
+      "step": 767
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.232000000000001e-05,
+      "loss": 0.563,
+      "step": 768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.231e-05,
+      "loss": 1.2324,
+      "step": 769
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230000000000001e-05,
+      "loss": 1.2178,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.229000000000001e-05,
+      "loss": 1.3075,
+      "step": 771
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.228e-05,
+      "loss": 0.8687,
+      "step": 772
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.227e-05,
+      "loss": 0.9937,
+      "step": 773
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.226e-05,
+      "loss": 1.1047,
+      "step": 774
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.225e-05,
+      "loss": 0.6287,
+      "step": 775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.224e-05,
+      "loss": 1.523,
+      "step": 776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.223e-05,
+      "loss": 0.5755,
+      "step": 777
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.222000000000001e-05,
+      "loss": 0.132,
+      "step": 778
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.221000000000001e-05,
+      "loss": 0.4426,
+      "step": 779
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.22e-05,
+      "loss": 0.3704,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.219000000000001e-05,
+      "loss": 0.6924,
+      "step": 781
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.218e-05,
+      "loss": 1.1541,
+      "step": 782
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.217000000000001e-05,
+      "loss": 1.1696,
+      "step": 783
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.216e-05,
+      "loss": 0.7751,
+      "step": 784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.215e-05,
+      "loss": 1.3308,
+      "step": 785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.214000000000001e-05,
+      "loss": 1.1422,
+      "step": 786
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.213e-05,
+      "loss": 1.5507,
+      "step": 787
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.212e-05,
+      "loss": 1.2644,
+      "step": 788
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.211000000000001e-05,
+      "loss": 0.9964,
+      "step": 789
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.21e-05,
+      "loss": 0.139,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.209000000000001e-05,
+      "loss": 0.118,
+      "step": 791
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.208e-05,
+      "loss": 0.1334,
+      "step": 792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.207e-05,
+      "loss": 0.2923,
+      "step": 793
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.206e-05,
+      "loss": 1.6501,
+      "step": 794
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.205e-05,
+      "loss": 1.3353,
+      "step": 795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.204e-05,
+      "loss": 1.155,
+      "step": 796
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.203000000000001e-05,
+      "loss": 1.2034,
+      "step": 797
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.202e-05,
+      "loss": 0.9721,
+      "step": 798
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.201000000000001e-05,
+      "loss": 1.3689,
+      "step": 799
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.3014,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.199e-05,
+      "loss": 1.2588,
+      "step": 801
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.198e-05,
+      "loss": 0.3503,
+      "step": 802
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.197e-05,
+      "loss": 0.5511,
+      "step": 803
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.196000000000001e-05,
+      "loss": 0.3011,
+      "step": 804
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.195e-05,
+      "loss": 0.91,
+      "step": 805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.194e-05,
+      "loss": 0.8268,
+      "step": 806
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.193000000000001e-05,
+      "loss": 1.5088,
+      "step": 807
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.192e-05,
+      "loss": 1.414,
+      "step": 808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.191e-05,
+      "loss": 1.2331,
+      "step": 809
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.190000000000001e-05,
+      "loss": 1.2194,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.189e-05,
+      "loss": 0.9149,
+      "step": 811
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.188e-05,
+      "loss": 0.9459,
+      "step": 812
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.187e-05,
+      "loss": 1.0878,
+      "step": 813
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.186e-05,
+      "loss": 1.0531,
+      "step": 814
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.185e-05,
+      "loss": 0.0213,
+      "step": 815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.184e-05,
+      "loss": 0.1047,
+      "step": 816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.183000000000001e-05,
+      "loss": 0.1161,
+      "step": 817
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.182000000000001e-05,
+      "loss": 0.5085,
+      "step": 818
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.181e-05,
+      "loss": 0.9932,
+      "step": 819
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 1.6133,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.179000000000001e-05,
+      "loss": 1.1411,
+      "step": 821
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.178e-05,
+      "loss": 1.0408,
+      "step": 822
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.177e-05,
+      "loss": 1.2002,
+      "step": 823
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.176e-05,
+      "loss": 1.161,
+      "step": 824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.175000000000001e-05,
+      "loss": 1.2134,
+      "step": 825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.174e-05,
+      "loss": 0.9627,
+      "step": 826
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.173e-05,
+      "loss": 0.4392,
+      "step": 827
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.172000000000001e-05,
+      "loss": 0.2666,
+      "step": 828
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.171e-05,
+      "loss": 0.1497,
+      "step": 829
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.17e-05,
+      "loss": 0.1445,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.169000000000001e-05,
+      "loss": 1.1215,
+      "step": 831
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.168e-05,
+      "loss": 1.0236,
+      "step": 832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.167e-05,
+      "loss": 1.3365,
+      "step": 833
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.166e-05,
+      "loss": 1.5384,
+      "step": 834
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.165e-05,
+      "loss": 1.3037,
+      "step": 835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.164000000000001e-05,
+      "loss": 1.1966,
+      "step": 836
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.163e-05,
+      "loss": 1.4053,
+      "step": 837
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.162000000000001e-05,
+      "loss": 1.6615,
+      "step": 838
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.161000000000001e-05,
+      "loss": 1.1001,
+      "step": 839
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.16e-05,
+      "loss": 0.022,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.159000000000001e-05,
+      "loss": 0.0856,
+      "step": 841
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.158e-05,
+      "loss": 0.0431,
+      "step": 842
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.157e-05,
+      "loss": 0.3837,
+      "step": 843
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.156e-05,
+      "loss": 1.4496,
+      "step": 844
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.155e-05,
+      "loss": 1.3456,
+      "step": 845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.154000000000001e-05,
+      "loss": 1.5505,
+      "step": 846
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.153e-05,
+      "loss": 1.1021,
+      "step": 847
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.152e-05,
+      "loss": 1.1787,
+      "step": 848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.151000000000001e-05,
+      "loss": 0.947,
+      "step": 849
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.15e-05,
+      "loss": 0.6364,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.149e-05,
+      "loss": 0.8999,
+      "step": 851
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.148e-05,
+      "loss": 0.4581,
+      "step": 852
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.147e-05,
+      "loss": 0.1326,
+      "step": 853
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.146e-05,
+      "loss": 0.0378,
+      "step": 854
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.145e-05,
+      "loss": 0.6563,
+      "step": 855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.144e-05,
+      "loss": 0.8997,
+      "step": 856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.143000000000001e-05,
+      "loss": 1.0617,
+      "step": 857
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.142e-05,
+      "loss": 1.3542,
+      "step": 858
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.141000000000001e-05,
+      "loss": 1.3233,
+      "step": 859
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.140000000000001e-05,
+      "loss": 1.3446,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.139e-05,
+      "loss": 1.4233,
+      "step": 861
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.138e-05,
+      "loss": 1.2892,
+      "step": 862
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.137e-05,
+      "loss": 1.5323,
+      "step": 863
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.136e-05,
+      "loss": 0.5489,
+      "step": 864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.135e-05,
+      "loss": 0.1882,
+      "step": 865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.134e-05,
+      "loss": 0.2188,
+      "step": 866
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.133000000000001e-05,
+      "loss": 0.2808,
+      "step": 867
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.132e-05,
+      "loss": 0.0169,
+      "step": 868
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.131e-05,
+      "loss": 0.8825,
+      "step": 869
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.130000000000001e-05,
+      "loss": 1.5045,
+      "step": 870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.129000000000002e-05,
+      "loss": 1.5223,
+      "step": 871
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.128e-05,
+      "loss": 1.5253,
+      "step": 872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.127e-05,
+      "loss": 1.4199,
+      "step": 873
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.126e-05,
+      "loss": 0.882,
+      "step": 874
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.125e-05,
+      "loss": 0.9242,
+      "step": 875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.124e-05,
+      "loss": 1.177,
+      "step": 876
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.123e-05,
+      "loss": 0.1219,
+      "step": 877
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.122000000000001e-05,
+      "loss": 0.2349,
+      "step": 878
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.121e-05,
+      "loss": 0.2104,
+      "step": 879
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.120000000000001e-05,
+      "loss": 0.3664,
+      "step": 880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.119000000000001e-05,
+      "loss": 0.8478,
+      "step": 881
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.118e-05,
+      "loss": 1.6416,
+      "step": 882
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.117e-05,
+      "loss": 1.1534,
+      "step": 883
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.116e-05,
+      "loss": 1.1073,
+      "step": 884
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.115e-05,
+      "loss": 1.1334,
+      "step": 885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.114e-05,
+      "loss": 1.0194,
+      "step": 886
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.113e-05,
+      "loss": 1.3613,
+      "step": 887
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.112000000000001e-05,
+      "loss": 1.432,
+      "step": 888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.111e-05,
+      "loss": 0.3827,
+      "step": 889
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.11e-05,
+      "loss": 0.1765,
+      "step": 890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.109000000000001e-05,
+      "loss": 0.0241,
+      "step": 891
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.108000000000002e-05,
+      "loss": 0.2739,
+      "step": 892
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.107e-05,
+      "loss": 0.698,
+      "step": 893
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.106e-05,
+      "loss": 1.295,
+      "step": 894
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.105e-05,
+      "loss": 1.375,
+      "step": 895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.104000000000001e-05,
+      "loss": 1.1992,
+      "step": 896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.103e-05,
+      "loss": 1.1557,
+      "step": 897
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.102e-05,
+      "loss": 1.806,
+      "step": 898
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.101000000000001e-05,
+      "loss": 1.3137,
+      "step": 899
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.1e-05,
+      "loss": 1.3867,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.099000000000001e-05,
+      "loss": 1.526,
+      "step": 901
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.098000000000001e-05,
+      "loss": 0.441,
+      "step": 902
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.097e-05,
+      "loss": 0.0558,
+      "step": 903
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.096e-05,
+      "loss": 0.1053,
+      "step": 904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.095e-05,
+      "loss": 0.673,
+      "step": 905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.094000000000001e-05,
+      "loss": 1.018,
+      "step": 906
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.093e-05,
+      "loss": 1.4499,
+      "step": 907
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.092e-05,
+      "loss": 1.6178,
+      "step": 908
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.091000000000001e-05,
+      "loss": 1.0608,
+      "step": 909
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.090000000000001e-05,
+      "loss": 1.459,
+      "step": 910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.089e-05,
+      "loss": 1.1478,
+      "step": 911
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.088000000000001e-05,
+      "loss": 1.2115,
+      "step": 912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.087e-05,
+      "loss": 1.2369,
+      "step": 913
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.086e-05,
+      "loss": 0.8332,
+      "step": 914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.085e-05,
+      "loss": 0.3152,
+      "step": 915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.084e-05,
+      "loss": 0.023,
+      "step": 916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.083000000000001e-05,
+      "loss": 0.039,
+      "step": 917
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.082e-05,
+      "loss": 0.8848,
+      "step": 918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.081e-05,
+      "loss": 1.2993,
+      "step": 919
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.080000000000001e-05,
+      "loss": 1.3162,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.079e-05,
+      "loss": 1.6017,
+      "step": 921
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.078000000000001e-05,
+      "loss": 1.3855,
+      "step": 922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.077e-05,
+      "loss": 1.6381,
+      "step": 923
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.076e-05,
+      "loss": 1.1817,
+      "step": 924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.075e-05,
+      "loss": 1.2006,
+      "step": 925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.074e-05,
+      "loss": 1.3664,
+      "step": 926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.073000000000001e-05,
+      "loss": 0.2274,
+      "step": 927
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.072e-05,
+      "loss": 0.1716,
+      "step": 928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.071e-05,
+      "loss": 0.1923,
+      "step": 929
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.070000000000001e-05,
+      "loss": 0.3284,
+      "step": 930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.069000000000001e-05,
+      "loss": 1.1887,
+      "step": 931
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.068e-05,
+      "loss": 1.7428,
+      "step": 932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.067e-05,
+      "loss": 1.441,
+      "step": 933
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.066e-05,
+      "loss": 1.2436,
+      "step": 934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.065000000000001e-05,
+      "loss": 1.4548,
+      "step": 935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.064e-05,
+      "loss": 1.2193,
+      "step": 936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.063e-05,
+      "loss": 1.5415,
+      "step": 937
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.062000000000001e-05,
+      "loss": 1.6813,
+      "step": 938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.061e-05,
+      "loss": 0.3475,
+      "step": 939
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.06e-05,
+      "loss": 0.0768,
+      "step": 940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.059000000000001e-05,
+      "loss": 0.0218,
+      "step": 941
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.058e-05,
+      "loss": 0.0049,
+      "step": 942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.057e-05,
+      "loss": 1.1634,
+      "step": 943
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.056e-05,
+      "loss": 1.0188,
+      "step": 944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.055e-05,
+      "loss": 1.4352,
+      "step": 945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.054e-05,
+      "loss": 1.0628,
+      "step": 946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.053e-05,
+      "loss": 1.3384,
+      "step": 947
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.052000000000001e-05,
+      "loss": 0.791,
+      "step": 948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.051000000000001e-05,
+      "loss": 0.7111,
+      "step": 949
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.05e-05,
+      "loss": 1.5631,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.049000000000001e-05,
+      "loss": 1.3879,
+      "step": 951
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.048000000000001e-05,
+      "loss": 0.5586,
+      "step": 952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.046999999999999e-05,
+      "loss": 0.1971,
+      "step": 953
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.046e-05,
+      "loss": 0.0346,
+      "step": 954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.045e-05,
+      "loss": 0.2026,
+      "step": 955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.044000000000001e-05,
+      "loss": 0.8764,
+      "step": 956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.043e-05,
+      "loss": 0.7577,
+      "step": 957
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.042e-05,
+      "loss": 1.1374,
+      "step": 958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.041000000000001e-05,
+      "loss": 1.2919,
+      "step": 959
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.04e-05,
+      "loss": 1.2646,
+      "step": 960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.039000000000001e-05,
+      "loss": 1.4474,
+      "step": 961
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.038000000000001e-05,
+      "loss": 1.3584,
+      "step": 962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.037e-05,
+      "loss": 1.4003,
+      "step": 963
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.036e-05,
+      "loss": 0.4215,
+      "step": 964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.035e-05,
+      "loss": 0.0969,
+      "step": 965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.034e-05,
+      "loss": 0.2179,
+      "step": 966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.033e-05,
+      "loss": 0.1305,
+      "step": 967
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.032e-05,
+      "loss": 0.2123,
+      "step": 968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.031000000000001e-05,
+      "loss": 1.281,
+      "step": 969
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.030000000000001e-05,
+      "loss": 1.1166,
+      "step": 970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.029e-05,
+      "loss": 1.2959,
+      "step": 971
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.028000000000001e-05,
+      "loss": 1.1708,
+      "step": 972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.027e-05,
+      "loss": 1.1994,
+      "step": 973
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.026e-05,
+      "loss": 1.4219,
+      "step": 974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.025e-05,
+      "loss": 1.441,
+      "step": 975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.024e-05,
+      "loss": 1.1061,
+      "step": 976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.023000000000001e-05,
+      "loss": 0.3593,
+      "step": 977
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.022e-05,
+      "loss": 0.0633,
+      "step": 978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.021e-05,
+      "loss": 0.6212,
+      "step": 979
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.020000000000001e-05,
+      "loss": 0.4422,
+      "step": 980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.019e-05,
+      "loss": 0.2878,
+      "step": 981
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.018000000000001e-05,
+      "loss": 0.9917,
+      "step": 982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.017e-05,
+      "loss": 1.0038,
+      "step": 983
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.016e-05,
+      "loss": 1.2868,
+      "step": 984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.015e-05,
+      "loss": 1.4016,
+      "step": 985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.014e-05,
+      "loss": 1.2808,
+      "step": 986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.013e-05,
+      "loss": 1.6921,
+      "step": 987
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.012e-05,
+      "loss": 1.1417,
+      "step": 988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.011e-05,
+      "loss": 0.2865,
+      "step": 989
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.010000000000001e-05,
+      "loss": 0.0186,
+      "step": 990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.009000000000001e-05,
+      "loss": 0.2466,
+      "step": 991
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.008e-05,
+      "loss": 0.148,
+      "step": 992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.007e-05,
+      "loss": 0.0572,
+      "step": 993
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.006e-05,
+      "loss": 0.6698,
+      "step": 994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.005000000000001e-05,
+      "loss": 1.0906,
+      "step": 995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.004e-05,
+      "loss": 1.3354,
+      "step": 996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.003e-05,
+      "loss": 1.5321,
+      "step": 997
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.002000000000001e-05,
+      "loss": 1.3122,
+      "step": 998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.001e-05,
+      "loss": 1.3238,
+      "step": 999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-05,
+      "loss": 1.5962,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.999000000000001e-05,
+      "loss": 1.3743,
+      "step": 1001
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.998e-05,
+      "loss": 0.3065,
+      "step": 1002
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.997000000000001e-05,
+      "loss": 0.3175,
+      "step": 1003
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.996e-05,
+      "loss": 0.1372,
+      "step": 1004
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.995e-05,
+      "loss": 0.8372,
+      "step": 1005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.994e-05,
+      "loss": 1.1121,
+      "step": 1006
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.993e-05,
+      "loss": 0.8716,
+      "step": 1007
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.992e-05,
+      "loss": 1.2302,
+      "step": 1008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.991000000000001e-05,
+      "loss": 1.5518,
+      "step": 1009
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.99e-05,
+      "loss": 1.3167,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.989000000000001e-05,
+      "loss": 0.8774,
+      "step": 1011
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.988000000000001e-05,
+      "loss": 1.4511,
+      "step": 1012
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.987e-05,
+      "loss": 1.8085,
+      "step": 1013
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.986e-05,
+      "loss": 0.5744,
+      "step": 1014
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.985e-05,
+      "loss": 0.3749,
+      "step": 1015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.984000000000001e-05,
+      "loss": 0.2141,
+      "step": 1016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.983e-05,
+      "loss": 0.2447,
+      "step": 1017
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.982e-05,
+      "loss": 0.7302,
+      "step": 1018
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.981000000000001e-05,
+      "loss": 0.7106,
+      "step": 1019
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.98e-05,
+      "loss": 0.7254,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.979e-05,
+      "loss": 0.9906,
+      "step": 1021
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.978000000000001e-05,
+      "loss": 1.139,
+      "step": 1022
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.977000000000002e-05,
+      "loss": 1.6114,
+      "step": 1023
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.976e-05,
+      "loss": 1.4742,
+      "step": 1024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.975e-05,
+      "loss": 1.5135,
+      "step": 1025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974e-05,
+      "loss": 0.9094,
+      "step": 1026
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.973e-05,
+      "loss": 0.3615,
+      "step": 1027
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.972e-05,
+      "loss": 0.6536,
+      "step": 1028
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.971e-05,
+      "loss": 0.489,
+      "step": 1029
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.970000000000001e-05,
+      "loss": 0.7336,
+      "step": 1030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.969e-05,
+      "loss": 1.3275,
+      "step": 1031
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.968000000000001e-05,
+      "loss": 1.1223,
+      "step": 1032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.967000000000001e-05,
+      "loss": 1.2549,
+      "step": 1033
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.966e-05,
+      "loss": 1.1483,
+      "step": 1034
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.965e-05,
+      "loss": 1.0439,
+      "step": 1035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.964e-05,
+      "loss": 1.2182,
+      "step": 1036
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.963000000000001e-05,
+      "loss": 0.904,
+      "step": 1037
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.962e-05,
+      "loss": 1.1297,
+      "step": 1038
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.961e-05,
+      "loss": 0.5224,
+      "step": 1039
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.960000000000001e-05,
+      "loss": 0.1288,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.959e-05,
+      "loss": 0.3041,
+      "step": 1041
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.958e-05,
+      "loss": 0.103,
+      "step": 1042
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.957000000000001e-05,
+      "loss": 0.3827,
+      "step": 1043
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.956e-05,
+      "loss": 0.4947,
+      "step": 1044
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.955e-05,
+      "loss": 1.2692,
+      "step": 1045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.954e-05,
+      "loss": 1.1734,
+      "step": 1046
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.953e-05,
+      "loss": 0.9327,
+      "step": 1047
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.952000000000001e-05,
+      "loss": 1.1212,
+      "step": 1048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.951e-05,
+      "loss": 1.5929,
+      "step": 1049
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.950000000000001e-05,
+      "loss": 1.8264,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.949000000000001e-05,
+      "loss": 1.2642,
+      "step": 1051
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.948e-05,
+      "loss": 0.1243,
+      "step": 1052
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.947000000000001e-05,
+      "loss": 0.1644,
+      "step": 1053
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.946e-05,
+      "loss": 0.1667,
+      "step": 1054
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.945e-05,
+      "loss": 1.0955,
+      "step": 1055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.944e-05,
+      "loss": 0.7578,
+      "step": 1056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.943e-05,
+      "loss": 0.9585,
+      "step": 1057
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.942000000000001e-05,
+      "loss": 1.088,
+      "step": 1058
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.941e-05,
+      "loss": 1.1316,
+      "step": 1059
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.94e-05,
+      "loss": 1.2026,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.939000000000001e-05,
+      "loss": 1.2855,
+      "step": 1061
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.938e-05,
+      "loss": 0.816,
+      "step": 1062
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.937e-05,
+      "loss": 1.6038,
+      "step": 1063
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.936e-05,
+      "loss": 0.9892,
+      "step": 1064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.935e-05,
+      "loss": 0.6237,
+      "step": 1065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.934e-05,
+      "loss": 0.1733,
+      "step": 1066
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.933e-05,
+      "loss": 0.0291,
+      "step": 1067
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.932e-05,
+      "loss": 0.3849,
+      "step": 1068
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.931000000000001e-05,
+      "loss": 1.0001,
+      "step": 1069
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.93e-05,
+      "loss": 0.9772,
+      "step": 1070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.929000000000001e-05,
+      "loss": 1.561,
+      "step": 1071
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.928000000000001e-05,
+      "loss": 1.4005,
+      "step": 1072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.927e-05,
+      "loss": 1.0788,
+      "step": 1073
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.926e-05,
+      "loss": 1.2336,
+      "step": 1074
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.925e-05,
+      "loss": 0.5944,
+      "step": 1075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.924e-05,
+      "loss": 1.4664,
+      "step": 1076
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.923e-05,
+      "loss": 0.0059,
+      "step": 1077
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.922e-05,
+      "loss": 0.0363,
+      "step": 1078
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.921000000000001e-05,
+      "loss": 0.0005,
+      "step": 1079
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.92e-05,
+      "loss": 0.4477,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.919e-05,
+      "loss": 0.7463,
+      "step": 1081
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.918000000000001e-05,
+      "loss": 1.4377,
+      "step": 1082
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.917000000000002e-05,
+      "loss": 0.9643,
+      "step": 1083
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.916e-05,
+      "loss": 1.3099,
+      "step": 1084
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.915e-05,
+      "loss": 1.2567,
+      "step": 1085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.914e-05,
+      "loss": 1.4699,
+      "step": 1086
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.913000000000001e-05,
+      "loss": 1.2641,
+      "step": 1087
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.912e-05,
+      "loss": 1.195,
+      "step": 1088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.911e-05,
+      "loss": 0.7761,
+      "step": 1089
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.910000000000001e-05,
+      "loss": 0.078,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.909e-05,
+      "loss": 0.1632,
+      "step": 1091
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.908000000000001e-05,
+      "loss": 0.446,
+      "step": 1092
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.907000000000001e-05,
+      "loss": 0.2036,
+      "step": 1093
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.906e-05,
+      "loss": 0.7195,
+      "step": 1094
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.905e-05,
+      "loss": 1.7357,
+      "step": 1095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.904e-05,
+      "loss": 1.0187,
+      "step": 1096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.903e-05,
+      "loss": 0.8509,
+      "step": 1097
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.902e-05,
+      "loss": 1.1824,
+      "step": 1098
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.901e-05,
+      "loss": 1.0598,
+      "step": 1099
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 0.7831,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.899e-05,
+      "loss": 0.8415,
+      "step": 1101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.898e-05,
+      "loss": 0.2407,
+      "step": 1102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.897000000000001e-05,
+      "loss": 0.0127,
+      "step": 1103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.896e-05,
+      "loss": 0.0304,
+      "step": 1104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.895e-05,
+      "loss": 0.4914,
+      "step": 1105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.894e-05,
+      "loss": 0.8399,
+      "step": 1106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.893e-05,
+      "loss": 1.4545,
+      "step": 1107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.892000000000001e-05,
+      "loss": 1.3075,
+      "step": 1108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.891e-05,
+      "loss": 1.029,
+      "step": 1109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.89e-05,
+      "loss": 1.465,
+      "step": 1110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.889000000000001e-05,
+      "loss": 0.9767,
+      "step": 1111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.888e-05,
+      "loss": 1.1518,
+      "step": 1112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.887000000000001e-05,
+      "loss": 2.0227,
+      "step": 1113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.886e-05,
+      "loss": 0.3723,
+      "step": 1114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.885e-05,
+      "loss": 0.1975,
+      "step": 1115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.884e-05,
+      "loss": 0.0322,
+      "step": 1116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.883e-05,
+      "loss": 0.218,
+      "step": 1117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.882000000000001e-05,
+      "loss": 0.8295,
+      "step": 1118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.881e-05,
+      "loss": 1.3967,
+      "step": 1119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.88e-05,
+      "loss": 1.5698,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.879000000000001e-05,
+      "loss": 1.5262,
+      "step": 1121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.878000000000001e-05,
+      "loss": 1.1344,
+      "step": 1122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.877e-05,
+      "loss": 0.9188,
+      "step": 1123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.876e-05,
+      "loss": 1.1241,
+      "step": 1124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.875e-05,
+      "loss": 0.9393,
+      "step": 1125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.874000000000001e-05,
+      "loss": 1.4001,
+      "step": 1126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.873e-05,
+      "loss": 0.7445,
+      "step": 1127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.872e-05,
+      "loss": 0.1727,
+      "step": 1128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.871000000000001e-05,
+      "loss": 0.0847,
+      "step": 1129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.87e-05,
+      "loss": 0.1196,
+      "step": 1130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.869e-05,
+      "loss": 0.8651,
+      "step": 1131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.868000000000001e-05,
+      "loss": 0.9129,
+      "step": 1132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.867e-05,
+      "loss": 1.2805,
+      "step": 1133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.866000000000001e-05,
+      "loss": 1.1508,
+      "step": 1134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.865e-05,
+      "loss": 1.378,
+      "step": 1135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.864e-05,
+      "loss": 1.2179,
+      "step": 1136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.863e-05,
+      "loss": 1.3126,
+      "step": 1137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.862e-05,
+      "loss": 1.2409,
+      "step": 1138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.861000000000001e-05,
+      "loss": 0.8703,
+      "step": 1139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.86e-05,
+      "loss": 0.2384,
+      "step": 1140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.859e-05,
+      "loss": 0.1896,
+      "step": 1141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.858000000000001e-05,
+      "loss": 0.2869,
+      "step": 1142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.857000000000001e-05,
+      "loss": 0.8127,
+      "step": 1143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.856e-05,
+      "loss": 1.0832,
+      "step": 1144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.855e-05,
+      "loss": 1.5058,
+      "step": 1145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.854e-05,
+      "loss": 1.3413,
+      "step": 1146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.853000000000001e-05,
+      "loss": 0.7954,
+      "step": 1147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.852e-05,
+      "loss": 1.658,
+      "step": 1148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.851e-05,
+      "loss": 1.0567,
+      "step": 1149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.850000000000001e-05,
+      "loss": 0.8692,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.849e-05,
+      "loss": 1.4261,
+      "step": 1151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.848e-05,
+      "loss": 0.3299,
+      "step": 1152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.847000000000001e-05,
+      "loss": 0.0105,
+      "step": 1153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846e-05,
+      "loss": 0.0346,
+      "step": 1154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.845e-05,
+      "loss": 0.4738,
+      "step": 1155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.844e-05,
+      "loss": 1.0952,
+      "step": 1156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.843e-05,
+      "loss": 0.7652,
+      "step": 1157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.842e-05,
+      "loss": 1.1775,
+      "step": 1158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.841e-05,
+      "loss": 0.8848,
+      "step": 1159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.840000000000001e-05,
+      "loss": 1.1689,
+      "step": 1160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.839000000000001e-05,
+      "loss": 1.1202,
+      "step": 1161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.838e-05,
+      "loss": 1.7578,
+      "step": 1162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.837000000000001e-05,
+      "loss": 1.6412,
+      "step": 1163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.836000000000001e-05,
+      "loss": 0.8346,
+      "step": 1164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.834999999999999e-05,
+      "loss": 0.1131,
+      "step": 1165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.834e-05,
+      "loss": 0.0505,
+      "step": 1166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.833e-05,
+      "loss": 0.8583,
+      "step": 1167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.832000000000001e-05,
+      "loss": 1.1103,
+      "step": 1168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.831e-05,
+      "loss": 1.2906,
+      "step": 1169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.83e-05,
+      "loss": 1.4,
+      "step": 1170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.829000000000001e-05,
+      "loss": 1.0,
+      "step": 1171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.828e-05,
+      "loss": 1.406,
+      "step": 1172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.827000000000001e-05,
+      "loss": 1.1522,
+      "step": 1173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.826000000000001e-05,
+      "loss": 1.3158,
+      "step": 1174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.825e-05,
+      "loss": 0.8481,
+      "step": 1175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.824e-05,
+      "loss": 1.4474,
+      "step": 1176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.823e-05,
+      "loss": 0.4434,
+      "step": 1177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.822e-05,
+      "loss": 0.4135,
+      "step": 1178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.821e-05,
+      "loss": 0.1493,
+      "step": 1179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.82e-05,
+      "loss": 0.4603,
+      "step": 1180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.819000000000001e-05,
+      "loss": 0.8852,
+      "step": 1181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.818000000000001e-05,
+      "loss": 0.9552,
+      "step": 1182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.817e-05,
+      "loss": 0.9208,
+      "step": 1183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.816000000000001e-05,
+      "loss": 1.2265,
+      "step": 1184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.815e-05,
+      "loss": 0.9191,
+      "step": 1185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.814e-05,
+      "loss": 1.1117,
+      "step": 1186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.813e-05,
+      "loss": 1.1378,
+      "step": 1187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.812e-05,
+      "loss": 1.8559,
+      "step": 1188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.811000000000001e-05,
+      "loss": 1.1104,
+      "step": 1189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.81e-05,
+      "loss": 0.6519,
+      "step": 1190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.809e-05,
+      "loss": 0.594,
+      "step": 1191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.808000000000001e-05,
+      "loss": 0.0824,
+      "step": 1192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.807e-05,
+      "loss": 1.0881,
+      "step": 1193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.806000000000001e-05,
+      "loss": 1.1528,
+      "step": 1194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.805e-05,
+      "loss": 1.4713,
+      "step": 1195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.804e-05,
+      "loss": 1.4873,
+      "step": 1196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.803e-05,
+      "loss": 0.6966,
+      "step": 1197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.802e-05,
+      "loss": 1.2489,
+      "step": 1198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.801e-05,
+      "loss": 1.2703,
+      "step": 1199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.2517,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.799e-05,
+      "loss": 1.3397,
+      "step": 1201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.798000000000001e-05,
+      "loss": 0.385,
+      "step": 1202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.797000000000001e-05,
+      "loss": 0.8203,
+      "step": 1203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.796e-05,
+      "loss": 0.1715,
+      "step": 1204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.795e-05,
+      "loss": 0.0496,
+      "step": 1205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.794e-05,
+      "loss": 0.5884,
+      "step": 1206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.793000000000001e-05,
+      "loss": 0.9601,
+      "step": 1207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.792e-05,
+      "loss": 1.3504,
+      "step": 1208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.791e-05,
+      "loss": 1.0078,
+      "step": 1209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.790000000000001e-05,
+      "loss": 1.1538,
+      "step": 1210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.789e-05,
+      "loss": 1.3192,
+      "step": 1211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.788e-05,
+      "loss": 1.0488,
+      "step": 1212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.787000000000001e-05,
+      "loss": 1.1575,
+      "step": 1213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.786e-05,
+      "loss": 0.8575,
+      "step": 1214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.785e-05,
+      "loss": 0.2729,
+      "step": 1215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.784e-05,
+      "loss": 0.108,
+      "step": 1216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.783e-05,
+      "loss": 0.0038,
+      "step": 1217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.782e-05,
+      "loss": 0.3868,
+      "step": 1218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.781e-05,
+      "loss": 1.1241,
+      "step": 1219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.78e-05,
+      "loss": 1.7387,
+      "step": 1220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.779000000000001e-05,
+      "loss": 1.2069,
+      "step": 1221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.778e-05,
+      "loss": 1.4388,
+      "step": 1222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.777000000000001e-05,
+      "loss": 1.1658,
+      "step": 1223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.776000000000001e-05,
+      "loss": 1.2188,
+      "step": 1224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.775e-05,
+      "loss": 0.9964,
+      "step": 1225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.774e-05,
+      "loss": 1.4439,
+      "step": 1226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.773e-05,
+      "loss": 0.3785,
+      "step": 1227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.772000000000001e-05,
+      "loss": 0.4296,
+      "step": 1228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.771e-05,
+      "loss": 0.2896,
+      "step": 1229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.77e-05,
+      "loss": 0.2888,
+      "step": 1230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.769000000000001e-05,
+      "loss": 0.6962,
+      "step": 1231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.768e-05,
+      "loss": 0.8002,
+      "step": 1232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.767e-05,
+      "loss": 1.1831,
+      "step": 1233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.766000000000001e-05,
+      "loss": 0.9699,
+      "step": 1234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.765e-05,
+      "loss": 0.9673,
+      "step": 1235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.764e-05,
+      "loss": 0.8987,
+      "step": 1236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.763e-05,
+      "loss": 1.4727,
+      "step": 1237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.762e-05,
+      "loss": 1.5307,
+      "step": 1238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.761000000000001e-05,
+      "loss": 0.8061,
+      "step": 1239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.76e-05,
+      "loss": 0.1268,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.759e-05,
+      "loss": 0.0511,
+      "step": 1241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.758000000000001e-05,
+      "loss": 0.2462,
+      "step": 1242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.757e-05,
+      "loss": 0.8591,
+      "step": 1243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.756000000000001e-05,
+      "loss": 1.3775,
+      "step": 1244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.755e-05,
+      "loss": 0.9655,
+      "step": 1245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.754e-05,
+      "loss": 1.2537,
+      "step": 1246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.753e-05,
+      "loss": 1.2413,
+      "step": 1247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.752e-05,
+      "loss": 1.0898,
+      "step": 1248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.751000000000001e-05,
+      "loss": 0.9601,
+      "step": 1249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.75e-05,
+      "loss": 1.5517,
+      "step": 1250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.749e-05,
+      "loss": 1.1352,
+      "step": 1251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.748000000000001e-05,
+      "loss": 0.1444,
+      "step": 1252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.747e-05,
+      "loss": 0.3045,
+      "step": 1253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.746e-05,
+      "loss": 0.1614,
+      "step": 1254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.745000000000001e-05,
+      "loss": 0.7411,
+      "step": 1255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.744e-05,
+      "loss": 0.8515,
+      "step": 1256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.743e-05,
+      "loss": 1.2122,
+      "step": 1257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.742e-05,
+      "loss": 0.9876,
+      "step": 1258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.741e-05,
+      "loss": 1.1902,
+      "step": 1259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.740000000000001e-05,
+      "loss": 1.2368,
+      "step": 1260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.739e-05,
+      "loss": 1.3146,
+      "step": 1261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.738000000000001e-05,
+      "loss": 0.7156,
+      "step": 1262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.737000000000001e-05,
+      "loss": 1.5073,
+      "step": 1263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.736e-05,
+      "loss": 1.0137,
+      "step": 1264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.735000000000001e-05,
+      "loss": 0.2386,
+      "step": 1265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.734e-05,
+      "loss": 0.1624,
+      "step": 1266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.733e-05,
+      "loss": 0.2813,
+      "step": 1267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.732e-05,
+      "loss": 0.7383,
+      "step": 1268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.731e-05,
+      "loss": 1.1509,
+      "step": 1269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.730000000000001e-05,
+      "loss": 1.3669,
+      "step": 1270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.729e-05,
+      "loss": 1.2288,
+      "step": 1271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.728e-05,
+      "loss": 1.1819,
+      "step": 1272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.727000000000001e-05,
+      "loss": 1.3884,
+      "step": 1273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.726000000000001e-05,
+      "loss": 1.8685,
+      "step": 1274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.725e-05,
+      "loss": 1.0052,
+      "step": 1275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.724e-05,
+      "loss": 1.4146,
+      "step": 1276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.723e-05,
+      "loss": 0.787,
+      "step": 1277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.722e-05,
+      "loss": 0.4124,
+      "step": 1278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.721e-05,
+      "loss": 0.1604,
+      "step": 1279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.72e-05,
+      "loss": 0.4224,
+      "step": 1280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.719000000000001e-05,
+      "loss": 1.2442,
+      "step": 1281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.718e-05,
+      "loss": 1.1807,
+      "step": 1282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.717000000000001e-05,
+      "loss": 1.1931,
+      "step": 1283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.716000000000001e-05,
+      "loss": 1.3501,
+      "step": 1284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.715e-05,
+      "loss": 1.4041,
+      "step": 1285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.714e-05,
+      "loss": 1.5015,
+      "step": 1286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.713e-05,
+      "loss": 0.7208,
+      "step": 1287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.712e-05,
+      "loss": 1.553,
+      "step": 1288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.711e-05,
+      "loss": 1.0215,
+      "step": 1289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.71e-05,
+      "loss": 0.3598,
+      "step": 1290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.709000000000001e-05,
+      "loss": 0.3048,
+      "step": 1291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.708e-05,
+      "loss": 0.1577,
+      "step": 1292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.707e-05,
+      "loss": 0.2632,
+      "step": 1293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.706000000000001e-05,
+      "loss": 1.14,
+      "step": 1294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.705000000000002e-05,
+      "loss": 1.2045,
+      "step": 1295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.704e-05,
+      "loss": 1.0461,
+      "step": 1296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.703e-05,
+      "loss": 1.3678,
+      "step": 1297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.702e-05,
+      "loss": 1.188,
+      "step": 1298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.701000000000001e-05,
+      "loss": 1.1636,
+      "step": 1299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.7e-05,
+      "loss": 1.3342,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.699e-05,
+      "loss": 0.927,
+      "step": 1301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.698000000000001e-05,
+      "loss": 0.0994,
+      "step": 1302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.697e-05,
+      "loss": 0.1364,
+      "step": 1303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.696000000000001e-05,
+      "loss": 0.1937,
+      "step": 1304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.695000000000001e-05,
+      "loss": 1.1216,
+      "step": 1305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.694e-05,
+      "loss": 1.4041,
+      "step": 1306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.693e-05,
+      "loss": 0.9776,
+      "step": 1307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.692e-05,
+      "loss": 0.9204,
+      "step": 1308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.691e-05,
+      "loss": 1.3731,
+      "step": 1309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.69e-05,
+      "loss": 0.9801,
+      "step": 1310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.689e-05,
+      "loss": 1.1647,
+      "step": 1311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.688000000000001e-05,
+      "loss": 1.418,
+      "step": 1312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.687000000000001e-05,
+      "loss": 1.0965,
+      "step": 1313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.686e-05,
+      "loss": 1.0709,
+      "step": 1314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.685000000000001e-05,
+      "loss": 0.1448,
+      "step": 1315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.684e-05,
+      "loss": 0.1765,
+      "step": 1316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.683e-05,
+      "loss": 0.2338,
+      "step": 1317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.682e-05,
+      "loss": 0.9911,
+      "step": 1318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.681e-05,
+      "loss": 1.1046,
+      "step": 1319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.680000000000001e-05,
+      "loss": 1.3755,
+      "step": 1320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.679e-05,
+      "loss": 0.7861,
+      "step": 1321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.678e-05,
+      "loss": 0.9843,
+      "step": 1322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.677000000000001e-05,
+      "loss": 0.9916,
+      "step": 1323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.676e-05,
+      "loss": 1.5365,
+      "step": 1324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.675000000000001e-05,
+      "loss": 1.3862,
+      "step": 1325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.674e-05,
+      "loss": 1.2165,
+      "step": 1326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.673e-05,
+      "loss": 0.4519,
+      "step": 1327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.672e-05,
+      "loss": 0.1421,
+      "step": 1328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.671e-05,
+      "loss": 0.3774,
+      "step": 1329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.67e-05,
+      "loss": 0.145,
+      "step": 1330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.669e-05,
+      "loss": 1.0922,
+      "step": 1331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.668e-05,
+      "loss": 1.4475,
+      "step": 1332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.667000000000001e-05,
+      "loss": 1.3426,
+      "step": 1333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.666000000000001e-05,
+      "loss": 1.2098,
+      "step": 1334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.665e-05,
+      "loss": 1.2204,
+      "step": 1335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.664e-05,
+      "loss": 1.002,
+      "step": 1336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.663e-05,
+      "loss": 0.8314,
+      "step": 1337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.662000000000001e-05,
+      "loss": 1.1246,
+      "step": 1338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.661e-05,
+      "loss": 0.4483,
+      "step": 1339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.66e-05,
+      "loss": 0.1804,
+      "step": 1340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.659000000000001e-05,
+      "loss": 0.0034,
+      "step": 1341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.658e-05,
+      "loss": 0.1017,
+      "step": 1342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.657e-05,
+      "loss": 0.4526,
+      "step": 1343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.656000000000001e-05,
+      "loss": 0.9501,
+      "step": 1344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.655e-05,
+      "loss": 1.3009,
+      "step": 1345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.654e-05,
+      "loss": 1.4697,
+      "step": 1346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.653e-05,
+      "loss": 1.3343,
+      "step": 1347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.652e-05,
+      "loss": 0.7259,
+      "step": 1348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.651e-05,
+      "loss": 1.1162,
+      "step": 1349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.65e-05,
+      "loss": 1.5277,
+      "step": 1350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.649000000000001e-05,
+      "loss": 1.4873,
+      "step": 1351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.648e-05,
+      "loss": 0.4781,
+      "step": 1352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.647e-05,
+      "loss": 0.054,
+      "step": 1353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.646000000000001e-05,
+      "loss": 0.1653,
+      "step": 1354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.645000000000001e-05,
+      "loss": 0.0642,
+      "step": 1355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.643999999999999e-05,
+      "loss": 0.4815,
+      "step": 1356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.643e-05,
+      "loss": 1.1678,
+      "step": 1357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.642e-05,
+      "loss": 1.2819,
+      "step": 1358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.641000000000001e-05,
+      "loss": 0.738,
+      "step": 1359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.64e-05,
+      "loss": 1.0799,
+      "step": 1360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.639e-05,
+      "loss": 1.5521,
+      "step": 1361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.638000000000001e-05,
+      "loss": 1.0022,
+      "step": 1362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.637e-05,
+      "loss": 1.5617,
+      "step": 1363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.636e-05,
+      "loss": 0.5654,
+      "step": 1364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.635000000000001e-05,
+      "loss": 0.1956,
+      "step": 1365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.634e-05,
+      "loss": 0.1727,
+      "step": 1366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.633e-05,
+      "loss": 0.0258,
+      "step": 1367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.632e-05,
+      "loss": 0.1282,
+      "step": 1368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.631e-05,
+      "loss": 0.1602,
+      "step": 1369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.63e-05,
+      "loss": 1.0248,
+      "step": 1370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.629e-05,
+      "loss": 1.1997,
+      "step": 1371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.628000000000001e-05,
+      "loss": 0.8434,
+      "step": 1372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.627000000000001e-05,
+      "loss": 1.2444,
+      "step": 1373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.626e-05,
+      "loss": 1.0582,
+      "step": 1374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.625000000000001e-05,
+      "loss": 1.4697,
+      "step": 1375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.624000000000001e-05,
+      "loss": 1.1599,
+      "step": 1376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.623e-05,
+      "loss": 0.2253,
+      "step": 1377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.622e-05,
+      "loss": 0.0817,
+      "step": 1378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.621e-05,
+      "loss": 0.1506,
+      "step": 1379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.620000000000001e-05,
+      "loss": 0.3451,
+      "step": 1380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.619e-05,
+      "loss": 0.4701,
+      "step": 1381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.618e-05,
+      "loss": 1.2811,
+      "step": 1382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.617000000000001e-05,
+      "loss": 1.1709,
+      "step": 1383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.616e-05,
+      "loss": 1.272,
+      "step": 1384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.615000000000001e-05,
+      "loss": 0.7617,
+      "step": 1385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.614000000000001e-05,
+      "loss": 1.0422,
+      "step": 1386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.613e-05,
+      "loss": 1.1104,
+      "step": 1387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.612e-05,
+      "loss": 1.3263,
+      "step": 1388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.611e-05,
+      "loss": 0.5128,
+      "step": 1389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.61e-05,
+      "loss": 0.3197,
+      "step": 1390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.609e-05,
+      "loss": 0.3354,
+      "step": 1391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.608e-05,
+      "loss": 0.0038,
+      "step": 1392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.607000000000001e-05,
+      "loss": 0.1645,
+      "step": 1393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.606000000000001e-05,
+      "loss": 0.7544,
+      "step": 1394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.605e-05,
+      "loss": 1.2847,
+      "step": 1395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.604000000000001e-05,
+      "loss": 1.3424,
+      "step": 1396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.603e-05,
+      "loss": 1.0528,
+      "step": 1397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.602e-05,
+      "loss": 0.9796,
+      "step": 1398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.601e-05,
+      "loss": 1.6801,
+      "step": 1399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.6e-05,
+      "loss": 1.5361,
+      "step": 1400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.599000000000001e-05,
+      "loss": 1.1128,
+      "step": 1401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.598e-05,
+      "loss": 0.3828,
+      "step": 1402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.597e-05,
+      "loss": 0.1529,
+      "step": 1403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.596000000000001e-05,
+      "loss": 0.0115,
+      "step": 1404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.595e-05,
+      "loss": 1.1863,
+      "step": 1405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.594000000000001e-05,
+      "loss": 1.1756,
+      "step": 1406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.593e-05,
+      "loss": 0.9941,
+      "step": 1407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.592e-05,
+      "loss": 1.0449,
+      "step": 1408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.591e-05,
+      "loss": 1.3395,
+      "step": 1409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.59e-05,
+      "loss": 1.0528,
+      "step": 1410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.589e-05,
+      "loss": 0.8828,
+      "step": 1411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.588000000000001e-05,
+      "loss": 1.9049,
+      "step": 1412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.587e-05,
+      "loss": 1.1449,
+      "step": 1413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.586000000000001e-05,
+      "loss": 0.8673,
+      "step": 1414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.585000000000001e-05,
+      "loss": 0.2925,
+      "step": 1415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.584e-05,
+      "loss": 0.1374,
+      "step": 1416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.583e-05,
+      "loss": 0.0817,
+      "step": 1417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.582e-05,
+      "loss": 0.3979,
+      "step": 1418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.581000000000001e-05,
+      "loss": 0.2616,
+      "step": 1419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.58e-05,
+      "loss": 1.3779,
+      "step": 1420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.579e-05,
+      "loss": 1.0147,
+      "step": 1421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.578000000000001e-05,
+      "loss": 1.4001,
+      "step": 1422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.577e-05,
+      "loss": 0.8834,
+      "step": 1423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.576e-05,
+      "loss": 1.0444,
+      "step": 1424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.575000000000001e-05,
+      "loss": 1.3286,
+      "step": 1425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.574000000000002e-05,
+      "loss": 1.1586,
+      "step": 1426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.573e-05,
+      "loss": 0.0231,
+      "step": 1427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.572e-05,
+      "loss": 0.0286,
+      "step": 1428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.571e-05,
+      "loss": 0.3597,
+      "step": 1429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.57e-05,
+      "loss": 0.6224,
+      "step": 1430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.569e-05,
+      "loss": 1.3727,
+      "step": 1431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.568e-05,
+      "loss": 1.0839,
+      "step": 1432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.567000000000001e-05,
+      "loss": 1.6172,
+      "step": 1433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.566e-05,
+      "loss": 0.9535,
+      "step": 1434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.565000000000001e-05,
+      "loss": 1.0071,
+      "step": 1435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.564000000000001e-05,
+      "loss": 1.7271,
+      "step": 1436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.563e-05,
+      "loss": 1.0929,
+      "step": 1437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.562e-05,
+      "loss": 1.0101,
+      "step": 1438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.561e-05,
+      "loss": 0.9694,
+      "step": 1439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.560000000000001e-05,
+      "loss": 0.279,
+      "step": 1440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.559e-05,
+      "loss": 0.0728,
+      "step": 1441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.558e-05,
+      "loss": 0.5847,
+      "step": 1442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.557000000000001e-05,
+      "loss": 1.0482,
+      "step": 1443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.556e-05,
+      "loss": 1.2168,
+      "step": 1444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.555e-05,
+      "loss": 0.9507,
+      "step": 1445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.554000000000001e-05,
+      "loss": 1.1062,
+      "step": 1446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.553e-05,
+      "loss": 1.5942,
+      "step": 1447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.552e-05,
+      "loss": 1.6499,
+      "step": 1448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.551e-05,
+      "loss": 1.2806,
+      "step": 1449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.55e-05,
+      "loss": 1.5793,
+      "step": 1450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.549000000000001e-05,
+      "loss": 1.2023,
+      "step": 1451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.548e-05,
+      "loss": 0.1926,
+      "step": 1452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.547e-05,
+      "loss": 0.0564,
+      "step": 1453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.546000000000001e-05,
+      "loss": 0.2101,
+      "step": 1454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.545e-05,
+      "loss": 0.7195,
+      "step": 1455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.544000000000001e-05,
+      "loss": 0.4535,
+      "step": 1456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.543e-05,
+      "loss": 1.2486,
+      "step": 1457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.542e-05,
+      "loss": 1.2086,
+      "step": 1458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.541e-05,
+      "loss": 0.9602,
+      "step": 1459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.54e-05,
+      "loss": 1.0585,
+      "step": 1460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.539000000000001e-05,
+      "loss": 1.3707,
+      "step": 1461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.538e-05,
+      "loss": 1.1945,
+      "step": 1462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.537e-05,
+      "loss": 0.88,
+      "step": 1463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.536000000000001e-05,
+      "loss": 0.8441,
+      "step": 1464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.535e-05,
+      "loss": 0.089,
+      "step": 1465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.534e-05,
+      "loss": 0.0283,
+      "step": 1466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.533e-05,
+      "loss": 0.0498,
+      "step": 1467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.532e-05,
+      "loss": 0.8122,
+      "step": 1468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.531e-05,
+      "loss": 1.3065,
+      "step": 1469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.53e-05,
+      "loss": 1.2429,
+      "step": 1470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.529e-05,
+      "loss": 1.0318,
+      "step": 1471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.528000000000001e-05,
+      "loss": 1.269,
+      "step": 1472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.527e-05,
+      "loss": 1.5559,
+      "step": 1473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.526000000000001e-05,
+      "loss": 1.334,
+      "step": 1474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.525000000000001e-05,
+      "loss": 1.4665,
+      "step": 1475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.524e-05,
+      "loss": 1.6995,
+      "step": 1476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.523e-05,
+      "loss": 0.5045,
+      "step": 1477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.522e-05,
+      "loss": 0.0146,
+      "step": 1478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.521e-05,
+      "loss": 0.2258,
+      "step": 1479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.52e-05,
+      "loss": 0.2817,
+      "step": 1480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.519e-05,
+      "loss": 0.7953,
+      "step": 1481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.518000000000001e-05,
+      "loss": 1.1296,
+      "step": 1482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.517e-05,
+      "loss": 1.2529,
+      "step": 1483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.516e-05,
+      "loss": 1.088,
+      "step": 1484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.515000000000001e-05,
+      "loss": 1.0231,
+      "step": 1485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.514000000000001e-05,
+      "loss": 0.892,
+      "step": 1486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.512999999999999e-05,
+      "loss": 1.1239,
+      "step": 1487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.512e-05,
+      "loss": 1.2009,
+      "step": 1488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.511e-05,
+      "loss": 1.326,
+      "step": 1489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.510000000000001e-05,
+      "loss": 0.4104,
+      "step": 1490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.509e-05,
+      "loss": 0.0754,
+      "step": 1491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.508e-05,
+      "loss": 0.0788,
+      "step": 1492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.507000000000001e-05,
+      "loss": 0.3067,
+      "step": 1493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.506e-05,
+      "loss": 0.6052,
+      "step": 1494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.505000000000001e-05,
+      "loss": 1.0136,
+      "step": 1495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.504000000000001e-05,
+      "loss": 1.6004,
+      "step": 1496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.503e-05,
+      "loss": 0.9155,
+      "step": 1497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.502e-05,
+      "loss": 1.2083,
+      "step": 1498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.501e-05,
+      "loss": 1.093,
+      "step": 1499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.5e-05,
+      "loss": 1.1413,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.499e-05,
+      "loss": 1.5221,
+      "step": 1501
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.498e-05,
+      "loss": 0.2851,
+      "step": 1502
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.497000000000001e-05,
+      "loss": 0.038,
+      "step": 1503
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.496e-05,
+      "loss": 0.3393,
+      "step": 1504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.495e-05,
+      "loss": 0.6004,
+      "step": 1505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.494000000000001e-05,
+      "loss": 1.2116,
+      "step": 1506
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.493000000000002e-05,
+      "loss": 1.3457,
+      "step": 1507
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.492e-05,
+      "loss": 1.3142,
+      "step": 1508
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.491e-05,
+      "loss": 1.7683,
+      "step": 1509
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.49e-05,
+      "loss": 0.8813,
+      "step": 1510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.489000000000001e-05,
+      "loss": 0.8974,
+      "step": 1511
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.488e-05,
+      "loss": 1.6439,
+      "step": 1512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.487e-05,
+      "loss": 1.2002,
+      "step": 1513
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.486000000000001e-05,
+      "loss": 0.5976,
+      "step": 1514
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.485e-05,
+      "loss": 0.3634,
+      "step": 1515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.484000000000001e-05,
+      "loss": 0.2557,
+      "step": 1516
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.483000000000001e-05,
+      "loss": 0.1071,
+      "step": 1517
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.482e-05,
+      "loss": 0.3642,
+      "step": 1518
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.481e-05,
+      "loss": 1.2751,
+      "step": 1519
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.48e-05,
+      "loss": 1.264,
+      "step": 1520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.479e-05,
+      "loss": 0.8788,
+      "step": 1521
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.478e-05,
+      "loss": 1.2356,
+      "step": 1522
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.477e-05,
+      "loss": 1.188,
+      "step": 1523
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.476000000000001e-05,
+      "loss": 1.3084,
+      "step": 1524
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.475000000000001e-05,
+      "loss": 1.4012,
+      "step": 1525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.474e-05,
+      "loss": 1.0788,
+      "step": 1526
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.473000000000001e-05,
+      "loss": 0.0982,
+      "step": 1527
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.472e-05,
+      "loss": 0.1413,
+      "step": 1528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.471000000000001e-05,
+      "loss": 0.1359,
+      "step": 1529
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.47e-05,
+      "loss": 0.4631,
+      "step": 1530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.469e-05,
+      "loss": 0.7463,
+      "step": 1531
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.468000000000001e-05,
+      "loss": 0.9792,
+      "step": 1532
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.467e-05,
+      "loss": 1.2364,
+      "step": 1533
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.466e-05,
+      "loss": 1.1983,
+      "step": 1534
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.465000000000001e-05,
+      "loss": 0.9112,
+      "step": 1535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.464e-05,
+      "loss": 0.9289,
+      "step": 1536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.463000000000001e-05,
+      "loss": 1.2416,
+      "step": 1537
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.462e-05,
+      "loss": 0.6739,
+      "step": 1538
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.461e-05,
+      "loss": 0.313,
+      "step": 1539
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.46e-05,
+      "loss": 0.084,
+      "step": 1540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.459e-05,
+      "loss": 0.3425,
+      "step": 1541
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.458e-05,
+      "loss": 0.1295,
+      "step": 1542
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.457e-05,
+      "loss": 0.5796,
+      "step": 1543
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.456e-05,
+      "loss": 0.7461,
+      "step": 1544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.455000000000001e-05,
+      "loss": 1.3738,
+      "step": 1545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.454000000000001e-05,
+      "loss": 1.0414,
+      "step": 1546
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.453e-05,
+      "loss": 1.0139,
+      "step": 1547
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.452e-05,
+      "loss": 1.2025,
+      "step": 1548
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.451e-05,
+      "loss": 0.8901,
+      "step": 1549
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.450000000000001e-05,
+      "loss": 1.6388,
+      "step": 1550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.449e-05,
+      "loss": 1.1283,
+      "step": 1551
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.448e-05,
+      "loss": 0.387,
+      "step": 1552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.447000000000001e-05,
+      "loss": 0.0885,
+      "step": 1553
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.446e-05,
+      "loss": 0.3769,
+      "step": 1554
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.445e-05,
+      "loss": 0.1771,
+      "step": 1555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.444000000000001e-05,
+      "loss": 0.4022,
+      "step": 1556
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.443e-05,
+      "loss": 1.3763,
+      "step": 1557
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.442e-05,
+      "loss": 1.2266,
+      "step": 1558
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.441e-05,
+      "loss": 1.5931,
+      "step": 1559
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.44e-05,
+      "loss": 1.1327,
+      "step": 1560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.439e-05,
+      "loss": 1.2425,
+      "step": 1561
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.438e-05,
+      "loss": 0.905,
+      "step": 1562
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.437000000000001e-05,
+      "loss": 1.0419,
+      "step": 1563
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.436000000000001e-05,
+      "loss": 0.4736,
+      "step": 1564
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.435e-05,
+      "loss": 0.2715,
+      "step": 1565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.434000000000001e-05,
+      "loss": 0.0765,
+      "step": 1566
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.433000000000001e-05,
+      "loss": 0.1286,
+      "step": 1567
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.431999999999999e-05,
+      "loss": 1.0593,
+      "step": 1568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.431e-05,
+      "loss": 1.267,
+      "step": 1569
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.43e-05,
+      "loss": 1.0118,
+      "step": 1570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.429000000000001e-05,
+      "loss": 1.0232,
+      "step": 1571
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.428e-05,
+      "loss": 1.4037,
+      "step": 1572
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.427e-05,
+      "loss": 1.2,
+      "step": 1573
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.426000000000001e-05,
+      "loss": 0.9334,
+      "step": 1574
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.425e-05,
+      "loss": 1.2353,
+      "step": 1575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.424e-05,
+      "loss": 1.1749,
+      "step": 1576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.423000000000001e-05,
+      "loss": 0.2275,
+      "step": 1577
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.422e-05,
+      "loss": 0.3243,
+      "step": 1578
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.421e-05,
+      "loss": 0.3528,
+      "step": 1579
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.42e-05,
+      "loss": 0.3445,
+      "step": 1580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.419e-05,
+      "loss": 0.2481,
+      "step": 1581
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.418e-05,
+      "loss": 0.4332,
+      "step": 1582
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.417e-05,
+      "loss": 1.2285,
+      "step": 1583
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.416000000000001e-05,
+      "loss": 1.1347,
+      "step": 1584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.415000000000001e-05,
+      "loss": 0.8447,
+      "step": 1585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.414e-05,
+      "loss": 0.7755,
+      "step": 1586
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.413000000000001e-05,
+      "loss": 1.5119,
+      "step": 1587
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.412e-05,
+      "loss": 1.3775,
+      "step": 1588
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.411e-05,
+      "loss": 1.278,
+      "step": 1589
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.41e-05,
+      "loss": 0.2279,
+      "step": 1590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.409e-05,
+      "loss": 0.0614,
+      "step": 1591
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.408000000000001e-05,
+      "loss": 0.162,
+      "step": 1592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.407e-05,
+      "loss": 1.1414,
+      "step": 1593
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.406e-05,
+      "loss": 1.2212,
+      "step": 1594
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.405000000000001e-05,
+      "loss": 0.6978,
+      "step": 1595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.404e-05,
+      "loss": 1.2207,
+      "step": 1596
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.403000000000001e-05,
+      "loss": 1.2966,
+      "step": 1597
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.402e-05,
+      "loss": 1.2618,
+      "step": 1598
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.401e-05,
+      "loss": 0.9388,
+      "step": 1599
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.4e-05,
+      "loss": 1.1213,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.399e-05,
+      "loss": 1.147,
+      "step": 1601
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.398e-05,
+      "loss": 0.1723,
+      "step": 1602
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.397000000000001e-05,
+      "loss": 0.0014,
+      "step": 1603
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.396e-05,
+      "loss": 0.153,
+      "step": 1604
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.395000000000001e-05,
+      "loss": 0.4254,
+      "step": 1605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.394000000000001e-05,
+      "loss": 1.0129,
+      "step": 1606
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.393e-05,
+      "loss": 1.6441,
+      "step": 1607
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.392e-05,
+      "loss": 1.4673,
+      "step": 1608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.391e-05,
+      "loss": 1.1559,
+      "step": 1609
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.39e-05,
+      "loss": 1.5358,
+      "step": 1610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.389e-05,
+      "loss": 1.053,
+      "step": 1611
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.388e-05,
+      "loss": 0.9695,
+      "step": 1612
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.387000000000001e-05,
+      "loss": 1.2523,
+      "step": 1613
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.386e-05,
+      "loss": 0.7087,
+      "step": 1614
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.385e-05,
+      "loss": 0.347,
+      "step": 1615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.384000000000001e-05,
+      "loss": 0.2387,
+      "step": 1616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.383e-05,
+      "loss": 0.0339,
+      "step": 1617
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.382e-05,
+      "loss": 0.5403,
+      "step": 1618
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.381e-05,
+      "loss": 1.0061,
+      "step": 1619
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.38e-05,
+      "loss": 1.0266,
+      "step": 1620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.379e-05,
+      "loss": 1.3971,
+      "step": 1621
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.378e-05,
+      "loss": 1.3653,
+      "step": 1622
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.377e-05,
+      "loss": 0.9329,
+      "step": 1623
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.376000000000001e-05,
+      "loss": 1.3532,
+      "step": 1624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.375e-05,
+      "loss": 1.3251,
+      "step": 1625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.374000000000001e-05,
+      "loss": 1.3898,
+      "step": 1626
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.373000000000001e-05,
+      "loss": 0.4197,
+      "step": 1627
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.372e-05,
+      "loss": 0.1309,
+      "step": 1628
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.371e-05,
+      "loss": 0.0136,
+      "step": 1629
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.37e-05,
+      "loss": 0.6229,
+      "step": 1630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.369000000000001e-05,
+      "loss": 0.9985,
+      "step": 1631
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.368e-05,
+      "loss": 0.7686,
+      "step": 1632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.367e-05,
+      "loss": 1.1319,
+      "step": 1633
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.366000000000001e-05,
+      "loss": 1.3611,
+      "step": 1634
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.365e-05,
+      "loss": 1.2136,
+      "step": 1635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.364e-05,
+      "loss": 1.2052,
+      "step": 1636
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.363000000000001e-05,
+      "loss": 1.5283,
+      "step": 1637
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.362000000000002e-05,
+      "loss": 0.9484,
+      "step": 1638
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.361e-05,
+      "loss": 1.0905,
+      "step": 1639
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.36e-05,
+      "loss": 0.1906,
+      "step": 1640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.359e-05,
+      "loss": 0.216,
+      "step": 1641
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.358e-05,
+      "loss": 0.2293,
+      "step": 1642
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.357e-05,
+      "loss": 0.1886,
+      "step": 1643
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.356e-05,
+      "loss": 0.5489,
+      "step": 1644
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.355000000000001e-05,
+      "loss": 1.3487,
+      "step": 1645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.354e-05,
+      "loss": 1.0497,
+      "step": 1646
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.353000000000001e-05,
+      "loss": 1.2508,
+      "step": 1647
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.352000000000001e-05,
+      "loss": 1.531,
+      "step": 1648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.351e-05,
+      "loss": 0.818,
+      "step": 1649
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.35e-05,
+      "loss": 1.8414,
+      "step": 1650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.349e-05,
+      "loss": 0.9576,
+      "step": 1651
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.348000000000001e-05,
+      "loss": 0.152,
+      "step": 1652
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.347e-05,
+      "loss": 0.1315,
+      "step": 1653
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.346e-05,
+      "loss": 0.1627,
+      "step": 1654
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.345000000000001e-05,
+      "loss": 0.1102,
+      "step": 1655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.344e-05,
+      "loss": 0.9908,
+      "step": 1656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.343e-05,
+      "loss": 1.3256,
+      "step": 1657
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.342000000000001e-05,
+      "loss": 1.1806,
+      "step": 1658
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.341e-05,
+      "loss": 1.0491,
+      "step": 1659
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.34e-05,
+      "loss": 0.8469,
+      "step": 1660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.339e-05,
+      "loss": 1.1785,
+      "step": 1661
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.338e-05,
+      "loss": 1.6718,
+      "step": 1662
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.337000000000001e-05,
+      "loss": 1.2276,
+      "step": 1663
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.336e-05,
+      "loss": 0.734,
+      "step": 1664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.335e-05,
+      "loss": 0.144,
+      "step": 1665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.334000000000001e-05,
+      "loss": 0.1523,
+      "step": 1666
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.333e-05,
+      "loss": 0.2861,
+      "step": 1667
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.332000000000001e-05,
+      "loss": 1.01,
+      "step": 1668
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.331e-05,
+      "loss": 1.1334,
+      "step": 1669
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.33e-05,
+      "loss": 1.3685,
+      "step": 1670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.329e-05,
+      "loss": 0.9891,
+      "step": 1671
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.328e-05,
+      "loss": 1.2639,
+      "step": 1672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.327000000000001e-05,
+      "loss": 1.2786,
+      "step": 1673
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.326e-05,
+      "loss": 1.237,
+      "step": 1674
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.325e-05,
+      "loss": 0.8028,
+      "step": 1675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.324000000000001e-05,
+      "loss": 1.0331,
+      "step": 1676
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.323000000000001e-05,
+      "loss": 0.2437,
+      "step": 1677
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.322e-05,
+      "loss": 0.3843,
+      "step": 1678
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.321e-05,
+      "loss": 0.3339,
+      "step": 1679
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.32e-05,
+      "loss": 0.3027,
+      "step": 1680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.319e-05,
+      "loss": 0.9179,
+      "step": 1681
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.318e-05,
+      "loss": 1.0071,
+      "step": 1682
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.317e-05,
+      "loss": 1.1878,
+      "step": 1683
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.316000000000001e-05,
+      "loss": 1.24,
+      "step": 1684
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.315e-05,
+      "loss": 1.443,
+      "step": 1685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.314000000000001e-05,
+      "loss": 1.1622,
+      "step": 1686
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.313000000000001e-05,
+      "loss": 1.4388,
+      "step": 1687
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.312e-05,
+      "loss": 1.979,
+      "step": 1688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.311e-05,
+      "loss": 1.1205,
+      "step": 1689
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.31e-05,
+      "loss": 0.2712,
+      "step": 1690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.309e-05,
+      "loss": 0.1602,
+      "step": 1691
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.308e-05,
+      "loss": 0.2223,
+      "step": 1692
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.307e-05,
+      "loss": 0.699,
+      "step": 1693
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.306000000000001e-05,
+      "loss": 0.982,
+      "step": 1694
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.305e-05,
+      "loss": 0.753,
+      "step": 1695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.304e-05,
+      "loss": 1.1122,
+      "step": 1696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.303000000000001e-05,
+      "loss": 1.1071,
+      "step": 1697
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.302000000000001e-05,
+      "loss": 1.3267,
+      "step": 1698
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.300999999999999e-05,
+      "loss": 1.2468,
+      "step": 1699
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.3e-05,
+      "loss": 1.4817,
+      "step": 1700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.299e-05,
+      "loss": 1.1168,
+      "step": 1701
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.298000000000001e-05,
+      "loss": 0.252,
+      "step": 1702
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.297e-05,
+      "loss": 0.1169,
+      "step": 1703
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.296e-05,
+      "loss": 0.0314,
+      "step": 1704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.295000000000001e-05,
+      "loss": 0.3224,
+      "step": 1705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.294e-05,
+      "loss": 1.0449,
+      "step": 1706
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.293000000000001e-05,
+      "loss": 1.2208,
+      "step": 1707
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.292000000000001e-05,
+      "loss": 1.0519,
+      "step": 1708
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.291e-05,
+      "loss": 1.34,
+      "step": 1709
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.29e-05,
+      "loss": 1.0522,
+      "step": 1710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.289e-05,
+      "loss": 1.4127,
+      "step": 1711
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.288e-05,
+      "loss": 0.9634,
+      "step": 1712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.287e-05,
+      "loss": 1.9191,
+      "step": 1713
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.286e-05,
+      "loss": 0.8681,
+      "step": 1714
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.285000000000001e-05,
+      "loss": 0.3608,
+      "step": 1715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.284000000000001e-05,
+      "loss": 0.1476,
+      "step": 1716
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.283e-05,
+      "loss": 0.1206,
+      "step": 1717
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.282000000000001e-05,
+      "loss": 0.5164,
+      "step": 1718
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.281e-05,
+      "loss": 1.3831,
+      "step": 1719
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.28e-05,
+      "loss": 1.3453,
+      "step": 1720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.279e-05,
+      "loss": 1.5508,
+      "step": 1721
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.278e-05,
+      "loss": 0.9817,
+      "step": 1722
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.277000000000001e-05,
+      "loss": 1.1503,
+      "step": 1723
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.276e-05,
+      "loss": 0.7413,
+      "step": 1724
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.275e-05,
+      "loss": 1.2544,
+      "step": 1725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.274000000000001e-05,
+      "loss": 1.4772,
+      "step": 1726
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.273e-05,
+      "loss": 0.1664,
+      "step": 1727
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.272000000000001e-05,
+      "loss": 0.1514,
+      "step": 1728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.271e-05,
+      "loss": 0.3768,
+      "step": 1729
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.27e-05,
+      "loss": 0.6645,
+      "step": 1730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.269e-05,
+      "loss": 0.8073,
+      "step": 1731
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.268e-05,
+      "loss": 1.0054,
+      "step": 1732
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.267e-05,
+      "loss": 0.993,
+      "step": 1733
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.266e-05,
+      "loss": 1.0263,
+      "step": 1734
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.265e-05,
+      "loss": 1.2187,
+      "step": 1735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.264000000000001e-05,
+      "loss": 0.8114,
+      "step": 1736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.263000000000001e-05,
+      "loss": 1.0905,
+      "step": 1737
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.262e-05,
+      "loss": 0.8568,
+      "step": 1738
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.261e-05,
+      "loss": 0.9313,
+      "step": 1739
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.26e-05,
+      "loss": 0.24,
+      "step": 1740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.259000000000001e-05,
+      "loss": 0.0356,
+      "step": 1741
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.258e-05,
+      "loss": 0.0325,
+      "step": 1742
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.257e-05,
+      "loss": 0.7291,
+      "step": 1743
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.256000000000001e-05,
+      "loss": 1.0836,
+      "step": 1744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.255e-05,
+      "loss": 1.3508,
+      "step": 1745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.254e-05,
+      "loss": 0.9242,
+      "step": 1746
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.253000000000001e-05,
+      "loss": 1.4823,
+      "step": 1747
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.252e-05,
+      "loss": 1.1169,
+      "step": 1748
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.251e-05,
+      "loss": 0.6823,
+      "step": 1749
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.25e-05,
+      "loss": 0.6962,
+      "step": 1750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.249e-05,
+      "loss": 1.2221,
+      "step": 1751
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.248e-05,
+      "loss": 0.8152,
+      "step": 1752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.247e-05,
+      "loss": 0.255,
+      "step": 1753
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.246e-05,
+      "loss": 0.1211,
+      "step": 1754
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.245e-05,
+      "loss": 0.7931,
+      "step": 1755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.244e-05,
+      "loss": 0.9798,
+      "step": 1756
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.243000000000001e-05,
+      "loss": 1.4428,
+      "step": 1757
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.242000000000001e-05,
+      "loss": 0.9309,
+      "step": 1758
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.241e-05,
+      "loss": 0.9858,
+      "step": 1759
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.24e-05,
+      "loss": 1.0031,
+      "step": 1760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.239e-05,
+      "loss": 1.3937,
+      "step": 1761
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.238000000000001e-05,
+      "loss": 0.8648,
+      "step": 1762
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.237e-05,
+      "loss": 1.5896,
+      "step": 1763
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.236e-05,
+      "loss": 0.4855,
+      "step": 1764
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.235000000000001e-05,
+      "loss": 0.2466,
+      "step": 1765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.234e-05,
+      "loss": 0.2184,
+      "step": 1766
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.233e-05,
+      "loss": 0.3429,
+      "step": 1767
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.232000000000001e-05,
+      "loss": 1.1573,
+      "step": 1768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.231e-05,
+      "loss": 0.7747,
+      "step": 1769
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.23e-05,
+      "loss": 0.8613,
+      "step": 1770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.229e-05,
+      "loss": 0.9504,
+      "step": 1771
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.228e-05,
+      "loss": 1.1733,
+      "step": 1772
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.227e-05,
+      "loss": 1.1181,
+      "step": 1773
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.226e-05,
+      "loss": 1.2626,
+      "step": 1774
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.225000000000001e-05,
+      "loss": 1.3,
+      "step": 1775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.224000000000001e-05,
+      "loss": 0.7666,
+      "step": 1776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.223e-05,
+      "loss": 0.2018,
+      "step": 1777
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.222000000000001e-05,
+      "loss": 0.4964,
+      "step": 1778
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.221000000000001e-05,
+      "loss": 0.3836,
+      "step": 1779
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.22e-05,
+      "loss": 0.1543,
+      "step": 1780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.219e-05,
+      "loss": 1.124,
+      "step": 1781
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.218e-05,
+      "loss": 1.4117,
+      "step": 1782
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.217000000000001e-05,
+      "loss": 1.5504,
+      "step": 1783
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.216e-05,
+      "loss": 1.3455,
+      "step": 1784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.215e-05,
+      "loss": 1.0382,
+      "step": 1785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.214000000000001e-05,
+      "loss": 1.219,
+      "step": 1786
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.213e-05,
+      "loss": 1.2271,
+      "step": 1787
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.212e-05,
+      "loss": 1.2384,
+      "step": 1788
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.211000000000001e-05,
+      "loss": 0.9152,
+      "step": 1789
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.21e-05,
+      "loss": 0.4736,
+      "step": 1790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.209e-05,
+      "loss": 0.0438,
+      "step": 1791
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.208e-05,
+      "loss": 0.1609,
+      "step": 1792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.207e-05,
+      "loss": 1.0492,
+      "step": 1793
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.206e-05,
+      "loss": 1.507,
+      "step": 1794
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.205e-05,
+      "loss": 1.1995,
+      "step": 1795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.204000000000001e-05,
+      "loss": 1.112,
+      "step": 1796
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.203000000000001e-05,
+      "loss": 1.384,
+      "step": 1797
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.202e-05,
+      "loss": 1.0986,
+      "step": 1798
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.201000000000001e-05,
+      "loss": 1.3507,
+      "step": 1799
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.2e-05,
+      "loss": 1.2672,
+      "step": 1800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.199e-05,
+      "loss": 0.98,
+      "step": 1801
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.198e-05,
+      "loss": 0.3346,
+      "step": 1802
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.197e-05,
+      "loss": 0.0383,
+      "step": 1803
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.196000000000001e-05,
+      "loss": 0.0597,
+      "step": 1804
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.195e-05,
+      "loss": 0.1088,
+      "step": 1805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.194e-05,
+      "loss": 0.9097,
+      "step": 1806
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.193000000000001e-05,
+      "loss": 1.0563,
+      "step": 1807
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.192e-05,
+      "loss": 1.2608,
+      "step": 1808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.191000000000001e-05,
+      "loss": 1.0544,
+      "step": 1809
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.19e-05,
+      "loss": 1.1822,
+      "step": 1810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.189e-05,
+      "loss": 1.4678,
+      "step": 1811
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.188e-05,
+      "loss": 0.8682,
+      "step": 1812
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.187e-05,
+      "loss": 1.5867,
+      "step": 1813
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.186e-05,
+      "loss": 0.695,
+      "step": 1814
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.185000000000001e-05,
+      "loss": 0.2228,
+      "step": 1815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.184e-05,
+      "loss": 0.3093,
+      "step": 1816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.183000000000001e-05,
+      "loss": 0.0497,
+      "step": 1817
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.182000000000001e-05,
+      "loss": 0.5756,
+      "step": 1818
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.181e-05,
+      "loss": 1.7945,
+      "step": 1819
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.18e-05,
+      "loss": 1.2283,
+      "step": 1820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.179e-05,
+      "loss": 1.1857,
+      "step": 1821
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.178e-05,
+      "loss": 1.2509,
+      "step": 1822
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.177e-05,
+      "loss": 1.2847,
+      "step": 1823
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.176e-05,
+      "loss": 1.253,
+      "step": 1824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.175000000000001e-05,
+      "loss": 1.5618,
+      "step": 1825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.174e-05,
+      "loss": 1.684,
+      "step": 1826
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.173e-05,
+      "loss": 0.2984,
+      "step": 1827
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.172000000000001e-05,
+      "loss": 0.1755,
+      "step": 1828
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.171e-05,
+      "loss": 0.4283,
+      "step": 1829
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.17e-05,
+      "loss": 0.0681,
+      "step": 1830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.169e-05,
+      "loss": 0.8127,
+      "step": 1831
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.168e-05,
+      "loss": 1.1282,
+      "step": 1832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.167e-05,
+      "loss": 1.185,
+      "step": 1833
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.166e-05,
+      "loss": 1.3669,
+      "step": 1834
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.165e-05,
+      "loss": 1.3339,
+      "step": 1835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.164000000000001e-05,
+      "loss": 1.1068,
+      "step": 1836
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.163e-05,
+      "loss": 1.2081,
+      "step": 1837
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.162000000000001e-05,
+      "loss": 1.0546,
+      "step": 1838
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.161000000000001e-05,
+      "loss": 0.6636,
+      "step": 1839
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.16e-05,
+      "loss": 0.0988,
+      "step": 1840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.159e-05,
+      "loss": 0.3256,
+      "step": 1841
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.158e-05,
+      "loss": 0.2529,
+      "step": 1842
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.157e-05,
+      "loss": 0.4722,
+      "step": 1843
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.156e-05,
+      "loss": 1.0792,
+      "step": 1844
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.155e-05,
+      "loss": 1.0704,
+      "step": 1845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.154000000000001e-05,
+      "loss": 0.8635,
+      "step": 1846
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.153e-05,
+      "loss": 0.8731,
+      "step": 1847
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.152e-05,
+      "loss": 0.935,
+      "step": 1848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.151000000000001e-05,
+      "loss": 1.2374,
+      "step": 1849
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.15e-05,
+      "loss": 1.0448,
+      "step": 1850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.149e-05,
+      "loss": 1.2109,
+      "step": 1851
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.148e-05,
+      "loss": 0.2154,
+      "step": 1852
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.147e-05,
+      "loss": 0.136,
+      "step": 1853
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.146000000000001e-05,
+      "loss": 0.1451,
+      "step": 1854
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.145e-05,
+      "loss": 0.4234,
+      "step": 1855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.144e-05,
+      "loss": 0.6852,
+      "step": 1856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.143000000000001e-05,
+      "loss": 1.5393,
+      "step": 1857
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.142e-05,
+      "loss": 1.4481,
+      "step": 1858
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.141000000000001e-05,
+      "loss": 1.1193,
+      "step": 1859
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.14e-05,
+      "loss": 1.1672,
+      "step": 1860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.139e-05,
+      "loss": 1.2653,
+      "step": 1861
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.138e-05,
+      "loss": 1.7862,
+      "step": 1862
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.137e-05,
+      "loss": 1.4683,
+      "step": 1863
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.136000000000001e-05,
+      "loss": 0.8654,
+      "step": 1864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.135e-05,
+      "loss": 0.0626,
+      "step": 1865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.134e-05,
+      "loss": 0.2516,
+      "step": 1866
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.133000000000001e-05,
+      "loss": 1.0205,
+      "step": 1867
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.132e-05,
+      "loss": 1.2191,
+      "step": 1868
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.131e-05,
+      "loss": 1.2775,
+      "step": 1869
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.13e-05,
+      "loss": 0.906,
+      "step": 1870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.129e-05,
+      "loss": 1.2931,
+      "step": 1871
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.128e-05,
+      "loss": 1.1881,
+      "step": 1872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.127e-05,
+      "loss": 1.3301,
+      "step": 1873
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.126e-05,
+      "loss": 0.8952,
+      "step": 1874
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 1.829,
+      "step": 1875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.124e-05,
+      "loss": 1.3323,
+      "step": 1876
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.123e-05,
+      "loss": 0.3372,
+      "step": 1877
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.122000000000001e-05,
+      "loss": 0.1561,
+      "step": 1878
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.121e-05,
+      "loss": 0.0414,
+      "step": 1879
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.120000000000001e-05,
+      "loss": 0.6018,
+      "step": 1880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.119e-05,
+      "loss": 0.7546,
+      "step": 1881
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.118e-05,
+      "loss": 1.0935,
+      "step": 1882
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.117e-05,
+      "loss": 0.9688,
+      "step": 1883
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.116e-05,
+      "loss": 1.1831,
+      "step": 1884
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.115000000000001e-05,
+      "loss": 1.3645,
+      "step": 1885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.114e-05,
+      "loss": 0.6727,
+      "step": 1886
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.113e-05,
+      "loss": 1.3246,
+      "step": 1887
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.112000000000001e-05,
+      "loss": 1.6789,
+      "step": 1888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.111000000000001e-05,
+      "loss": 0.8891,
+      "step": 1889
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.11e-05,
+      "loss": 0.3361,
+      "step": 1890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.109e-05,
+      "loss": 0.0098,
+      "step": 1891
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.108e-05,
+      "loss": 0.0902,
+      "step": 1892
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.107000000000001e-05,
+      "loss": 0.1163,
+      "step": 1893
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.106e-05,
+      "loss": 1.0655,
+      "step": 1894
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.105e-05,
+      "loss": 1.1341,
+      "step": 1895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.104000000000001e-05,
+      "loss": 1.319,
+      "step": 1896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.103e-05,
+      "loss": 1.1109,
+      "step": 1897
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.102000000000001e-05,
+      "loss": 1.2536,
+      "step": 1898
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.101000000000001e-05,
+      "loss": 0.9869,
+      "step": 1899
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.1e-05,
+      "loss": 1.6697,
+      "step": 1900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.099e-05,
+      "loss": 1.3259,
+      "step": 1901
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.098e-05,
+      "loss": 0.0728,
+      "step": 1902
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.097e-05,
+      "loss": 0.1223,
+      "step": 1903
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.096e-05,
+      "loss": 0.488,
+      "step": 1904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.095e-05,
+      "loss": 1.0128,
+      "step": 1905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.094000000000001e-05,
+      "loss": 1.165,
+      "step": 1906
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.093e-05,
+      "loss": 1.1897,
+      "step": 1907
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.092e-05,
+      "loss": 1.2121,
+      "step": 1908
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.091000000000001e-05,
+      "loss": 1.1722,
+      "step": 1909
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.090000000000001e-05,
+      "loss": 0.9041,
+      "step": 1910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.088999999999999e-05,
+      "loss": 1.0713,
+      "step": 1911
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.088e-05,
+      "loss": 1.1291,
+      "step": 1912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.087e-05,
+      "loss": 1.6658,
+      "step": 1913
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.086000000000001e-05,
+      "loss": 1.1876,
+      "step": 1914
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.085e-05,
+      "loss": 0.3078,
+      "step": 1915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.084e-05,
+      "loss": 0.8302,
+      "step": 1916
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.083000000000001e-05,
+      "loss": 0.0839,
+      "step": 1917
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.082e-05,
+      "loss": 0.1323,
+      "step": 1918
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.081000000000001e-05,
+      "loss": 0.5403,
+      "step": 1919
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.080000000000001e-05,
+      "loss": 0.9221,
+      "step": 1920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.079e-05,
+      "loss": 1.0876,
+      "step": 1921
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.078e-05,
+      "loss": 1.1222,
+      "step": 1922
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.077e-05,
+      "loss": 1.2188,
+      "step": 1923
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.076e-05,
+      "loss": 0.7847,
+      "step": 1924
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.075e-05,
+      "loss": 1.2721,
+      "step": 1925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.074e-05,
+      "loss": 1.4164,
+      "step": 1926
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.073000000000001e-05,
+      "loss": 0.1872,
+      "step": 1927
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.072000000000001e-05,
+      "loss": 0.1514,
+      "step": 1928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.071e-05,
+      "loss": 0.2412,
+      "step": 1929
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.070000000000001e-05,
+      "loss": 0.2986,
+      "step": 1930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.069e-05,
+      "loss": 0.4961,
+      "step": 1931
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.068e-05,
+      "loss": 1.4863,
+      "step": 1932
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.067e-05,
+      "loss": 1.2842,
+      "step": 1933
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.066e-05,
+      "loss": 1.2967,
+      "step": 1934
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.065000000000001e-05,
+      "loss": 1.3499,
+      "step": 1935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.064e-05,
+      "loss": 1.2417,
+      "step": 1936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.063e-05,
+      "loss": 1.5527,
+      "step": 1937
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.062000000000001e-05,
+      "loss": 1.5535,
+      "step": 1938
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.061e-05,
+      "loss": 1.1955,
+      "step": 1939
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.060000000000001e-05,
+      "loss": 0.2328,
+      "step": 1940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.059e-05,
+      "loss": 0.1166,
+      "step": 1941
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.058e-05,
+      "loss": 0.1006,
+      "step": 1942
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.057e-05,
+      "loss": 0.1739,
+      "step": 1943
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.056e-05,
+      "loss": 0.879,
+      "step": 1944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.055e-05,
+      "loss": 1.2838,
+      "step": 1945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.054e-05,
+      "loss": 1.3868,
+      "step": 1946
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.053e-05,
+      "loss": 1.0288,
+      "step": 1947
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.052000000000001e-05,
+      "loss": 1.0541,
+      "step": 1948
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.051000000000001e-05,
+      "loss": 1.1773,
+      "step": 1949
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.05e-05,
+      "loss": 0.9477,
+      "step": 1950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.049e-05,
+      "loss": 1.2427,
+      "step": 1951
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.048e-05,
+      "loss": 0.2992,
+      "step": 1952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.047000000000001e-05,
+      "loss": 0.1646,
+      "step": 1953
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.046e-05,
+      "loss": 0.1854,
+      "step": 1954
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.045e-05,
+      "loss": 0.6296,
+      "step": 1955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.044000000000001e-05,
+      "loss": 1.1305,
+      "step": 1956
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.043e-05,
+      "loss": 1.4134,
+      "step": 1957
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.042e-05,
+      "loss": 1.1609,
+      "step": 1958
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.041000000000001e-05,
+      "loss": 1.3019,
+      "step": 1959
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.04e-05,
+      "loss": 1.7876,
+      "step": 1960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.039e-05,
+      "loss": 1.3208,
+      "step": 1961
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.038e-05,
+      "loss": 1.3679,
+      "step": 1962
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.037e-05,
+      "loss": 1.1665,
+      "step": 1963
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.036e-05,
+      "loss": 0.9095,
+      "step": 1964
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.035e-05,
+      "loss": 0.1553,
+      "step": 1965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.034e-05,
+      "loss": 0.2477,
+      "step": 1966
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.033000000000001e-05,
+      "loss": 0.1853,
+      "step": 1967
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.032e-05,
+      "loss": 0.1546,
+      "step": 1968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.031000000000001e-05,
+      "loss": 1.5783,
+      "step": 1969
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.030000000000001e-05,
+      "loss": 1.1856,
+      "step": 1970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.028999999999999e-05,
+      "loss": 1.1521,
+      "step": 1971
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.028e-05,
+      "loss": 1.1456,
+      "step": 1972
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.027e-05,
+      "loss": 0.9632,
+      "step": 1973
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.026000000000001e-05,
+      "loss": 0.8131,
+      "step": 1974
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.025e-05,
+      "loss": 0.7412,
+      "step": 1975
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.024e-05,
+      "loss": 1.6912,
+      "step": 1976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.023000000000001e-05,
+      "loss": 0.2576,
+      "step": 1977
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.022e-05,
+      "loss": 0.1012,
+      "step": 1978
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.021e-05,
+      "loss": 0.2303,
+      "step": 1979
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.020000000000001e-05,
+      "loss": 0.0971,
+      "step": 1980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.019e-05,
+      "loss": 0.8302,
+      "step": 1981
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.018e-05,
+      "loss": 1.6749,
+      "step": 1982
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.017e-05,
+      "loss": 1.1195,
+      "step": 1983
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.016e-05,
+      "loss": 1.2414,
+      "step": 1984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.015e-05,
+      "loss": 1.1132,
+      "step": 1985
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.014e-05,
+      "loss": 1.288,
+      "step": 1986
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.013000000000001e-05,
+      "loss": 1.1474,
+      "step": 1987
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.012000000000001e-05,
+      "loss": 1.2894,
+      "step": 1988
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.011e-05,
+      "loss": 0.3909,
+      "step": 1989
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.010000000000001e-05,
+      "loss": 0.5532,
+      "step": 1990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.009e-05,
+      "loss": 0.0355,
+      "step": 1991
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.008e-05,
+      "loss": 0.4947,
+      "step": 1992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.007e-05,
+      "loss": 1.0357,
+      "step": 1993
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.006e-05,
+      "loss": 1.3709,
+      "step": 1994
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.005000000000001e-05,
+      "loss": 0.9066,
+      "step": 1995
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.004e-05,
+      "loss": 1.0372,
+      "step": 1996
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.003e-05,
+      "loss": 1.3376,
+      "step": 1997
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.002000000000001e-05,
+      "loss": 1.7055,
+      "step": 1998
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.001e-05,
+      "loss": 0.7143,
+      "step": 1999
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8e-05,
+      "loss": 0.9939,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.999000000000001e-05,
+      "loss": 1.0992,
+      "step": 2001
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.998e-05,
+      "loss": 0.049,
+      "step": 2002
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.997e-05,
+      "loss": 0.0045,
+      "step": 2003
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.996e-05,
+      "loss": 0.1159,
+      "step": 2004
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.995e-05,
+      "loss": 0.1569,
+      "step": 2005
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.994000000000001e-05,
+      "loss": 0.4377,
+      "step": 2006
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.993e-05,
+      "loss": 1.1258,
+      "step": 2007
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.992000000000001e-05,
+      "loss": 1.1372,
+      "step": 2008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.991000000000001e-05,
+      "loss": 1.1335,
+      "step": 2009
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.99e-05,
+      "loss": 1.1351,
+      "step": 2010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.989000000000001e-05,
+      "loss": 1.1311,
+      "step": 2011
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.988e-05,
+      "loss": 1.2729,
+      "step": 2012
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.987e-05,
+      "loss": 1.6681,
+      "step": 2013
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.986e-05,
+      "loss": 0.5075,
+      "step": 2014
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.985e-05,
+      "loss": 0.2166,
+      "step": 2015
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.984000000000001e-05,
+      "loss": 0.2387,
+      "step": 2016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.983e-05,
+      "loss": 0.4079,
+      "step": 2017
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.982e-05,
+      "loss": 0.2535,
+      "step": 2018
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.981000000000001e-05,
+      "loss": 1.5508,
+      "step": 2019
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.98e-05,
+      "loss": 1.3107,
+      "step": 2020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.979000000000001e-05,
+      "loss": 1.3341,
+      "step": 2021
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.978e-05,
+      "loss": 1.1021,
+      "step": 2022
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.977e-05,
+      "loss": 1.2546,
+      "step": 2023
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.976e-05,
+      "loss": 1.1921,
+      "step": 2024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.975e-05,
+      "loss": 0.893,
+      "step": 2025
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.974e-05,
+      "loss": 1.3977,
+      "step": 2026
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.973000000000001e-05,
+      "loss": 0.3016,
+      "step": 2027
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.972e-05,
+      "loss": 0.3007,
+      "step": 2028
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.971000000000001e-05,
+      "loss": 0.0731,
+      "step": 2029
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.970000000000001e-05,
+      "loss": 0.2482,
+      "step": 2030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.969e-05,
+      "loss": 1.1195,
+      "step": 2031
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.968e-05,
+      "loss": 0.6833,
+      "step": 2032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.967e-05,
+      "loss": 1.0914,
+      "step": 2033
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.966e-05,
+      "loss": 0.8315,
+      "step": 2034
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.965e-05,
+      "loss": 1.2783,
+      "step": 2035
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.964e-05,
+      "loss": 0.656,
+      "step": 2036
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.963000000000001e-05,
+      "loss": 1.0266,
+      "step": 2037
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.962e-05,
+      "loss": 0.9112,
+      "step": 2038
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.961e-05,
+      "loss": 0.9255,
+      "step": 2039
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.960000000000001e-05,
+      "loss": 0.1923,
+      "step": 2040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.959000000000002e-05,
+      "loss": 0.1861,
+      "step": 2041
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.958e-05,
+      "loss": 0.0085,
+      "step": 2042
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.957e-05,
+      "loss": 0.4009,
+      "step": 2043
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.956e-05,
+      "loss": 1.0642,
+      "step": 2044
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.955e-05,
+      "loss": 1.1068,
+      "step": 2045
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.954e-05,
+      "loss": 1.4298,
+      "step": 2046
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.953e-05,
+      "loss": 1.2247,
+      "step": 2047
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.952000000000001e-05,
+      "loss": 1.212,
+      "step": 2048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.951e-05,
+      "loss": 2.0007,
+      "step": 2049
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.950000000000001e-05,
+      "loss": 0.9195,
+      "step": 2050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.949000000000001e-05,
+      "loss": 1.3404,
+      "step": 2051
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.948e-05,
+      "loss": 0.2268,
+      "step": 2052
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.947e-05,
+      "loss": 0.0411,
+      "step": 2053
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.946e-05,
+      "loss": 0.1531,
+      "step": 2054
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.945e-05,
+      "loss": 0.0714,
+      "step": 2055
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.944e-05,
+      "loss": 1.0323,
+      "step": 2056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.943e-05,
+      "loss": 1.5979,
+      "step": 2057
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.942000000000001e-05,
+      "loss": 1.2377,
+      "step": 2058
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.941e-05,
+      "loss": 0.6765,
+      "step": 2059
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.94e-05,
+      "loss": 1.3115,
+      "step": 2060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.939000000000001e-05,
+      "loss": 1.8215,
+      "step": 2061
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.938e-05,
+      "loss": 1.5104,
+      "step": 2062
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.937e-05,
+      "loss": 0.885,
+      "step": 2063
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.936e-05,
+      "loss": 1.1124,
+      "step": 2064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.935e-05,
+      "loss": 0.4084,
+      "step": 2065
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.934000000000001e-05,
+      "loss": 0.0365,
+      "step": 2066
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.933e-05,
+      "loss": 0.2432,
+      "step": 2067
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.932e-05,
+      "loss": 0.2367,
+      "step": 2068
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.931000000000001e-05,
+      "loss": 1.1981,
+      "step": 2069
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.93e-05,
+      "loss": 1.0813,
+      "step": 2070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.929000000000001e-05,
+      "loss": 0.7601,
+      "step": 2071
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.928e-05,
+      "loss": 1.8364,
+      "step": 2072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.927e-05,
+      "loss": 1.523,
+      "step": 2073
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.926e-05,
+      "loss": 0.5863,
+      "step": 2074
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.925e-05,
+      "loss": 1.4332,
+      "step": 2075
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.924000000000001e-05,
+      "loss": 1.2106,
+      "step": 2076
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.923e-05,
+      "loss": 0.2328,
+      "step": 2077
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.922e-05,
+      "loss": 0.3482,
+      "step": 2078
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.921000000000001e-05,
+      "loss": 0.3931,
+      "step": 2079
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.920000000000001e-05,
+      "loss": 0.5662,
+      "step": 2080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.919e-05,
+      "loss": 0.8476,
+      "step": 2081
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.918e-05,
+      "loss": 0.9285,
+      "step": 2082
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.917e-05,
+      "loss": 1.2364,
+      "step": 2083
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.916e-05,
+      "loss": 1.1947,
+      "step": 2084
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.915e-05,
+      "loss": 0.5775,
+      "step": 2085
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.914e-05,
+      "loss": 1.4359,
+      "step": 2086
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.913000000000001e-05,
+      "loss": 1.1445,
+      "step": 2087
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.912e-05,
+      "loss": 1.1593,
+      "step": 2088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.911e-05,
+      "loss": 0.7197,
+      "step": 2089
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.910000000000001e-05,
+      "loss": 0.0867,
+      "step": 2090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.909e-05,
+      "loss": 0.2231,
+      "step": 2091
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.908e-05,
+      "loss": 0.1684,
+      "step": 2092
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.907e-05,
+      "loss": 0.5571,
+      "step": 2093
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.906e-05,
+      "loss": 1.3652,
+      "step": 2094
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.905e-05,
+      "loss": 1.298,
+      "step": 2095
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.904e-05,
+      "loss": 1.285,
+      "step": 2096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.903000000000001e-05,
+      "loss": 1.1643,
+      "step": 2097
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.902e-05,
+      "loss": 1.599,
+      "step": 2098
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.901e-05,
+      "loss": 0.9421,
+      "step": 2099
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 0.8673,
+      "step": 2100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.899000000000001e-05,
+      "loss": 1.5188,
+      "step": 2101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.897999999999999e-05,
+      "loss": 0.4541,
+      "step": 2102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.897e-05,
+      "loss": 0.3355,
+      "step": 2103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.896e-05,
+      "loss": 0.2611,
+      "step": 2104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.895000000000001e-05,
+      "loss": 0.2728,
+      "step": 2105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.894e-05,
+      "loss": 1.0077,
+      "step": 2106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.893e-05,
+      "loss": 0.9632,
+      "step": 2107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.892000000000001e-05,
+      "loss": 1.0997,
+      "step": 2108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.891e-05,
+      "loss": 1.3949,
+      "step": 2109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.890000000000001e-05,
+      "loss": 1.0669,
+      "step": 2110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.889000000000001e-05,
+      "loss": 1.0606,
+      "step": 2111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.888e-05,
+      "loss": 1.5075,
+      "step": 2112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.887e-05,
+      "loss": 1.1108,
+      "step": 2113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.886e-05,
+      "loss": 0.719,
+      "step": 2114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.885e-05,
+      "loss": 0.0104,
+      "step": 2115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.884e-05,
+      "loss": 0.2182,
+      "step": 2116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.883e-05,
+      "loss": 0.2185,
+      "step": 2117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.882000000000001e-05,
+      "loss": 0.4693,
+      "step": 2118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.881e-05,
+      "loss": 0.5767,
+      "step": 2119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.88e-05,
+      "loss": 1.5664,
+      "step": 2120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.879000000000001e-05,
+      "loss": 0.9786,
+      "step": 2121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.878e-05,
+      "loss": 1.199,
+      "step": 2122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.876999999999999e-05,
+      "loss": 1.1329,
+      "step": 2123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.876e-05,
+      "loss": 1.2273,
+      "step": 2124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.875e-05,
+      "loss": 1.1532,
+      "step": 2125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.874000000000001e-05,
+      "loss": 1.3797,
+      "step": 2126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.873e-05,
+      "loss": 0.4431,
+      "step": 2127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.872e-05,
+      "loss": 0.0092,
+      "step": 2128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.871000000000001e-05,
+      "loss": 0.1814,
+      "step": 2129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.87e-05,
+      "loss": 0.0525,
+      "step": 2130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.869000000000001e-05,
+      "loss": 0.806,
+      "step": 2131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.868000000000001e-05,
+      "loss": 0.7571,
+      "step": 2132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.867e-05,
+      "loss": 1.1332,
+      "step": 2133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.866e-05,
+      "loss": 1.4643,
+      "step": 2134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.865e-05,
+      "loss": 1.2914,
+      "step": 2135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.864e-05,
+      "loss": 0.9492,
+      "step": 2136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.863e-05,
+      "loss": 1.0338,
+      "step": 2137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.862e-05,
+      "loss": 2.0818,
+      "step": 2138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.861000000000001e-05,
+      "loss": 0.5962,
+      "step": 2139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.860000000000001e-05,
+      "loss": 0.1704,
+      "step": 2140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.859e-05,
+      "loss": 0.2211,
+      "step": 2141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.858000000000001e-05,
+      "loss": 0.2312,
+      "step": 2142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.857e-05,
+      "loss": 0.1191,
+      "step": 2143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.856000000000001e-05,
+      "loss": 0.7122,
+      "step": 2144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.855e-05,
+      "loss": 1.1821,
+      "step": 2145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.854e-05,
+      "loss": 1.3169,
+      "step": 2146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.853000000000001e-05,
+      "loss": 1.0559,
+      "step": 2147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.852e-05,
+      "loss": 0.9535,
+      "step": 2148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.851e-05,
+      "loss": 1.4142,
+      "step": 2149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.850000000000001e-05,
+      "loss": 1.4048,
+      "step": 2150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.849e-05,
+      "loss": 1.2392,
+      "step": 2151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.848000000000001e-05,
+      "loss": 0.1712,
+      "step": 2152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.847e-05,
+      "loss": 0.327,
+      "step": 2153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.846e-05,
+      "loss": 0.0672,
+      "step": 2154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.845e-05,
+      "loss": 0.3875,
+      "step": 2155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.844e-05,
+      "loss": 1.2741,
+      "step": 2156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.843e-05,
+      "loss": 1.2992,
+      "step": 2157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.842e-05,
+      "loss": 1.3409,
+      "step": 2158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.841e-05,
+      "loss": 1.1146,
+      "step": 2159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.840000000000001e-05,
+      "loss": 1.3818,
+      "step": 2160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.839000000000001e-05,
+      "loss": 1.0259,
+      "step": 2161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.838e-05,
+      "loss": 1.1598,
+      "step": 2162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.837e-05,
+      "loss": 1.6042,
+      "step": 2163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.836e-05,
+      "loss": 0.606,
+      "step": 2164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.835000000000001e-05,
+      "loss": 0.2252,
+      "step": 2165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.834e-05,
+      "loss": 0.1498,
+      "step": 2166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.833e-05,
+      "loss": 0.1229,
+      "step": 2167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.832000000000001e-05,
+      "loss": 0.2794,
+      "step": 2168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.831e-05,
+      "loss": 1.3409,
+      "step": 2169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.83e-05,
+      "loss": 1.013,
+      "step": 2170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.829000000000001e-05,
+      "loss": 1.1378,
+      "step": 2171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.828e-05,
+      "loss": 1.5603,
+      "step": 2172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.827e-05,
+      "loss": 1.3811,
+      "step": 2173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.826e-05,
+      "loss": 1.1709,
+      "step": 2174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.825e-05,
+      "loss": 0.3038,
+      "step": 2175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.824e-05,
+      "loss": 1.1223,
+      "step": 2176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.823e-05,
+      "loss": 0.5627,
+      "step": 2177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.822e-05,
+      "loss": 0.1173,
+      "step": 2178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.821000000000001e-05,
+      "loss": 0.5263,
+      "step": 2179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.82e-05,
+      "loss": 0.1486,
+      "step": 2180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.819000000000001e-05,
+      "loss": 0.813,
+      "step": 2181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.818000000000001e-05,
+      "loss": 1.0889,
+      "step": 2182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.817e-05,
+      "loss": 1.283,
+      "step": 2183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.816e-05,
+      "loss": 1.015,
+      "step": 2184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.815e-05,
+      "loss": 1.3519,
+      "step": 2185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.814000000000001e-05,
+      "loss": 1.1266,
+      "step": 2186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.813e-05,
+      "loss": 1.3933,
+      "step": 2187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.812e-05,
+      "loss": 1.1166,
+      "step": 2188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.811000000000001e-05,
+      "loss": 0.467,
+      "step": 2189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.81e-05,
+      "loss": 0.1438,
+      "step": 2190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.809e-05,
+      "loss": 0.1691,
+      "step": 2191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.808000000000001e-05,
+      "loss": 0.1639,
+      "step": 2192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.807e-05,
+      "loss": 0.9543,
+      "step": 2193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.806e-05,
+      "loss": 1.8086,
+      "step": 2194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.805e-05,
+      "loss": 0.9281,
+      "step": 2195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.804e-05,
+      "loss": 1.0491,
+      "step": 2196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.803e-05,
+      "loss": 1.1234,
+      "step": 2197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.802e-05,
+      "loss": 1.6201,
+      "step": 2198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.801000000000001e-05,
+      "loss": 1.0488,
+      "step": 2199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 1.4119,
+      "step": 2200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.799e-05,
+      "loss": 1.2316,
+      "step": 2201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.798000000000001e-05,
+      "loss": 0.2031,
+      "step": 2202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.797e-05,
+      "loss": 0.4917,
+      "step": 2203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.796e-05,
+      "loss": 0.3862,
+      "step": 2204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.795e-05,
+      "loss": 0.0923,
+      "step": 2205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.794e-05,
+      "loss": 0.6698,
+      "step": 2206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.793000000000001e-05,
+      "loss": 0.7486,
+      "step": 2207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.792e-05,
+      "loss": 1.1455,
+      "step": 2208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.791e-05,
+      "loss": 1.6763,
+      "step": 2209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.790000000000001e-05,
+      "loss": 1.1955,
+      "step": 2210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.789e-05,
+      "loss": 1.2824,
+      "step": 2211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.788e-05,
+      "loss": 1.0519,
+      "step": 2212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.787e-05,
+      "loss": 1.1821,
+      "step": 2213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.786e-05,
+      "loss": 1.0958,
+      "step": 2214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.785e-05,
+      "loss": 0.2802,
+      "step": 2215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.784e-05,
+      "loss": 0.1978,
+      "step": 2216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.783e-05,
+      "loss": 0.1316,
+      "step": 2217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.782000000000001e-05,
+      "loss": 0.6123,
+      "step": 2218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.781e-05,
+      "loss": 0.5516,
+      "step": 2219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 0.8666,
+      "step": 2220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.779000000000001e-05,
+      "loss": 1.3144,
+      "step": 2221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.778e-05,
+      "loss": 0.8847,
+      "step": 2222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.777e-05,
+      "loss": 1.011,
+      "step": 2223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.776e-05,
+      "loss": 1.1044,
+      "step": 2224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.775e-05,
+      "loss": 1.1053,
+      "step": 2225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.774e-05,
+      "loss": 1.1528,
+      "step": 2226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.773e-05,
+      "loss": 0.4328,
+      "step": 2227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.772000000000001e-05,
+      "loss": 0.4596,
+      "step": 2228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.771e-05,
+      "loss": 0.1262,
+      "step": 2229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.77e-05,
+      "loss": 0.5131,
+      "step": 2230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.769000000000001e-05,
+      "loss": 0.7247,
+      "step": 2231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.768e-05,
+      "loss": 0.825,
+      "step": 2232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.767e-05,
+      "loss": 1.1341,
+      "step": 2233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.766e-05,
+      "loss": 0.7827,
+      "step": 2234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.765e-05,
+      "loss": 1.1887,
+      "step": 2235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.764e-05,
+      "loss": 1.6993,
+      "step": 2236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.763e-05,
+      "loss": 1.2741,
+      "step": 2237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.762e-05,
+      "loss": 1.4981,
+      "step": 2238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.761000000000001e-05,
+      "loss": 0.7974,
+      "step": 2239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.76e-05,
+      "loss": 0.2508,
+      "step": 2240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.759000000000001e-05,
+      "loss": 0.2012,
+      "step": 2241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.758000000000001e-05,
+      "loss": 0.4645,
+      "step": 2242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.757e-05,
+      "loss": 0.5628,
+      "step": 2243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.756e-05,
+      "loss": 0.9943,
+      "step": 2244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.755e-05,
+      "loss": 1.5573,
+      "step": 2245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.754e-05,
+      "loss": 1.1518,
+      "step": 2246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.753e-05,
+      "loss": 1.2775,
+      "step": 2247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.752e-05,
+      "loss": 1.1563,
+      "step": 2248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.751000000000001e-05,
+      "loss": 0.7386,
+      "step": 2249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.75e-05,
+      "loss": 1.0864,
+      "step": 2250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.749e-05,
+      "loss": 1.1762,
+      "step": 2251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.748000000000001e-05,
+      "loss": 0.3494,
+      "step": 2252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.747000000000002e-05,
+      "loss": 0.1197,
+      "step": 2253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.746e-05,
+      "loss": 0.221,
+      "step": 2254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.745e-05,
+      "loss": 0.8454,
+      "step": 2255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.744e-05,
+      "loss": 1.5691,
+      "step": 2256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.743000000000001e-05,
+      "loss": 0.9209,
+      "step": 2257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.742e-05,
+      "loss": 1.0894,
+      "step": 2258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.741e-05,
+      "loss": 1.5197,
+      "step": 2259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.740000000000001e-05,
+      "loss": 1.5367,
+      "step": 2260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.739e-05,
+      "loss": 0.8675,
+      "step": 2261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.738000000000001e-05,
+      "loss": 1.3704,
+      "step": 2262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.737000000000001e-05,
+      "loss": 1.3666,
+      "step": 2263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.736e-05,
+      "loss": 0.6096,
+      "step": 2264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.735e-05,
+      "loss": 0.2091,
+      "step": 2265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.734e-05,
+      "loss": 0.1524,
+      "step": 2266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.733e-05,
+      "loss": 0.3979,
+      "step": 2267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.732e-05,
+      "loss": 0.3763,
+      "step": 2268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.731e-05,
+      "loss": 1.0697,
+      "step": 2269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.730000000000001e-05,
+      "loss": 1.242,
+      "step": 2270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.729e-05,
+      "loss": 0.9895,
+      "step": 2271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.728e-05,
+      "loss": 1.1027,
+      "step": 2272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.727000000000001e-05,
+      "loss": 1.2107,
+      "step": 2273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.726e-05,
+      "loss": 0.5361,
+      "step": 2274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.725e-05,
+      "loss": 1.5134,
+      "step": 2275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.724e-05,
+      "loss": 1.0599,
+      "step": 2276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.723e-05,
+      "loss": 0.4974,
+      "step": 2277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.722000000000001e-05,
+      "loss": 0.0464,
+      "step": 2278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.721e-05,
+      "loss": 0.2624,
+      "step": 2279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.72e-05,
+      "loss": 0.355,
+      "step": 2280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.719000000000001e-05,
+      "loss": 0.2266,
+      "step": 2281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.718e-05,
+      "loss": 0.9901,
+      "step": 2282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.717000000000001e-05,
+      "loss": 1.1893,
+      "step": 2283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.716e-05,
+      "loss": 1.4928,
+      "step": 2284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.715e-05,
+      "loss": 1.1633,
+      "step": 2285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.714e-05,
+      "loss": 1.4005,
+      "step": 2286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.713e-05,
+      "loss": 1.0707,
+      "step": 2287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.712000000000001e-05,
+      "loss": 1.7134,
+      "step": 2288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.711e-05,
+      "loss": 0.6191,
+      "step": 2289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.71e-05,
+      "loss": 0.293,
+      "step": 2290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.709000000000001e-05,
+      "loss": 0.1581,
+      "step": 2291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.708000000000001e-05,
+      "loss": 0.5268,
+      "step": 2292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.707e-05,
+      "loss": 0.5081,
+      "step": 2293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.706e-05,
+      "loss": 0.835,
+      "step": 2294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.705e-05,
+      "loss": 1.3907,
+      "step": 2295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.704000000000001e-05,
+      "loss": 1.0456,
+      "step": 2296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.703e-05,
+      "loss": 1.1928,
+      "step": 2297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.702e-05,
+      "loss": 1.1828,
+      "step": 2298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.701000000000001e-05,
+      "loss": 1.4221,
+      "step": 2299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.7e-05,
+      "loss": 0.6576,
+      "step": 2300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.699e-05,
+      "loss": 1.0293,
+      "step": 2301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.698000000000001e-05,
+      "loss": 0.099,
+      "step": 2302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.697e-05,
+      "loss": 0.098,
+      "step": 2303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.696e-05,
+      "loss": 0.0708,
+      "step": 2304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.695e-05,
+      "loss": 0.3944,
+      "step": 2305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.694e-05,
+      "loss": 0.3268,
+      "step": 2306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.693e-05,
+      "loss": 1.1246,
+      "step": 2307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.692e-05,
+      "loss": 0.9719,
+      "step": 2308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.691000000000001e-05,
+      "loss": 1.4936,
+      "step": 2309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.69e-05,
+      "loss": 1.8613,
+      "step": 2310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.689e-05,
+      "loss": 1.0574,
+      "step": 2311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.688000000000001e-05,
+      "loss": 1.0829,
+      "step": 2312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.687000000000001e-05,
+      "loss": 1.3643,
+      "step": 2313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.685999999999999e-05,
+      "loss": 1.0927,
+      "step": 2314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.685e-05,
+      "loss": 0.2802,
+      "step": 2315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.684e-05,
+      "loss": 0.2422,
+      "step": 2316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.683000000000001e-05,
+      "loss": 0.7675,
+      "step": 2317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.682e-05,
+      "loss": 0.8283,
+      "step": 2318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.681e-05,
+      "loss": 0.876,
+      "step": 2319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.680000000000001e-05,
+      "loss": 1.5577,
+      "step": 2320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.679e-05,
+      "loss": 1.093,
+      "step": 2321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.678000000000001e-05,
+      "loss": 1.2512,
+      "step": 2322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.677000000000001e-05,
+      "loss": 1.3236,
+      "step": 2323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.676e-05,
+      "loss": 1.249,
+      "step": 2324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.675e-05,
+      "loss": 1.0048,
+      "step": 2325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.674e-05,
+      "loss": 1.3709,
+      "step": 2326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.673e-05,
+      "loss": 0.9547,
+      "step": 2327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.672e-05,
+      "loss": 0.2512,
+      "step": 2328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.671e-05,
+      "loss": 0.2072,
+      "step": 2329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.670000000000001e-05,
+      "loss": 0.1041,
+      "step": 2330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.669000000000001e-05,
+      "loss": 0.2375,
+      "step": 2331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.668e-05,
+      "loss": 0.6822,
+      "step": 2332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.667000000000001e-05,
+      "loss": 1.2447,
+      "step": 2333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.666e-05,
+      "loss": 1.102,
+      "step": 2334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.664999999999999e-05,
+      "loss": 1.3294,
+      "step": 2335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.664e-05,
+      "loss": 0.8655,
+      "step": 2336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.663e-05,
+      "loss": 0.8401,
+      "step": 2337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.662000000000001e-05,
+      "loss": 0.9142,
+      "step": 2338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.661e-05,
+      "loss": 0.6895,
+      "step": 2339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.66e-05,
+      "loss": 0.3375,
+      "step": 2340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.659000000000001e-05,
+      "loss": 0.157,
+      "step": 2341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.658e-05,
+      "loss": 0.425,
+      "step": 2342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.657000000000001e-05,
+      "loss": 0.4456,
+      "step": 2343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.656e-05,
+      "loss": 1.1501,
+      "step": 2344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.655e-05,
+      "loss": 1.2035,
+      "step": 2345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.654e-05,
+      "loss": 1.2124,
+      "step": 2346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.653e-05,
+      "loss": 1.3894,
+      "step": 2347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.652e-05,
+      "loss": 1.0953,
+      "step": 2348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.651e-05,
+      "loss": 1.9246,
+      "step": 2349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.65e-05,
+      "loss": 0.855,
+      "step": 2350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.649000000000001e-05,
+      "loss": 1.0383,
+      "step": 2351
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.648000000000001e-05,
+      "loss": 0.1929,
+      "step": 2352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.647e-05,
+      "loss": 0.1962,
+      "step": 2353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.646e-05,
+      "loss": 0.0882,
+      "step": 2354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.645e-05,
+      "loss": 0.3083,
+      "step": 2355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.644e-05,
+      "loss": 0.3718,
+      "step": 2356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.643e-05,
+      "loss": 1.1209,
+      "step": 2357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.642e-05,
+      "loss": 1.358,
+      "step": 2358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.641000000000001e-05,
+      "loss": 0.9761,
+      "step": 2359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.64e-05,
+      "loss": 1.2789,
+      "step": 2360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.639e-05,
+      "loss": 1.4398,
+      "step": 2361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.638000000000001e-05,
+      "loss": 1.1993,
+      "step": 2362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.637e-05,
+      "loss": 1.1357,
+      "step": 2363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.636e-05,
+      "loss": 0.4677,
+      "step": 2364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.635e-05,
+      "loss": 0.0428,
+      "step": 2365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.634e-05,
+      "loss": 0.0036,
+      "step": 2366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.633e-05,
+      "loss": 0.0756,
+      "step": 2367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.632e-05,
+      "loss": 1.1385,
+      "step": 2368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.631e-05,
+      "loss": 1.316,
+      "step": 2369
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.630000000000001e-05,
+      "loss": 1.7185,
+      "step": 2370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.629e-05,
+      "loss": 1.0905,
+      "step": 2371
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.628000000000001e-05,
+      "loss": 0.7955,
+      "step": 2372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.627000000000001e-05,
+      "loss": 1.181,
+      "step": 2373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.625999999999999e-05,
+      "loss": 1.9186,
+      "step": 2374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.625e-05,
+      "loss": 1.2861,
+      "step": 2375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.624e-05,
+      "loss": 1.6822,
+      "step": 2376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.623000000000001e-05,
+      "loss": 0.8947,
+      "step": 2377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.622e-05,
+      "loss": 0.0943,
+      "step": 2378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.621e-05,
+      "loss": 0.1586,
+      "step": 2379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.620000000000001e-05,
+      "loss": 0.2758,
+      "step": 2380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.619e-05,
+      "loss": 0.8033,
+      "step": 2381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.618e-05,
+      "loss": 0.9547,
+      "step": 2382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.617000000000001e-05,
+      "loss": 1.1897,
+      "step": 2383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.616e-05,
+      "loss": 0.845,
+      "step": 2384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.615e-05,
+      "loss": 0.8495,
+      "step": 2385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.614e-05,
+      "loss": 1.5416,
+      "step": 2386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.613e-05,
+      "loss": 1.7412,
+      "step": 2387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.612e-05,
+      "loss": 1.6649,
+      "step": 2388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.611e-05,
+      "loss": 0.771,
+      "step": 2389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.61e-05,
+      "loss": 0.1224,
+      "step": 2390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.609000000000001e-05,
+      "loss": 0.0196,
+      "step": 2391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.608e-05,
+      "loss": 0.2318,
+      "step": 2392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.607000000000001e-05,
+      "loss": 0.6892,
+      "step": 2393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.606000000000001e-05,
+      "loss": 1.1793,
+      "step": 2394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.605e-05,
+      "loss": 1.4926,
+      "step": 2395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.604e-05,
+      "loss": 1.3694,
+      "step": 2396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.603e-05,
+      "loss": 0.9209,
+      "step": 2397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.602000000000001e-05,
+      "loss": 1.1886,
+      "step": 2398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.601e-05,
+      "loss": 1.2491,
+      "step": 2399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.6e-05,
+      "loss": 1.6417,
+      "step": 2400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.599000000000001e-05,
+      "loss": 1.1276,
+      "step": 2401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.598e-05,
+      "loss": 0.3928,
+      "step": 2402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.597e-05,
+      "loss": 0.486,
+      "step": 2403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.596000000000001e-05,
+      "loss": 0.3122,
+      "step": 2404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.595e-05,
+      "loss": 0.6623,
+      "step": 2405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.594e-05,
+      "loss": 0.5827,
+      "step": 2406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.593e-05,
+      "loss": 1.348,
+      "step": 2407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.592e-05,
+      "loss": 1.1682,
+      "step": 2408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.591e-05,
+      "loss": 1.2591,
+      "step": 2409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.59e-05,
+      "loss": 1.4147,
+      "step": 2410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.589000000000001e-05,
+      "loss": 1.2215,
+      "step": 2411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.588000000000001e-05,
+      "loss": 1.6053,
+      "step": 2412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.587e-05,
+      "loss": 1.1542,
+      "step": 2413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.586000000000001e-05,
+      "loss": 0.5561,
+      "step": 2414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.585e-05,
+      "loss": 0.4328,
+      "step": 2415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.584e-05,
+      "loss": 0.0464,
+      "step": 2416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.583e-05,
+      "loss": 0.4232,
+      "step": 2417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.582e-05,
+      "loss": 0.8551,
+      "step": 2418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.581000000000001e-05,
+      "loss": 0.9492,
+      "step": 2419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.58e-05,
+      "loss": 1.0987,
+      "step": 2420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.579e-05,
+      "loss": 0.905,
+      "step": 2421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.578000000000001e-05,
+      "loss": 1.1566,
+      "step": 2422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.577e-05,
+      "loss": 1.3069,
+      "step": 2423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.576e-05,
+      "loss": 0.9265,
+      "step": 2424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.575e-05,
+      "loss": 1.9256,
+      "step": 2425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.574e-05,
+      "loss": 1.5678,
+      "step": 2426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.573e-05,
+      "loss": 0.6226,
+      "step": 2427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.572e-05,
+      "loss": 0.1403,
+      "step": 2428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.571e-05,
+      "loss": 0.2844,
+      "step": 2429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.570000000000001e-05,
+      "loss": 0.0599,
+      "step": 2430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.569e-05,
+      "loss": 1.021,
+      "step": 2431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.568000000000001e-05,
+      "loss": 1.5568,
+      "step": 2432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.567000000000001e-05,
+      "loss": 1.1643,
+      "step": 2433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.566e-05,
+      "loss": 1.2123,
+      "step": 2434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.565e-05,
+      "loss": 0.9686,
+      "step": 2435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.564e-05,
+      "loss": 1.4012,
+      "step": 2436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.563e-05,
+      "loss": 1.1953,
+      "step": 2437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.562e-05,
+      "loss": 1.1997,
+      "step": 2438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.561e-05,
+      "loss": 0.6182,
+      "step": 2439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.560000000000001e-05,
+      "loss": 0.1308,
+      "step": 2440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.559e-05,
+      "loss": 0.1523,
+      "step": 2441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.558e-05,
+      "loss": 0.1027,
+      "step": 2442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.557000000000001e-05,
+      "loss": 0.711,
+      "step": 2443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.556000000000002e-05,
+      "loss": 1.446,
+      "step": 2444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.555e-05,
+      "loss": 1.3242,
+      "step": 2445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.554e-05,
+      "loss": 1.1352,
+      "step": 2446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.553e-05,
+      "loss": 1.2089,
+      "step": 2447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.552e-05,
+      "loss": 1.1945,
+      "step": 2448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.551e-05,
+      "loss": 1.0187,
+      "step": 2449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.55e-05,
+      "loss": 1.2318,
+      "step": 2450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.549000000000001e-05,
+      "loss": 1.6821,
+      "step": 2451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.548e-05,
+      "loss": 0.15,
+      "step": 2452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.547000000000001e-05,
+      "loss": 0.0197,
+      "step": 2453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.546000000000001e-05,
+      "loss": 0.5903,
+      "step": 2454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.545e-05,
+      "loss": 0.5157,
+      "step": 2455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.544e-05,
+      "loss": 1.3992,
+      "step": 2456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.543e-05,
+      "loss": 0.7129,
+      "step": 2457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.542e-05,
+      "loss": 1.1208,
+      "step": 2458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.541e-05,
+      "loss": 1.4561,
+      "step": 2459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.54e-05,
+      "loss": 1.149,
+      "step": 2460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.539000000000001e-05,
+      "loss": 0.9847,
+      "step": 2461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.538e-05,
+      "loss": 1.1773,
+      "step": 2462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.537e-05,
+      "loss": 1.5129,
+      "step": 2463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.536000000000001e-05,
+      "loss": 0.8911,
+      "step": 2464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.535e-05,
+      "loss": 0.2006,
+      "step": 2465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.534e-05,
+      "loss": 0.1343,
+      "step": 2466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.533e-05,
+      "loss": 0.1572,
+      "step": 2467
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.532e-05,
+      "loss": 0.497,
+      "step": 2468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.531000000000001e-05,
+      "loss": 0.9013,
+      "step": 2469
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.53e-05,
+      "loss": 1.3691,
+      "step": 2470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.529e-05,
+      "loss": 0.9717,
+      "step": 2471
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.528000000000001e-05,
+      "loss": 1.6434,
+      "step": 2472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.527e-05,
+      "loss": 1.1048,
+      "step": 2473
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.526000000000001e-05,
+      "loss": 1.2628,
+      "step": 2474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.525e-05,
+      "loss": 1.2943,
+      "step": 2475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.524e-05,
+      "loss": 1.1986,
+      "step": 2476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.523e-05,
+      "loss": 0.0817,
+      "step": 2477
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.522e-05,
+      "loss": 0.0244,
+      "step": 2478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.521e-05,
+      "loss": 0.3094,
+      "step": 2479
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.52e-05,
+      "loss": 0.4829,
+      "step": 2480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.519e-05,
+      "loss": 1.2385,
+      "step": 2481
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.518000000000001e-05,
+      "loss": 1.4142,
+      "step": 2482
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.517000000000001e-05,
+      "loss": 1.2109,
+      "step": 2483
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.516e-05,
+      "loss": 0.7574,
+      "step": 2484
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.515e-05,
+      "loss": 1.565,
+      "step": 2485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.514e-05,
+      "loss": 1.0242,
+      "step": 2486
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.513e-05,
+      "loss": 0.8786,
+      "step": 2487
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.512e-05,
+      "loss": 1.5154,
+      "step": 2488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.511e-05,
+      "loss": 0.8129,
+      "step": 2489
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.510000000000001e-05,
+      "loss": 0.2194,
+      "step": 2490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.509e-05,
+      "loss": 0.1996,
+      "step": 2491
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.508e-05,
+      "loss": 0.0693,
+      "step": 2492
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.507000000000001e-05,
+      "loss": 0.1179,
+      "step": 2493
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.506e-05,
+      "loss": 0.9615,
+      "step": 2494
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.505e-05,
+      "loss": 1.2832,
+      "step": 2495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.504e-05,
+      "loss": 1.1832,
+      "step": 2496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.503e-05,
+      "loss": 0.9153,
+      "step": 2497
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.502e-05,
+      "loss": 1.2498,
+      "step": 2498
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.501e-05,
+      "loss": 1.4213,
+      "step": 2499
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.4874,
+      "step": 2500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.499e-05,
+      "loss": 1.2463,
+      "step": 2501
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.498e-05,
+      "loss": 0.3451,
+      "step": 2502
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.497000000000001e-05,
+      "loss": 0.0881,
+      "step": 2503
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.496000000000001e-05,
+      "loss": 0.1824,
+      "step": 2504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.495e-05,
+      "loss": 0.5552,
+      "step": 2505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.494e-05,
+      "loss": 0.5189,
+      "step": 2506
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.493e-05,
+      "loss": 1.1593,
+      "step": 2507
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.492000000000001e-05,
+      "loss": 0.9503,
+      "step": 2508
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.491e-05,
+      "loss": 0.8416,
+      "step": 2509
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.49e-05,
+      "loss": 1.1281,
+      "step": 2510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.489000000000001e-05,
+      "loss": 1.4628,
+      "step": 2511
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.488e-05,
+      "loss": 0.9833,
+      "step": 2512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.487e-05,
+      "loss": 1.7454,
+      "step": 2513
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.486000000000001e-05,
+      "loss": 0.9197,
+      "step": 2514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.485e-05,
+      "loss": 0.1637,
+      "step": 2515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.484e-05,
+      "loss": 0.1071,
+      "step": 2516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.483e-05,
+      "loss": 0.0663,
+      "step": 2517
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.482e-05,
+      "loss": 0.5189,
+      "step": 2518
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.481e-05,
+      "loss": 0.7168,
+      "step": 2519
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.48e-05,
+      "loss": 0.9125,
+      "step": 2520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.479000000000001e-05,
+      "loss": 1.2746,
+      "step": 2521
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.478e-05,
+      "loss": 1.3224,
+      "step": 2522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.477e-05,
+      "loss": 1.1976,
+      "step": 2523
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.476000000000001e-05,
+      "loss": 0.804,
+      "step": 2524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.475000000000001e-05,
+      "loss": 1.0654,
+      "step": 2525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.473999999999999e-05,
+      "loss": 0.8428,
+      "step": 2526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.473e-05,
+      "loss": 0.1773,
+      "step": 2527
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.472e-05,
+      "loss": 0.1143,
+      "step": 2528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.471000000000001e-05,
+      "loss": 0.0543,
+      "step": 2529
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.47e-05,
+      "loss": 0.3355,
+      "step": 2530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.469e-05,
+      "loss": 0.7769,
+      "step": 2531
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.468000000000001e-05,
+      "loss": 0.9539,
+      "step": 2532
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.467e-05,
+      "loss": 1.309,
+      "step": 2533
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.466000000000001e-05,
+      "loss": 1.445,
+      "step": 2534
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.465000000000001e-05,
+      "loss": 1.1102,
+      "step": 2535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.464e-05,
+      "loss": 1.4563,
+      "step": 2536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.463e-05,
+      "loss": 1.5694,
+      "step": 2537
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.462e-05,
+      "loss": 1.8791,
+      "step": 2538
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.461e-05,
+      "loss": 0.8273,
+      "step": 2539
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.46e-05,
+      "loss": 0.3719,
+      "step": 2540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.459e-05,
+      "loss": 0.108,
+      "step": 2541
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.458000000000001e-05,
+      "loss": 0.3482,
+      "step": 2542
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.457000000000001e-05,
+      "loss": 0.8084,
+      "step": 2543
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.456e-05,
+      "loss": 1.2452,
+      "step": 2544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.455000000000001e-05,
+      "loss": 0.989,
+      "step": 2545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.454e-05,
+      "loss": 1.4711,
+      "step": 2546
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.453e-05,
+      "loss": 0.8101,
+      "step": 2547
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.452e-05,
+      "loss": 1.2454,
+      "step": 2548
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.451e-05,
+      "loss": 0.8217,
+      "step": 2549
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.450000000000001e-05,
+      "loss": 1.7369,
+      "step": 2550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.449e-05,
+      "loss": 0.6816,
+      "step": 2551
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.448e-05,
+      "loss": 0.3034,
+      "step": 2552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.447000000000001e-05,
+      "loss": 0.1359,
+      "step": 2553
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.446e-05,
+      "loss": 0.0077,
+      "step": 2554
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.445000000000001e-05,
+      "loss": 0.2803,
+      "step": 2555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.444e-05,
+      "loss": 1.311,
+      "step": 2556
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.443e-05,
+      "loss": 1.2976,
+      "step": 2557
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.442e-05,
+      "loss": 1.2696,
+      "step": 2558
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.441e-05,
+      "loss": 1.0781,
+      "step": 2559
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.44e-05,
+      "loss": 0.9991,
+      "step": 2560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.439e-05,
+      "loss": 1.3812,
+      "step": 2561
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.438e-05,
+      "loss": 1.2339,
+      "step": 2562
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.437000000000001e-05,
+      "loss": 1.2408,
+      "step": 2563
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.436000000000001e-05,
+      "loss": 0.6257,
+      "step": 2564
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.435e-05,
+      "loss": 0.1781,
+      "step": 2565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.434e-05,
+      "loss": 0.7016,
+      "step": 2566
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.433e-05,
+      "loss": 0.0682,
+      "step": 2567
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.432e-05,
+      "loss": 0.0313,
+      "step": 2568
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.431e-05,
+      "loss": 1.1641,
+      "step": 2569
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.43e-05,
+      "loss": 1.0182,
+      "step": 2570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.429000000000001e-05,
+      "loss": 1.1021,
+      "step": 2571
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.428e-05,
+      "loss": 1.3371,
+      "step": 2572
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.427e-05,
+      "loss": 1.1474,
+      "step": 2573
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.426000000000001e-05,
+      "loss": 0.8752,
+      "step": 2574
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.425e-05,
+      "loss": 0.9227,
+      "step": 2575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.424e-05,
+      "loss": 1.3778,
+      "step": 2576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.423e-05,
+      "loss": 0.4452,
+      "step": 2577
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.422e-05,
+      "loss": 0.1873,
+      "step": 2578
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.421e-05,
+      "loss": 0.0766,
+      "step": 2579
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.42e-05,
+      "loss": 0.0342,
+      "step": 2580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.419e-05,
+      "loss": 0.8886,
+      "step": 2581
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.418000000000001e-05,
+      "loss": 1.2801,
+      "step": 2582
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.417e-05,
+      "loss": 1.604,
+      "step": 2583
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.416000000000001e-05,
+      "loss": 1.1383,
+      "step": 2584
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.415000000000001e-05,
+      "loss": 0.8572,
+      "step": 2585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.414e-05,
+      "loss": 0.9112,
+      "step": 2586
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.413e-05,
+      "loss": 1.0911,
+      "step": 2587
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.412e-05,
+      "loss": 0.9876,
+      "step": 2588
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.411000000000001e-05,
+      "loss": 0.5822,
+      "step": 2589
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.41e-05,
+      "loss": 0.4355,
+      "step": 2590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.409e-05,
+      "loss": 0.2925,
+      "step": 2591
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.408000000000001e-05,
+      "loss": 0.3497,
+      "step": 2592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.407e-05,
+      "loss": 1.1342,
+      "step": 2593
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.406e-05,
+      "loss": 1.786,
+      "step": 2594
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.405000000000001e-05,
+      "loss": 1.3401,
+      "step": 2595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.404e-05,
+      "loss": 1.1329,
+      "step": 2596
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.403e-05,
+      "loss": 1.303,
+      "step": 2597
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.402e-05,
+      "loss": 0.819,
+      "step": 2598
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.401e-05,
+      "loss": 1.27,
+      "step": 2599
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4e-05,
+      "loss": 1.6199,
+      "step": 2600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.399e-05,
+      "loss": 1.1295,
+      "step": 2601
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.398e-05,
+      "loss": 0.3916,
+      "step": 2602
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.397000000000001e-05,
+      "loss": 0.2175,
+      "step": 2603
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.396e-05,
+      "loss": 0.0671,
+      "step": 2604
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.395000000000001e-05,
+      "loss": 0.0407,
+      "step": 2605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.394e-05,
+      "loss": 0.8839,
+      "step": 2606
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.393e-05,
+      "loss": 0.8485,
+      "step": 2607
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.392e-05,
+      "loss": 1.6846,
+      "step": 2608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.391e-05,
+      "loss": 1.0063,
+      "step": 2609
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.390000000000001e-05,
+      "loss": 1.2594,
+      "step": 2610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.389e-05,
+      "loss": 1.5111,
+      "step": 2611
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.388e-05,
+      "loss": 1.2918,
+      "step": 2612
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.387000000000001e-05,
+      "loss": 1.4181,
+      "step": 2613
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.386e-05,
+      "loss": 0.6632,
+      "step": 2614
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.385e-05,
+      "loss": 0.087,
+      "step": 2615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.384e-05,
+      "loss": 0.0834,
+      "step": 2616
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.383e-05,
+      "loss": 0.0092,
+      "step": 2617
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.382e-05,
+      "loss": 0.2664,
+      "step": 2618
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.381e-05,
+      "loss": 0.9005,
+      "step": 2619
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.38e-05,
+      "loss": 1.4556,
+      "step": 2620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.379000000000001e-05,
+      "loss": 0.9238,
+      "step": 2621
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.378e-05,
+      "loss": 1.2171,
+      "step": 2622
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.377000000000001e-05,
+      "loss": 1.0797,
+      "step": 2623
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.376000000000001e-05,
+      "loss": 1.2645,
+      "step": 2624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.375e-05,
+      "loss": 1.1069,
+      "step": 2625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.374000000000001e-05,
+      "loss": 1.1316,
+      "step": 2626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.373e-05,
+      "loss": 0.2592,
+      "step": 2627
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.372e-05,
+      "loss": 0.1969,
+      "step": 2628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.371e-05,
+      "loss": 0.2398,
+      "step": 2629
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.37e-05,
+      "loss": 0.2272,
+      "step": 2630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.369000000000001e-05,
+      "loss": 1.2483,
+      "step": 2631
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.368e-05,
+      "loss": 1.1542,
+      "step": 2632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.367e-05,
+      "loss": 1.4295,
+      "step": 2633
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.366000000000001e-05,
+      "loss": 1.0263,
+      "step": 2634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.365e-05,
+      "loss": 1.3952,
+      "step": 2635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.364e-05,
+      "loss": 0.9686,
+      "step": 2636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.363e-05,
+      "loss": 1.3465,
+      "step": 2637
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.362e-05,
+      "loss": 1.1777,
+      "step": 2638
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.361e-05,
+      "loss": 0.5581,
+      "step": 2639
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.36e-05,
+      "loss": 0.2397,
+      "step": 2640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.359e-05,
+      "loss": 0.0592,
+      "step": 2641
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.358000000000001e-05,
+      "loss": 0.1871,
+      "step": 2642
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.357e-05,
+      "loss": 0.5897,
+      "step": 2643
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.356000000000001e-05,
+      "loss": 0.9971,
+      "step": 2644
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.355000000000001e-05,
+      "loss": 1.0221,
+      "step": 2645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.354e-05,
+      "loss": 0.7807,
+      "step": 2646
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.353e-05,
+      "loss": 1.3116,
+      "step": 2647
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.352e-05,
+      "loss": 1.2509,
+      "step": 2648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.351e-05,
+      "loss": 1.2463,
+      "step": 2649
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.35e-05,
+      "loss": 0.7865,
+      "step": 2650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.349e-05,
+      "loss": 1.241,
+      "step": 2651
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.348000000000001e-05,
+      "loss": 0.2859,
+      "step": 2652
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.347e-05,
+      "loss": 0.0868,
+      "step": 2653
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.346e-05,
+      "loss": 0.2823,
+      "step": 2654
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.345000000000001e-05,
+      "loss": 0.2016,
+      "step": 2655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.344000000000002e-05,
+      "loss": 1.2945,
+      "step": 2656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.342999999999999e-05,
+      "loss": 1.3753,
+      "step": 2657
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.342e-05,
+      "loss": 1.4703,
+      "step": 2658
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.341e-05,
+      "loss": 1.1301,
+      "step": 2659
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.340000000000001e-05,
+      "loss": 0.8681,
+      "step": 2660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.339e-05,
+      "loss": 1.4355,
+      "step": 2661
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.338e-05,
+      "loss": 1.798,
+      "step": 2662
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.337000000000001e-05,
+      "loss": 1.0591,
+      "step": 2663
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.336e-05,
+      "loss": 0.7405,
+      "step": 2664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.335000000000001e-05,
+      "loss": 0.1381,
+      "step": 2665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.334000000000001e-05,
+      "loss": 0.0213,
+      "step": 2666
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.333e-05,
+      "loss": 0.4748,
+      "step": 2667
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.332e-05,
+      "loss": 0.5425,
+      "step": 2668
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.331e-05,
+      "loss": 1.1139,
+      "step": 2669
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.33e-05,
+      "loss": 0.9646,
+      "step": 2670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.329e-05,
+      "loss": 0.7593,
+      "step": 2671
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.328e-05,
+      "loss": 1.2811,
+      "step": 2672
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.327000000000001e-05,
+      "loss": 1.2473,
+      "step": 2673
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.326e-05,
+      "loss": 1.0035,
+      "step": 2674
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.325e-05,
+      "loss": 1.5062,
+      "step": 2675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.324000000000001e-05,
+      "loss": 1.041,
+      "step": 2676
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.323e-05,
+      "loss": 0.1668,
+      "step": 2677
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.322e-05,
+      "loss": 0.0165,
+      "step": 2678
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.321e-05,
+      "loss": 0.0958,
+      "step": 2679
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.32e-05,
+      "loss": 0.344,
+      "step": 2680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.319000000000001e-05,
+      "loss": 0.9569,
+      "step": 2681
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.318e-05,
+      "loss": 0.882,
+      "step": 2682
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.317e-05,
+      "loss": 1.5347,
+      "step": 2683
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.316000000000001e-05,
+      "loss": 0.9775,
+      "step": 2684
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.315e-05,
+      "loss": 1.0041,
+      "step": 2685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.314000000000001e-05,
+      "loss": 1.2868,
+      "step": 2686
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.313e-05,
+      "loss": 0.7274,
+      "step": 2687
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.312e-05,
+      "loss": 1.6542,
+      "step": 2688
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.311e-05,
+      "loss": 0.6589,
+      "step": 2689
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.31e-05,
+      "loss": 0.0184,
+      "step": 2690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.309e-05,
+      "loss": 0.2317,
+      "step": 2691
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.308e-05,
+      "loss": 0.1102,
+      "step": 2692
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.307e-05,
+      "loss": 0.5365,
+      "step": 2693
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.306000000000001e-05,
+      "loss": 0.7055,
+      "step": 2694
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.305000000000001e-05,
+      "loss": 1.3129,
+      "step": 2695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.304e-05,
+      "loss": 1.2232,
+      "step": 2696
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.303e-05,
+      "loss": 1.1473,
+      "step": 2697
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.302e-05,
+      "loss": 0.9829,
+      "step": 2698
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.301e-05,
+      "loss": 1.5008,
+      "step": 2699
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.3e-05,
+      "loss": 1.1405,
+      "step": 2700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.299e-05,
+      "loss": 1.1359,
+      "step": 2701
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.298000000000001e-05,
+      "loss": 0.6169,
+      "step": 2702
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.297e-05,
+      "loss": 0.321,
+      "step": 2703
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.296e-05,
+      "loss": 0.148,
+      "step": 2704
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.295000000000001e-05,
+      "loss": 0.1565,
+      "step": 2705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.294e-05,
+      "loss": 0.9218,
+      "step": 2706
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.293e-05,
+      "loss": 0.4985,
+      "step": 2707
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.292e-05,
+      "loss": 0.7152,
+      "step": 2708
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.291e-05,
+      "loss": 1.091,
+      "step": 2709
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.29e-05,
+      "loss": 1.3734,
+      "step": 2710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.289e-05,
+      "loss": 1.3239,
+      "step": 2711
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.288000000000001e-05,
+      "loss": 1.3242,
+      "step": 2712
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.287e-05,
+      "loss": 1.5261,
+      "step": 2713
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.286e-05,
+      "loss": 0.9326,
+      "step": 2714
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.285000000000001e-05,
+      "loss": 0.1358,
+      "step": 2715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.284000000000001e-05,
+      "loss": 0.3497,
+      "step": 2716
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.282999999999999e-05,
+      "loss": 0.4554,
+      "step": 2717
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.282e-05,
+      "loss": 1.0374,
+      "step": 2718
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.281e-05,
+      "loss": 0.4934,
+      "step": 2719
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.280000000000001e-05,
+      "loss": 1.3301,
+      "step": 2720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.279e-05,
+      "loss": 0.9917,
+      "step": 2721
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.278e-05,
+      "loss": 1.0011,
+      "step": 2722
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.277000000000001e-05,
+      "loss": 1.1539,
+      "step": 2723
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.276e-05,
+      "loss": 0.5797,
+      "step": 2724
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.275e-05,
+      "loss": 0.9712,
+      "step": 2725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.274000000000001e-05,
+      "loss": 0.9827,
+      "step": 2726
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.273e-05,
+      "loss": 0.1843,
+      "step": 2727
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.272e-05,
+      "loss": 0.0008,
+      "step": 2728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.271e-05,
+      "loss": 0.3239,
+      "step": 2729
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.27e-05,
+      "loss": 0.4411,
+      "step": 2730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.269e-05,
+      "loss": 0.8254,
+      "step": 2731
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.268e-05,
+      "loss": 1.1592,
+      "step": 2732
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.267000000000001e-05,
+      "loss": 1.0772,
+      "step": 2733
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.266000000000001e-05,
+      "loss": 1.5499,
+      "step": 2734
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.265e-05,
+      "loss": 1.5248,
+      "step": 2735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.264000000000001e-05,
+      "loss": 1.5131,
+      "step": 2736
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.263e-05,
+      "loss": 0.9399,
+      "step": 2737
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.261999999999999e-05,
+      "loss": 0.8992,
+      "step": 2738
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.261e-05,
+      "loss": 0.8133,
+      "step": 2739
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.26e-05,
+      "loss": 0.1247,
+      "step": 2740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.259000000000001e-05,
+      "loss": 0.0601,
+      "step": 2741
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.258e-05,
+      "loss": 0.0739,
+      "step": 2742
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.257e-05,
+      "loss": 1.0282,
+      "step": 2743
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.256000000000001e-05,
+      "loss": 1.1978,
+      "step": 2744
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.255e-05,
+      "loss": 1.4751,
+      "step": 2745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.254000000000001e-05,
+      "loss": 1.1205,
+      "step": 2746
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.253e-05,
+      "loss": 1.4281,
+      "step": 2747
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.252e-05,
+      "loss": 1.0587,
+      "step": 2748
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.251e-05,
+      "loss": 1.2318,
+      "step": 2749
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.25e-05,
+      "loss": 0.7207,
+      "step": 2750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.249e-05,
+      "loss": 1.4249,
+      "step": 2751
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.248e-05,
+      "loss": 0.0278,
+      "step": 2752
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.247e-05,
+      "loss": 0.0208,
+      "step": 2753
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.246000000000001e-05,
+      "loss": 0.0955,
+      "step": 2754
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.245000000000001e-05,
+      "loss": 0.7237,
+      "step": 2755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.244e-05,
+      "loss": 0.9644,
+      "step": 2756
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.243000000000001e-05,
+      "loss": 1.5274,
+      "step": 2757
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.242e-05,
+      "loss": 1.1776,
+      "step": 2758
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.241e-05,
+      "loss": 0.8806,
+      "step": 2759
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.24e-05,
+      "loss": 1.2115,
+      "step": 2760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.239e-05,
+      "loss": 1.2432,
+      "step": 2761
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.238000000000001e-05,
+      "loss": 0.9283,
+      "step": 2762
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.237e-05,
+      "loss": 1.299,
+      "step": 2763
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.236e-05,
+      "loss": 0.9026,
+      "step": 2764
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.235000000000001e-05,
+      "loss": 0.343,
+      "step": 2765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.234e-05,
+      "loss": 0.0474,
+      "step": 2766
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.233000000000001e-05,
+      "loss": 0.0914,
+      "step": 2767
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.232e-05,
+      "loss": 0.0176,
+      "step": 2768
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.231e-05,
+      "loss": 0.7801,
+      "step": 2769
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.23e-05,
+      "loss": 1.0561,
+      "step": 2770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.229e-05,
+      "loss": 1.1705,
+      "step": 2771
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.228e-05,
+      "loss": 1.0727,
+      "step": 2772
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.227000000000001e-05,
+      "loss": 0.8632,
+      "step": 2773
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.226e-05,
+      "loss": 0.7731,
+      "step": 2774
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.225000000000001e-05,
+      "loss": 1.1481,
+      "step": 2775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.224000000000001e-05,
+      "loss": 1.3245,
+      "step": 2776
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.223e-05,
+      "loss": 0.2039,
+      "step": 2777
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.222e-05,
+      "loss": 0.1784,
+      "step": 2778
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.221e-05,
+      "loss": 0.2104,
+      "step": 2779
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.22e-05,
+      "loss": 0.7484,
+      "step": 2780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.219e-05,
+      "loss": 0.9105,
+      "step": 2781
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.218e-05,
+      "loss": 1.3495,
+      "step": 2782
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.217000000000001e-05,
+      "loss": 1.3596,
+      "step": 2783
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.216e-05,
+      "loss": 0.9838,
+      "step": 2784
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.215e-05,
+      "loss": 1.2397,
+      "step": 2785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.214000000000001e-05,
+      "loss": 0.9355,
+      "step": 2786
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.213e-05,
+      "loss": 1.0168,
+      "step": 2787
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.212e-05,
+      "loss": 0.9701,
+      "step": 2788
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.211e-05,
+      "loss": 0.4254,
+      "step": 2789
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.21e-05,
+      "loss": 0.0036,
+      "step": 2790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.209e-05,
+      "loss": 0.2863,
+      "step": 2791
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.208e-05,
+      "loss": 0.6822,
+      "step": 2792
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.207e-05,
+      "loss": 0.9814,
+      "step": 2793
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.206000000000001e-05,
+      "loss": 1.1456,
+      "step": 2794
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.205e-05,
+      "loss": 1.0963,
+      "step": 2795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.204000000000001e-05,
+      "loss": 1.4037,
+      "step": 2796
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.203000000000001e-05,
+      "loss": 1.3508,
+      "step": 2797
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.202e-05,
+      "loss": 1.1597,
+      "step": 2798
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.201e-05,
+      "loss": 1.0569,
+      "step": 2799
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2e-05,
+      "loss": 1.2699,
+      "step": 2800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.199000000000001e-05,
+      "loss": 0.7112,
+      "step": 2801
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.198e-05,
+      "loss": 0.1283,
+      "step": 2802
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.197e-05,
+      "loss": 0.237,
+      "step": 2803
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.196000000000001e-05,
+      "loss": 0.0384,
+      "step": 2804
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.195e-05,
+      "loss": 0.3353,
+      "step": 2805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.194e-05,
+      "loss": 0.6909,
+      "step": 2806
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.193000000000001e-05,
+      "loss": 1.3838,
+      "step": 2807
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.192e-05,
+      "loss": 1.2289,
+      "step": 2808
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.191e-05,
+      "loss": 0.9786,
+      "step": 2809
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.19e-05,
+      "loss": 1.1884,
+      "step": 2810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.189e-05,
+      "loss": 1.0168,
+      "step": 2811
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.188e-05,
+      "loss": 1.2434,
+      "step": 2812
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.187e-05,
+      "loss": 0.8349,
+      "step": 2813
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.186e-05,
+      "loss": 1.0741,
+      "step": 2814
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.185000000000001e-05,
+      "loss": 0.1321,
+      "step": 2815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.184e-05,
+      "loss": 0.2275,
+      "step": 2816
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.183000000000001e-05,
+      "loss": 0.018,
+      "step": 2817
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.182e-05,
+      "loss": 0.5663,
+      "step": 2818
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.181e-05,
+      "loss": 0.9791,
+      "step": 2819
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.18e-05,
+      "loss": 1.3973,
+      "step": 2820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.179e-05,
+      "loss": 1.2765,
+      "step": 2821
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.178000000000001e-05,
+      "loss": 1.2097,
+      "step": 2822
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.177e-05,
+      "loss": 1.3243,
+      "step": 2823
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.176e-05,
+      "loss": 1.0863,
+      "step": 2824
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.175000000000001e-05,
+      "loss": 1.2691,
+      "step": 2825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.174e-05,
+      "loss": 1.3249,
+      "step": 2826
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.173e-05,
+      "loss": 0.2748,
+      "step": 2827
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.172e-05,
+      "loss": 0.0606,
+      "step": 2828
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.171e-05,
+      "loss": 0.0885,
+      "step": 2829
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.17e-05,
+      "loss": 0.8587,
+      "step": 2830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.169e-05,
+      "loss": 1.0754,
+      "step": 2831
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.168e-05,
+      "loss": 0.8934,
+      "step": 2832
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.167000000000001e-05,
+      "loss": 1.1504,
+      "step": 2833
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.166e-05,
+      "loss": 1.1163,
+      "step": 2834
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.165000000000001e-05,
+      "loss": 1.1175,
+      "step": 2835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.164000000000001e-05,
+      "loss": 1.5148,
+      "step": 2836
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.163e-05,
+      "loss": 1.1867,
+      "step": 2837
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.162e-05,
+      "loss": 1.3859,
+      "step": 2838
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.161e-05,
+      "loss": 0.8594,
+      "step": 2839
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.16e-05,
+      "loss": 0.1126,
+      "step": 2840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.159e-05,
+      "loss": 0.0548,
+      "step": 2841
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.158e-05,
+      "loss": 0.1801,
+      "step": 2842
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.157000000000001e-05,
+      "loss": 0.682,
+      "step": 2843
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.156e-05,
+      "loss": 1.15,
+      "step": 2844
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.155e-05,
+      "loss": 1.0573,
+      "step": 2845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.154000000000001e-05,
+      "loss": 0.9851,
+      "step": 2846
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.153000000000001e-05,
+      "loss": 1.1601,
+      "step": 2847
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.151999999999999e-05,
+      "loss": 1.2339,
+      "step": 2848
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.151e-05,
+      "loss": 1.2934,
+      "step": 2849
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.15e-05,
+      "loss": 1.1057,
+      "step": 2850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.149e-05,
+      "loss": 1.221,
+      "step": 2851
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.148e-05,
+      "loss": 0.24,
+      "step": 2852
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.147e-05,
+      "loss": 0.0592,
+      "step": 2853
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.146000000000001e-05,
+      "loss": 0.0112,
+      "step": 2854
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.145e-05,
+      "loss": 0.1603,
+      "step": 2855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.144000000000001e-05,
+      "loss": 0.8632,
+      "step": 2856
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.143000000000001e-05,
+      "loss": 1.4177,
+      "step": 2857
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.142e-05,
+      "loss": 1.3648,
+      "step": 2858
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.141e-05,
+      "loss": 1.2433,
+      "step": 2859
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.14e-05,
+      "loss": 1.3918,
+      "step": 2860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.139e-05,
+      "loss": 1.5406,
+      "step": 2861
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.138e-05,
+      "loss": 1.6734,
+      "step": 2862
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.137e-05,
+      "loss": 1.7578,
+      "step": 2863
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.136000000000001e-05,
+      "loss": 0.5267,
+      "step": 2864
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.135e-05,
+      "loss": 0.2118,
+      "step": 2865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.134e-05,
+      "loss": 0.0527,
+      "step": 2866
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.133000000000001e-05,
+      "loss": 0.1712,
+      "step": 2867
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.132e-05,
+      "loss": 0.4566,
+      "step": 2868
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.130999999999999e-05,
+      "loss": 1.4931,
+      "step": 2869
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.13e-05,
+      "loss": 1.2985,
+      "step": 2870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.129e-05,
+      "loss": 1.1328,
+      "step": 2871
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.128000000000001e-05,
+      "loss": 1.366,
+      "step": 2872
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.127e-05,
+      "loss": 1.1606,
+      "step": 2873
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.126e-05,
+      "loss": 1.0772,
+      "step": 2874
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.125000000000001e-05,
+      "loss": 0.7631,
+      "step": 2875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.124e-05,
+      "loss": 1.1462,
+      "step": 2876
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.123000000000001e-05,
+      "loss": 0.2155,
+      "step": 2877
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.122000000000001e-05,
+      "loss": 0.2108,
+      "step": 2878
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.121e-05,
+      "loss": 0.0154,
+      "step": 2879
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.12e-05,
+      "loss": 0.045,
+      "step": 2880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.119e-05,
+      "loss": 0.9511,
+      "step": 2881
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.118e-05,
+      "loss": 1.182,
+      "step": 2882
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.117e-05,
+      "loss": 1.3965,
+      "step": 2883
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.116e-05,
+      "loss": 1.3154,
+      "step": 2884
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.115000000000001e-05,
+      "loss": 1.4955,
+      "step": 2885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.114e-05,
+      "loss": 1.3237,
+      "step": 2886
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.113e-05,
+      "loss": 0.9018,
+      "step": 2887
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.112000000000001e-05,
+      "loss": 1.1523,
+      "step": 2888
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.111e-05,
+      "loss": 0.7058,
+      "step": 2889
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.11e-05,
+      "loss": 0.2866,
+      "step": 2890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.109e-05,
+      "loss": 0.0271,
+      "step": 2891
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.108e-05,
+      "loss": 0.0185,
+      "step": 2892
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.107000000000001e-05,
+      "loss": 0.4383,
+      "step": 2893
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.106e-05,
+      "loss": 0.8376,
+      "step": 2894
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.105e-05,
+      "loss": 0.8696,
+      "step": 2895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.104000000000001e-05,
+      "loss": 0.9568,
+      "step": 2896
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.103e-05,
+      "loss": 1.0828,
+      "step": 2897
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.102000000000001e-05,
+      "loss": 0.939,
+      "step": 2898
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.101e-05,
+      "loss": 1.2224,
+      "step": 2899
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.1e-05,
+      "loss": 1.1893,
+      "step": 2900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.099e-05,
+      "loss": 0.7299,
+      "step": 2901
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.098e-05,
+      "loss": 0.0633,
+      "step": 2902
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.097e-05,
+      "loss": 0.1157,
+      "step": 2903
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.096e-05,
+      "loss": 0.1057,
+      "step": 2904
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.095e-05,
+      "loss": 0.3563,
+      "step": 2905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.094000000000001e-05,
+      "loss": 1.2963,
+      "step": 2906
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.093000000000001e-05,
+      "loss": 1.4765,
+      "step": 2907
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.092e-05,
+      "loss": 0.9029,
+      "step": 2908
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.091e-05,
+      "loss": 1.5932,
+      "step": 2909
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.09e-05,
+      "loss": 1.1804,
+      "step": 2910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.089000000000001e-05,
+      "loss": 1.4417,
+      "step": 2911
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.088e-05,
+      "loss": 1.0615,
+      "step": 2912
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.087e-05,
+      "loss": 1.2907,
+      "step": 2913
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.086000000000001e-05,
+      "loss": 0.8021,
+      "step": 2914
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.085e-05,
+      "loss": 0.1911,
+      "step": 2915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.084e-05,
+      "loss": 0.1728,
+      "step": 2916
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.083000000000001e-05,
+      "loss": 0.7068,
+      "step": 2917
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.082e-05,
+      "loss": 0.784,
+      "step": 2918
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.081e-05,
+      "loss": 1.1691,
+      "step": 2919
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.08e-05,
+      "loss": 1.2612,
+      "step": 2920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.079e-05,
+      "loss": 1.4245,
+      "step": 2921
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.078e-05,
+      "loss": 1.1237,
+      "step": 2922
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.077e-05,
+      "loss": 0.8343,
+      "step": 2923
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.076000000000001e-05,
+      "loss": 1.1656,
+      "step": 2924
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.075e-05,
+      "loss": 1.6142,
+      "step": 2925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.074e-05,
+      "loss": 1.2029,
+      "step": 2926
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.073000000000001e-05,
+      "loss": 0.2963,
+      "step": 2927
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.072000000000001e-05,
+      "loss": 0.2404,
+      "step": 2928
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.070999999999999e-05,
+      "loss": 0.1851,
+      "step": 2929
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.07e-05,
+      "loss": 0.1959,
+      "step": 2930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.069e-05,
+      "loss": 0.4089,
+      "step": 2931
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.068000000000001e-05,
+      "loss": 1.3827,
+      "step": 2932
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.067e-05,
+      "loss": 1.0873,
+      "step": 2933
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.066e-05,
+      "loss": 1.2284,
+      "step": 2934
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.065000000000001e-05,
+      "loss": 0.8464,
+      "step": 2935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.064e-05,
+      "loss": 1.3404,
+      "step": 2936
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.063e-05,
+      "loss": 1.2681,
+      "step": 2937
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.062000000000001e-05,
+      "loss": 1.1253,
+      "step": 2938
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.061e-05,
+      "loss": 0.9386,
+      "step": 2939
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.06e-05,
+      "loss": 0.1224,
+      "step": 2940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.059e-05,
+      "loss": 0.1956,
+      "step": 2941
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.058e-05,
+      "loss": 0.2247,
+      "step": 2942
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.057e-05,
+      "loss": 0.2155,
+      "step": 2943
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.056e-05,
+      "loss": 0.6795,
+      "step": 2944
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.055000000000001e-05,
+      "loss": 0.609,
+      "step": 2945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.054000000000001e-05,
+      "loss": 1.2817,
+      "step": 2946
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.053e-05,
+      "loss": 1.3034,
+      "step": 2947
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.052000000000001e-05,
+      "loss": 1.1448,
+      "step": 2948
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.051e-05,
+      "loss": 1.41,
+      "step": 2949
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.05e-05,
+      "loss": 1.4401,
+      "step": 2950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.049e-05,
+      "loss": 0.9185,
+      "step": 2951
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.048e-05,
+      "loss": 0.3845,
+      "step": 2952
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.047000000000001e-05,
+      "loss": 0.0055,
+      "step": 2953
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.046e-05,
+      "loss": 0.4295,
+      "step": 2954
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.045e-05,
+      "loss": 0.3707,
+      "step": 2955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.044000000000001e-05,
+      "loss": 0.4239,
+      "step": 2956
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.043e-05,
+      "loss": 0.9958,
+      "step": 2957
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.042000000000001e-05,
+      "loss": 1.0596,
+      "step": 2958
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.041e-05,
+      "loss": 1.0059,
+      "step": 2959
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.04e-05,
+      "loss": 1.1784,
+      "step": 2960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.039e-05,
+      "loss": 1.3469,
+      "step": 2961
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.038e-05,
+      "loss": 0.8309,
+      "step": 2962
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.037e-05,
+      "loss": 1.6904,
+      "step": 2963
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.036e-05,
+      "loss": 0.7334,
+      "step": 2964
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.035e-05,
+      "loss": 0.7504,
+      "step": 2965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.034000000000001e-05,
+      "loss": 0.1185,
+      "step": 2966
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.033000000000001e-05,
+      "loss": 0.0383,
+      "step": 2967
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.032e-05,
+      "loss": 1.0591,
+      "step": 2968
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.031e-05,
+      "loss": 0.9483,
+      "step": 2969
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.03e-05,
+      "loss": 1.0115,
+      "step": 2970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.029e-05,
+      "loss": 0.5765,
+      "step": 2971
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.028e-05,
+      "loss": 1.049,
+      "step": 2972
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.027e-05,
+      "loss": 0.9468,
+      "step": 2973
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.026000000000001e-05,
+      "loss": 1.2788,
+      "step": 2974
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.025e-05,
+      "loss": 0.8278,
+      "step": 2975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.024e-05,
+      "loss": 1.3906,
+      "step": 2976
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.023000000000001e-05,
+      "loss": 0.4114,
+      "step": 2977
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.022e-05,
+      "loss": 0.3043,
+      "step": 2978
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.021e-05,
+      "loss": 0.0468,
+      "step": 2979
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.02e-05,
+      "loss": 0.2738,
+      "step": 2980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.019e-05,
+      "loss": 1.1189,
+      "step": 2981
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.018e-05,
+      "loss": 1.1767,
+      "step": 2982
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.017e-05,
+      "loss": 1.6103,
+      "step": 2983
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.016e-05,
+      "loss": 1.4296,
+      "step": 2984
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.015000000000001e-05,
+      "loss": 1.1159,
+      "step": 2985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.014e-05,
+      "loss": 0.8931,
+      "step": 2986
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.013000000000001e-05,
+      "loss": 0.6368,
+      "step": 2987
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.012000000000001e-05,
+      "loss": 1.5924,
+      "step": 2988
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.011e-05,
+      "loss": 0.9207,
+      "step": 2989
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.01e-05,
+      "loss": 0.3364,
+      "step": 2990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.009e-05,
+      "loss": 0.137,
+      "step": 2991
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.008e-05,
+      "loss": 0.1189,
+      "step": 2992
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.007e-05,
+      "loss": 0.467,
+      "step": 2993
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.006e-05,
+      "loss": 1.3476,
+      "step": 2994
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.005000000000001e-05,
+      "loss": 0.9044,
+      "step": 2995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.004e-05,
+      "loss": 1.3388,
+      "step": 2996
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.003e-05,
+      "loss": 1.3082,
+      "step": 2997
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.002000000000001e-05,
+      "loss": 1.135,
+      "step": 2998
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.001e-05,
+      "loss": 1.1362,
+      "step": 2999
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7e-05,
+      "loss": 1.0182,
+      "step": 3000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.999e-05,
+      "loss": 1.3128,
+      "step": 3001
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.998e-05,
+      "loss": 0.247,
+      "step": 3002
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.997e-05,
+      "loss": 0.1242,
+      "step": 3003
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.996e-05,
+      "loss": 0.488,
+      "step": 3004
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.995e-05,
+      "loss": 0.2147,
+      "step": 3005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.994000000000001e-05,
+      "loss": 1.3262,
+      "step": 3006
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.993e-05,
+      "loss": 1.1921,
+      "step": 3007
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.992000000000001e-05,
+      "loss": 1.3178,
+      "step": 3008
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.991000000000001e-05,
+      "loss": 0.9348,
+      "step": 3009
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.99e-05,
+      "loss": 1.0817,
+      "step": 3010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.989e-05,
+      "loss": 1.1232,
+      "step": 3011
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.988e-05,
+      "loss": 1.1186,
+      "step": 3012
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.987000000000001e-05,
+      "loss": 1.5223,
+      "step": 3013
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.986e-05,
+      "loss": 0.7574,
+      "step": 3014
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.985e-05,
+      "loss": 0.2122,
+      "step": 3015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.984000000000001e-05,
+      "loss": 0.0842,
+      "step": 3016
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.983e-05,
+      "loss": 0.4928,
+      "step": 3017
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.982e-05,
+      "loss": 0.4311,
+      "step": 3018
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.981000000000001e-05,
+      "loss": 1.2784,
+      "step": 3019
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.98e-05,
+      "loss": 1.3856,
+      "step": 3020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.979e-05,
+      "loss": 0.9823,
+      "step": 3021
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.978e-05,
+      "loss": 1.1406,
+      "step": 3022
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.977e-05,
+      "loss": 1.1377,
+      "step": 3023
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.976000000000001e-05,
+      "loss": 1.3914,
+      "step": 3024
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.975e-05,
+      "loss": 1.644,
+      "step": 3025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.974e-05,
+      "loss": 0.8921,
+      "step": 3026
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.973000000000001e-05,
+      "loss": 0.3156,
+      "step": 3027
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.972e-05,
+      "loss": 0.138,
+      "step": 3028
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.971000000000001e-05,
+      "loss": 0.1602,
+      "step": 3029
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.97e-05,
+      "loss": 0.5974,
+      "step": 3030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.969e-05,
+      "loss": 1.1345,
+      "step": 3031
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.968e-05,
+      "loss": 1.2663,
+      "step": 3032
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.967e-05,
+      "loss": 1.147,
+      "step": 3033
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.966000000000001e-05,
+      "loss": 1.1574,
+      "step": 3034
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.965e-05,
+      "loss": 1.1471,
+      "step": 3035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.964e-05,
+      "loss": 1.2572,
+      "step": 3036
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.963000000000001e-05,
+      "loss": 1.2755,
+      "step": 3037
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.962e-05,
+      "loss": 1.3158,
+      "step": 3038
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.961e-05,
+      "loss": 0.3835,
+      "step": 3039
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.96e-05,
+      "loss": 0.1349,
+      "step": 3040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.959e-05,
+      "loss": 0.0075,
+      "step": 3041
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.958e-05,
+      "loss": 0.0355,
+      "step": 3042
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.957e-05,
+      "loss": 1.1706,
+      "step": 3043
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.956e-05,
+      "loss": 1.1451,
+      "step": 3044
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.955000000000001e-05,
+      "loss": 1.0254,
+      "step": 3045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.954e-05,
+      "loss": 1.2095,
+      "step": 3046
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.953000000000001e-05,
+      "loss": 1.1437,
+      "step": 3047
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.952000000000001e-05,
+      "loss": 1.0994,
+      "step": 3048
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.951e-05,
+      "loss": 1.4262,
+      "step": 3049
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.95e-05,
+      "loss": 1.4812,
+      "step": 3050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.949e-05,
+      "loss": 1.2608,
+      "step": 3051
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.948e-05,
+      "loss": 0.4843,
+      "step": 3052
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.947e-05,
+      "loss": 0.093,
+      "step": 3053
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.946e-05,
+      "loss": 0.2608,
+      "step": 3054
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.945000000000001e-05,
+      "loss": 0.2209,
+      "step": 3055
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.944e-05,
+      "loss": 0.6774,
+      "step": 3056
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.943e-05,
+      "loss": 1.5472,
+      "step": 3057
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.942000000000001e-05,
+      "loss": 1.3523,
+      "step": 3058
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.941000000000001e-05,
+      "loss": 1.399,
+      "step": 3059
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 1.3329,
+      "step": 3060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.939e-05,
+      "loss": 1.1772,
+      "step": 3061
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.938e-05,
+      "loss": 1.1099,
+      "step": 3062
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.937000000000001e-05,
+      "loss": 1.348,
+      "step": 3063
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.936e-05,
+      "loss": 0.5422,
+      "step": 3064
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.935e-05,
+      "loss": 0.1836,
+      "step": 3065
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.934000000000001e-05,
+      "loss": 0.0174,
+      "step": 3066
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.933e-05,
+      "loss": 0.1068,
+      "step": 3067
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.932000000000001e-05,
+      "loss": 0.7414,
+      "step": 3068
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.931000000000001e-05,
+      "loss": 1.1677,
+      "step": 3069
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.93e-05,
+      "loss": 1.3427,
+      "step": 3070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.929e-05,
+      "loss": 1.3134,
+      "step": 3071
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.928e-05,
+      "loss": 1.3832,
+      "step": 3072
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.927e-05,
+      "loss": 1.3928,
+      "step": 3073
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.926e-05,
+      "loss": 0.9386,
+      "step": 3074
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.925e-05,
+      "loss": 1.2061,
+      "step": 3075
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.924000000000001e-05,
+      "loss": 1.0843,
+      "step": 3076
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.923e-05,
+      "loss": 0.0708,
+      "step": 3077
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.922e-05,
+      "loss": 0.104,
+      "step": 3078
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.921000000000001e-05,
+      "loss": 0.1634,
+      "step": 3079
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.92e-05,
+      "loss": 0.1968,
+      "step": 3080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.918999999999999e-05,
+      "loss": 1.2095,
+      "step": 3081
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.918e-05,
+      "loss": 0.9806,
+      "step": 3082
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.917e-05,
+      "loss": 0.9172,
+      "step": 3083
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.916000000000001e-05,
+      "loss": 0.9452,
+      "step": 3084
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.915e-05,
+      "loss": 1.1417,
+      "step": 3085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.914e-05,
+      "loss": 0.8134,
+      "step": 3086
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.913000000000001e-05,
+      "loss": 1.1967,
+      "step": 3087
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.912e-05,
+      "loss": 1.3486,
+      "step": 3088
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.911000000000001e-05,
+      "loss": 0.6399,
+      "step": 3089
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.91e-05,
+      "loss": 0.1069,
+      "step": 3090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.909e-05,
+      "loss": 0.1848,
+      "step": 3091
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.908e-05,
+      "loss": 0.0635,
+      "step": 3092
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.907e-05,
+      "loss": 0.6225,
+      "step": 3093
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.906e-05,
+      "loss": 1.0677,
+      "step": 3094
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.905e-05,
+      "loss": 0.964,
+      "step": 3095
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.904e-05,
+      "loss": 0.807,
+      "step": 3096
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.903000000000001e-05,
+      "loss": 1.1809,
+      "step": 3097
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.902000000000001e-05,
+      "loss": 1.2624,
+      "step": 3098
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.901e-05,
+      "loss": 1.0162,
+      "step": 3099
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9e-05,
+      "loss": 1.8074,
+      "step": 3100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.899e-05,
+      "loss": 1.4921,
+      "step": 3101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.898e-05,
+      "loss": 0.3439,
+      "step": 3102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.897e-05,
+      "loss": 0.2742,
+      "step": 3103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.896e-05,
+      "loss": 0.0149,
+      "step": 3104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.895000000000001e-05,
+      "loss": 0.3561,
+      "step": 3105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.894e-05,
+      "loss": 0.8228,
+      "step": 3106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.893e-05,
+      "loss": 1.1877,
+      "step": 3107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.892000000000001e-05,
+      "loss": 0.9388,
+      "step": 3108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.891e-05,
+      "loss": 1.6292,
+      "step": 3109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.89e-05,
+      "loss": 1.0561,
+      "step": 3110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.889e-05,
+      "loss": 1.4045,
+      "step": 3111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.888e-05,
+      "loss": 0.6656,
+      "step": 3112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.887e-05,
+      "loss": 0.9525,
+      "step": 3113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.886e-05,
+      "loss": 0.7491,
+      "step": 3114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.885e-05,
+      "loss": 0.3593,
+      "step": 3115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.884e-05,
+      "loss": 0.009,
+      "step": 3116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.883e-05,
+      "loss": 0.4011,
+      "step": 3117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.882000000000001e-05,
+      "loss": 0.6316,
+      "step": 3118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.881000000000001e-05,
+      "loss": 1.4138,
+      "step": 3119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.879999999999999e-05,
+      "loss": 1.1558,
+      "step": 3120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.879e-05,
+      "loss": 1.1121,
+      "step": 3121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.878e-05,
+      "loss": 1.4021,
+      "step": 3122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.877000000000001e-05,
+      "loss": 1.0586,
+      "step": 3123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.876e-05,
+      "loss": 1.2663,
+      "step": 3124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.875e-05,
+      "loss": 1.3868,
+      "step": 3125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.874000000000001e-05,
+      "loss": 1.0237,
+      "step": 3126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.873e-05,
+      "loss": 0.1386,
+      "step": 3127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.872e-05,
+      "loss": 0.1431,
+      "step": 3128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.871000000000001e-05,
+      "loss": 0.5624,
+      "step": 3129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.87e-05,
+      "loss": 0.9009,
+      "step": 3130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.869e-05,
+      "loss": 1.4253,
+      "step": 3131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.868e-05,
+      "loss": 1.2596,
+      "step": 3132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.867e-05,
+      "loss": 1.4735,
+      "step": 3133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.866e-05,
+      "loss": 1.0511,
+      "step": 3134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.865e-05,
+      "loss": 1.3217,
+      "step": 3135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.864000000000001e-05,
+      "loss": 1.8113,
+      "step": 3136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.863000000000001e-05,
+      "loss": 0.9743,
+      "step": 3137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.862e-05,
+      "loss": 1.2209,
+      "step": 3138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.861000000000001e-05,
+      "loss": 0.5694,
+      "step": 3139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.860000000000001e-05,
+      "loss": 0.423,
+      "step": 3140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.858999999999999e-05,
+      "loss": 0.049,
+      "step": 3141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.858e-05,
+      "loss": 0.1017,
+      "step": 3142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.857e-05,
+      "loss": 0.5166,
+      "step": 3143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.856000000000001e-05,
+      "loss": 0.9474,
+      "step": 3144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.855e-05,
+      "loss": 1.0497,
+      "step": 3145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.854e-05,
+      "loss": 1.3921,
+      "step": 3146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.853000000000001e-05,
+      "loss": 1.3406,
+      "step": 3147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.852e-05,
+      "loss": 1.3329,
+      "step": 3148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.851e-05,
+      "loss": 1.2793,
+      "step": 3149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.850000000000001e-05,
+      "loss": 1.1274,
+      "step": 3150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.849e-05,
+      "loss": 0.7589,
+      "step": 3151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.848e-05,
+      "loss": 0.279,
+      "step": 3152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.847e-05,
+      "loss": 0.1723,
+      "step": 3153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.846e-05,
+      "loss": 0.1969,
+      "step": 3154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.845e-05,
+      "loss": 0.4211,
+      "step": 3155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.844e-05,
+      "loss": 1.0248,
+      "step": 3156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.843000000000001e-05,
+      "loss": 1.5082,
+      "step": 3157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.842000000000001e-05,
+      "loss": 1.1842,
+      "step": 3158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.841e-05,
+      "loss": 1.286,
+      "step": 3159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.840000000000001e-05,
+      "loss": 1.4788,
+      "step": 3160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.839e-05,
+      "loss": 1.2866,
+      "step": 3161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.838e-05,
+      "loss": 1.2152,
+      "step": 3162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.837e-05,
+      "loss": 1.1848,
+      "step": 3163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.836e-05,
+      "loss": 0.7632,
+      "step": 3164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.835000000000001e-05,
+      "loss": 0.0301,
+      "step": 3165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.834e-05,
+      "loss": 0.0726,
+      "step": 3166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.833e-05,
+      "loss": 0.0319,
+      "step": 3167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.832000000000001e-05,
+      "loss": 0.1349,
+      "step": 3168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.831e-05,
+      "loss": 1.0649,
+      "step": 3169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.83e-05,
+      "loss": 0.9178,
+      "step": 3170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.829e-05,
+      "loss": 1.0248,
+      "step": 3171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.828e-05,
+      "loss": 1.3006,
+      "step": 3172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.827e-05,
+      "loss": 1.3207,
+      "step": 3173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.826e-05,
+      "loss": 1.259,
+      "step": 3174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.825e-05,
+      "loss": 1.729,
+      "step": 3175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.824e-05,
+      "loss": 1.4994,
+      "step": 3176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.823e-05,
+      "loss": 0.2615,
+      "step": 3177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.822000000000001e-05,
+      "loss": 0.0045,
+      "step": 3178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.821000000000001e-05,
+      "loss": 0.1213,
+      "step": 3179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.82e-05,
+      "loss": 0.5264,
+      "step": 3180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.819e-05,
+      "loss": 1.1289,
+      "step": 3181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.818e-05,
+      "loss": 1.1756,
+      "step": 3182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.817e-05,
+      "loss": 0.6686,
+      "step": 3183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.816e-05,
+      "loss": 1.3247,
+      "step": 3184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.815e-05,
+      "loss": 1.3374,
+      "step": 3185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.814000000000001e-05,
+      "loss": 1.4817,
+      "step": 3186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.813e-05,
+      "loss": 0.8699,
+      "step": 3187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.812e-05,
+      "loss": 1.1876,
+      "step": 3188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.811000000000001e-05,
+      "loss": 0.7546,
+      "step": 3189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.81e-05,
+      "loss": 0.008,
+      "step": 3190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.809e-05,
+      "loss": 0.1306,
+      "step": 3191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.808e-05,
+      "loss": 0.0307,
+      "step": 3192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.807e-05,
+      "loss": 0.6872,
+      "step": 3193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.806e-05,
+      "loss": 0.7602,
+      "step": 3194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.805e-05,
+      "loss": 1.1518,
+      "step": 3195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.804e-05,
+      "loss": 1.233,
+      "step": 3196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.803000000000001e-05,
+      "loss": 1.015,
+      "step": 3197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.802e-05,
+      "loss": 1.028,
+      "step": 3198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.801000000000001e-05,
+      "loss": 0.6933,
+      "step": 3199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.0891,
+      "step": 3200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.799e-05,
+      "loss": 1.0026,
+      "step": 3201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.798e-05,
+      "loss": 0.1516,
+      "step": 3202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.797e-05,
+      "loss": 0.0493,
+      "step": 3203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.796e-05,
+      "loss": 0.2567,
+      "step": 3204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.795e-05,
+      "loss": 0.5551,
+      "step": 3205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.794e-05,
+      "loss": 0.6551,
+      "step": 3206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.793000000000001e-05,
+      "loss": 1.2259,
+      "step": 3207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.792e-05,
+      "loss": 1.0459,
+      "step": 3208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.791e-05,
+      "loss": 0.9301,
+      "step": 3209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.790000000000001e-05,
+      "loss": 1.1111,
+      "step": 3210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.789e-05,
+      "loss": 1.4145,
+      "step": 3211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.788e-05,
+      "loss": 0.7389,
+      "step": 3212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.787e-05,
+      "loss": 1.6124,
+      "step": 3213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.786e-05,
+      "loss": 0.4751,
+      "step": 3214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.785e-05,
+      "loss": 0.1625,
+      "step": 3215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.784e-05,
+      "loss": 0.1543,
+      "step": 3216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.783e-05,
+      "loss": 0.5718,
+      "step": 3217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.782000000000001e-05,
+      "loss": 1.0154,
+      "step": 3218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.781e-05,
+      "loss": 0.6783,
+      "step": 3219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 1.0661,
+      "step": 3220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.779e-05,
+      "loss": 1.0589,
+      "step": 3221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.778e-05,
+      "loss": 1.3764,
+      "step": 3222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.777e-05,
+      "loss": 1.0537,
+      "step": 3223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.776e-05,
+      "loss": 1.4046,
+      "step": 3224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.775000000000001e-05,
+      "loss": 1.5702,
+      "step": 3225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.774e-05,
+      "loss": 1.1362,
+      "step": 3226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.773e-05,
+      "loss": 0.3198,
+      "step": 3227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.772000000000001e-05,
+      "loss": 0.2661,
+      "step": 3228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.771e-05,
+      "loss": 0.04,
+      "step": 3229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.77e-05,
+      "loss": 0.23,
+      "step": 3230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.769e-05,
+      "loss": 0.6693,
+      "step": 3231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.768e-05,
+      "loss": 1.1212,
+      "step": 3232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.767e-05,
+      "loss": 1.2811,
+      "step": 3233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.766e-05,
+      "loss": 1.329,
+      "step": 3234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.765e-05,
+      "loss": 1.3165,
+      "step": 3235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.764000000000001e-05,
+      "loss": 1.2247,
+      "step": 3236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.763e-05,
+      "loss": 1.0602,
+      "step": 3237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.762e-05,
+      "loss": 0.7676,
+      "step": 3238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.761000000000001e-05,
+      "loss": 0.6439,
+      "step": 3239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.76e-05,
+      "loss": 0.0563,
+      "step": 3240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.759e-05,
+      "loss": 0.0284,
+      "step": 3241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.758e-05,
+      "loss": 0.0404,
+      "step": 3242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.757e-05,
+      "loss": 1.1795,
+      "step": 3243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.756e-05,
+      "loss": 1.1522,
+      "step": 3244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.755e-05,
+      "loss": 1.5745,
+      "step": 3245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.754000000000001e-05,
+      "loss": 1.0022,
+      "step": 3246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.753e-05,
+      "loss": 1.1334,
+      "step": 3247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.752e-05,
+      "loss": 1.1882,
+      "step": 3248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.751000000000001e-05,
+      "loss": 1.1063,
+      "step": 3249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.750000000000001e-05,
+      "loss": 1.7736,
+      "step": 3250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.749e-05,
+      "loss": 1.0658,
+      "step": 3251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.748e-05,
+      "loss": 0.3608,
+      "step": 3252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.747e-05,
+      "loss": 0.3445,
+      "step": 3253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.746e-05,
+      "loss": 0.3201,
+      "step": 3254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.745e-05,
+      "loss": 0.0823,
+      "step": 3255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.744e-05,
+      "loss": 1.0223,
+      "step": 3256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.743000000000001e-05,
+      "loss": 1.3916,
+      "step": 3257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.742e-05,
+      "loss": 0.9534,
+      "step": 3258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.741000000000001e-05,
+      "loss": 1.3283,
+      "step": 3259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.740000000000001e-05,
+      "loss": 1.491,
+      "step": 3260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.739e-05,
+      "loss": 1.2873,
+      "step": 3261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.738e-05,
+      "loss": 1.036,
+      "step": 3262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.737e-05,
+      "loss": 1.4889,
+      "step": 3263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.736e-05,
+      "loss": 0.7304,
+      "step": 3264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.735e-05,
+      "loss": 0.2808,
+      "step": 3265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.734e-05,
+      "loss": 0.2121,
+      "step": 3266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.733000000000001e-05,
+      "loss": 0.2984,
+      "step": 3267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.732e-05,
+      "loss": 0.3536,
+      "step": 3268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.731e-05,
+      "loss": 1.0578,
+      "step": 3269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.730000000000001e-05,
+      "loss": 0.7704,
+      "step": 3270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.729000000000001e-05,
+      "loss": 0.9383,
+      "step": 3271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.727999999999999e-05,
+      "loss": 1.1327,
+      "step": 3272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.727e-05,
+      "loss": 1.2421,
+      "step": 3273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.726e-05,
+      "loss": 1.6608,
+      "step": 3274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.725000000000001e-05,
+      "loss": 1.685,
+      "step": 3275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.724e-05,
+      "loss": 1.2326,
+      "step": 3276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.723e-05,
+      "loss": 0.4179,
+      "step": 3277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.722000000000001e-05,
+      "loss": 0.1173,
+      "step": 3278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.721e-05,
+      "loss": 0.5692,
+      "step": 3279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.720000000000001e-05,
+      "loss": 0.0311,
+      "step": 3280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.719000000000001e-05,
+      "loss": 0.454,
+      "step": 3281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.718e-05,
+      "loss": 1.3903,
+      "step": 3282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.717e-05,
+      "loss": 1.2015,
+      "step": 3283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.716e-05,
+      "loss": 1.4637,
+      "step": 3284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.715e-05,
+      "loss": 1.7414,
+      "step": 3285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.714e-05,
+      "loss": 1.1216,
+      "step": 3286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.713e-05,
+      "loss": 1.1486,
+      "step": 3287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.712000000000001e-05,
+      "loss": 0.6141,
+      "step": 3288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.711e-05,
+      "loss": 0.5475,
+      "step": 3289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.71e-05,
+      "loss": 0.2917,
+      "step": 3290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.709000000000001e-05,
+      "loss": 0.2657,
+      "step": 3291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.708e-05,
+      "loss": 0.2854,
+      "step": 3292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.706999999999999e-05,
+      "loss": 1.1177,
+      "step": 3293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.706e-05,
+      "loss": 1.2068,
+      "step": 3294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.705e-05,
+      "loss": 0.7299,
+      "step": 3295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.704000000000001e-05,
+      "loss": 1.2547,
+      "step": 3296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.703e-05,
+      "loss": 1.8132,
+      "step": 3297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.702e-05,
+      "loss": 1.1472,
+      "step": 3298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.701000000000001e-05,
+      "loss": 1.0462,
+      "step": 3299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.7e-05,
+      "loss": 0.7898,
+      "step": 3300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.699000000000001e-05,
+      "loss": 1.0222,
+      "step": 3301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.698e-05,
+      "loss": 0.3866,
+      "step": 3302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.697e-05,
+      "loss": 0.2106,
+      "step": 3303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.696e-05,
+      "loss": 0.0772,
+      "step": 3304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.695e-05,
+      "loss": 0.0224,
+      "step": 3305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.694e-05,
+      "loss": 0.8954,
+      "step": 3306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.693e-05,
+      "loss": 1.1468,
+      "step": 3307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.692e-05,
+      "loss": 1.3712,
+      "step": 3308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.691000000000001e-05,
+      "loss": 1.4463,
+      "step": 3309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.690000000000001e-05,
+      "loss": 1.2349,
+      "step": 3310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.689e-05,
+      "loss": 1.0182,
+      "step": 3311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.688e-05,
+      "loss": 1.1124,
+      "step": 3312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.687e-05,
+      "loss": 1.2382,
+      "step": 3313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.686000000000001e-05,
+      "loss": 0.7216,
+      "step": 3314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.685e-05,
+      "loss": 0.0566,
+      "step": 3315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.684e-05,
+      "loss": 0.0177,
+      "step": 3316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.683000000000001e-05,
+      "loss": 0.0448,
+      "step": 3317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.682e-05,
+      "loss": 0.4696,
+      "step": 3318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.681e-05,
+      "loss": 1.0143,
+      "step": 3319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.680000000000001e-05,
+      "loss": 1.1933,
+      "step": 3320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.679e-05,
+      "loss": 1.5292,
+      "step": 3321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.678e-05,
+      "loss": 1.2751,
+      "step": 3322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.677e-05,
+      "loss": 1.182,
+      "step": 3323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.676e-05,
+      "loss": 1.0828,
+      "step": 3324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.675e-05,
+      "loss": 1.0361,
+      "step": 3325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.674e-05,
+      "loss": 1.1231,
+      "step": 3326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.673e-05,
+      "loss": 0.3599,
+      "step": 3327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.672e-05,
+      "loss": 0.0817,
+      "step": 3328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.671e-05,
+      "loss": 0.003,
+      "step": 3329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.670000000000001e-05,
+      "loss": 0.0395,
+      "step": 3330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.669000000000001e-05,
+      "loss": 1.0695,
+      "step": 3331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.667999999999999e-05,
+      "loss": 0.4698,
+      "step": 3332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.667e-05,
+      "loss": 0.8825,
+      "step": 3333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666e-05,
+      "loss": 1.339,
+      "step": 3334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.665000000000001e-05,
+      "loss": 1.1483,
+      "step": 3335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.664e-05,
+      "loss": 0.9696,
+      "step": 3336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.663e-05,
+      "loss": 1.1639,
+      "step": 3337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.662000000000001e-05,
+      "loss": 1.171,
+      "step": 3338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.661e-05,
+      "loss": 0.5493,
+      "step": 3339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.66e-05,
+      "loss": 0.3833,
+      "step": 3340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.659000000000001e-05,
+      "loss": 0.2755,
+      "step": 3341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.658e-05,
+      "loss": 0.333,
+      "step": 3342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.657e-05,
+      "loss": 0.2411,
+      "step": 3343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.656e-05,
+      "loss": 1.1389,
+      "step": 3344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.655e-05,
+      "loss": 0.9208,
+      "step": 3345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.654e-05,
+      "loss": 0.7557,
+      "step": 3346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.653e-05,
+      "loss": 1.1734,
+      "step": 3347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.652000000000001e-05,
+      "loss": 1.2713,
+      "step": 3348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.651000000000001e-05,
+      "loss": 0.736,
+      "step": 3349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.65e-05,
+      "loss": 1.5614,
+      "step": 3350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.649000000000001e-05,
+      "loss": 1.2169,
+      "step": 3351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.648e-05,
+      "loss": 0.3024,
+      "step": 3352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.647e-05,
+      "loss": 0.4213,
+      "step": 3353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.646e-05,
+      "loss": 0.1012,
+      "step": 3354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.645e-05,
+      "loss": 0.3134,
+      "step": 3355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.644000000000001e-05,
+      "loss": 1.0738,
+      "step": 3356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.643e-05,
+      "loss": 1.1808,
+      "step": 3357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.642e-05,
+      "loss": 1.2574,
+      "step": 3358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.641000000000001e-05,
+      "loss": 0.7578,
+      "step": 3359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.64e-05,
+      "loss": 1.2682,
+      "step": 3360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.639e-05,
+      "loss": 1.0236,
+      "step": 3361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.638e-05,
+      "loss": 1.3383,
+      "step": 3362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.637e-05,
+      "loss": 1.2041,
+      "step": 3363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.636e-05,
+      "loss": 0.5175,
+      "step": 3364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.635e-05,
+      "loss": 0.1849,
+      "step": 3365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.634e-05,
+      "loss": 0.0138,
+      "step": 3366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.633e-05,
+      "loss": 0.1851,
+      "step": 3367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.632e-05,
+      "loss": 0.2432,
+      "step": 3368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.631000000000001e-05,
+      "loss": 0.5411,
+      "step": 3369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.630000000000001e-05,
+      "loss": 1.4363,
+      "step": 3370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.629e-05,
+      "loss": 1.4847,
+      "step": 3371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.628e-05,
+      "loss": 1.2711,
+      "step": 3372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.627e-05,
+      "loss": 1.4472,
+      "step": 3373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.626e-05,
+      "loss": 1.554,
+      "step": 3374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.625e-05,
+      "loss": 1.1965,
+      "step": 3375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.624e-05,
+      "loss": 1.5876,
+      "step": 3376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.623000000000001e-05,
+      "loss": 0.5709,
+      "step": 3377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.622e-05,
+      "loss": 0.138,
+      "step": 3378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.621e-05,
+      "loss": 0.0181,
+      "step": 3379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.620000000000001e-05,
+      "loss": 0.4664,
+      "step": 3380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.619e-05,
+      "loss": 0.7688,
+      "step": 3381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.618e-05,
+      "loss": 1.0852,
+      "step": 3382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.617e-05,
+      "loss": 1.3368,
+      "step": 3383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.616e-05,
+      "loss": 0.8559,
+      "step": 3384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.615e-05,
+      "loss": 1.5204,
+      "step": 3385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.614e-05,
+      "loss": 1.3118,
+      "step": 3386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.613e-05,
+      "loss": 1.1743,
+      "step": 3387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.612000000000001e-05,
+      "loss": 0.8267,
+      "step": 3388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.611e-05,
+      "loss": 1.1495,
+      "step": 3389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.610000000000001e-05,
+      "loss": 0.2884,
+      "step": 3390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.609000000000001e-05,
+      "loss": 0.2171,
+      "step": 3391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.608e-05,
+      "loss": 0.036,
+      "step": 3392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.607e-05,
+      "loss": 0.1665,
+      "step": 3393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.606e-05,
+      "loss": 0.3443,
+      "step": 3394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.605e-05,
+      "loss": 1.6018,
+      "step": 3395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.604e-05,
+      "loss": 1.1461,
+      "step": 3396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.603e-05,
+      "loss": 0.9132,
+      "step": 3397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.602000000000001e-05,
+      "loss": 1.0039,
+      "step": 3398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.601e-05,
+      "loss": 1.2384,
+      "step": 3399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.6e-05,
+      "loss": 1.6536,
+      "step": 3400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.599000000000001e-05,
+      "loss": 1.063,
+      "step": 3401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.598e-05,
+      "loss": 0.291,
+      "step": 3402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.597e-05,
+      "loss": 0.1515,
+      "step": 3403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.596e-05,
+      "loss": 0.026,
+      "step": 3404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.595e-05,
+      "loss": 0.2531,
+      "step": 3405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.594e-05,
+      "loss": 0.8372,
+      "step": 3406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.593e-05,
+      "loss": 1.4934,
+      "step": 3407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.592e-05,
+      "loss": 1.2919,
+      "step": 3408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.591000000000001e-05,
+      "loss": 1.3796,
+      "step": 3409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.59e-05,
+      "loss": 1.1036,
+      "step": 3410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.589000000000001e-05,
+      "loss": 1.471,
+      "step": 3411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.588000000000001e-05,
+      "loss": 1.3733,
+      "step": 3412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.587e-05,
+      "loss": 0.9998,
+      "step": 3413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.586e-05,
+      "loss": 0.8716,
+      "step": 3414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.585e-05,
+      "loss": 0.153,
+      "step": 3415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.584e-05,
+      "loss": 0.2802,
+      "step": 3416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.583e-05,
+      "loss": 0.1705,
+      "step": 3417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.582e-05,
+      "loss": 0.7021,
+      "step": 3418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.581000000000001e-05,
+      "loss": 1.4884,
+      "step": 3419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.58e-05,
+      "loss": 1.1219,
+      "step": 3420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.579e-05,
+      "loss": 1.1167,
+      "step": 3421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.578000000000001e-05,
+      "loss": 1.1686,
+      "step": 3422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.577e-05,
+      "loss": 1.0598,
+      "step": 3423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.576e-05,
+      "loss": 1.1693,
+      "step": 3424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.575e-05,
+      "loss": 1.0351,
+      "step": 3425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.574e-05,
+      "loss": 1.4463,
+      "step": 3426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.573000000000001e-05,
+      "loss": 0.6237,
+      "step": 3427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.572e-05,
+      "loss": 0.3952,
+      "step": 3428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.571e-05,
+      "loss": 0.0922,
+      "step": 3429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.570000000000001e-05,
+      "loss": 0.5488,
+      "step": 3430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.569e-05,
+      "loss": 0.9456,
+      "step": 3431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.568000000000001e-05,
+      "loss": 1.3526,
+      "step": 3432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.567e-05,
+      "loss": 1.4057,
+      "step": 3433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.566e-05,
+      "loss": 1.2275,
+      "step": 3434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.565e-05,
+      "loss": 1.1486,
+      "step": 3435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.564e-05,
+      "loss": 1.2242,
+      "step": 3436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.563000000000001e-05,
+      "loss": 1.8781,
+      "step": 3437
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.562e-05,
+      "loss": 0.8304,
+      "step": 3438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.561e-05,
+      "loss": 0.3863,
+      "step": 3439
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.560000000000001e-05,
+      "loss": 0.1783,
+      "step": 3440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.559e-05,
+      "loss": 0.0336,
+      "step": 3441
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.558e-05,
+      "loss": 0.4171,
+      "step": 3442
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.557e-05,
+      "loss": 0.7934,
+      "step": 3443
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.556e-05,
+      "loss": 0.7269,
+      "step": 3444
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.555e-05,
+      "loss": 1.4343,
+      "step": 3445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.554e-05,
+      "loss": 1.4465,
+      "step": 3446
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.553e-05,
+      "loss": 1.4865,
+      "step": 3447
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.552000000000001e-05,
+      "loss": 1.0718,
+      "step": 3448
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.551e-05,
+      "loss": 1.1824,
+      "step": 3449
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.55e-05,
+      "loss": 1.6964,
+      "step": 3450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.549000000000001e-05,
+      "loss": 1.295,
+      "step": 3451
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.548e-05,
+      "loss": 0.0076,
+      "step": 3452
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.547e-05,
+      "loss": 0.0696,
+      "step": 3453
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.546e-05,
+      "loss": 0.1164,
+      "step": 3454
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.545e-05,
+      "loss": 0.3945,
+      "step": 3455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.544e-05,
+      "loss": 0.5417,
+      "step": 3456
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.543e-05,
+      "loss": 1.3044,
+      "step": 3457
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.542000000000001e-05,
+      "loss": 1.0081,
+      "step": 3458
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.541e-05,
+      "loss": 1.3976,
+      "step": 3459
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.54e-05,
+      "loss": 1.4641,
+      "step": 3460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.539000000000001e-05,
+      "loss": 1.0945,
+      "step": 3461
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.538000000000001e-05,
+      "loss": 1.1539,
+      "step": 3462
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.536999999999999e-05,
+      "loss": 1.3713,
+      "step": 3463
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.536e-05,
+      "loss": 0.1796,
+      "step": 3464
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.535e-05,
+      "loss": 0.1703,
+      "step": 3465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.534e-05,
+      "loss": 0.0233,
+      "step": 3466
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.533e-05,
+      "loss": 0.0225,
+      "step": 3467
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.532e-05,
+      "loss": 0.6819,
+      "step": 3468
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.531000000000001e-05,
+      "loss": 0.8825,
+      "step": 3469
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.53e-05,
+      "loss": 1.2163,
+      "step": 3470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.529e-05,
+      "loss": 1.0525,
+      "step": 3471
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.528000000000001e-05,
+      "loss": 1.1864,
+      "step": 3472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.527e-05,
+      "loss": 1.3074,
+      "step": 3473
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.526e-05,
+      "loss": 1.2016,
+      "step": 3474
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.525e-05,
+      "loss": 1.5222,
+      "step": 3475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.524e-05,
+      "loss": 1.4279,
+      "step": 3476
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.523e-05,
+      "loss": 0.1821,
+      "step": 3477
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.522e-05,
+      "loss": 0.0636,
+      "step": 3478
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.521000000000001e-05,
+      "loss": 0.4173,
+      "step": 3479
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.52e-05,
+      "loss": 1.0357,
+      "step": 3480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.519e-05,
+      "loss": 1.1788,
+      "step": 3481
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.518000000000001e-05,
+      "loss": 0.9179,
+      "step": 3482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.517e-05,
+      "loss": 1.4002,
+      "step": 3483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.515999999999999e-05,
+      "loss": 1.3645,
+      "step": 3484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.515e-05,
+      "loss": 1.0751,
+      "step": 3485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.514e-05,
+      "loss": 1.1431,
+      "step": 3486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.513000000000001e-05,
+      "loss": 1.1623,
+      "step": 3487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.512e-05,
+      "loss": 1.2853,
+      "step": 3488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.511e-05,
+      "loss": 1.04,
+      "step": 3489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.510000000000001e-05,
+      "loss": 0.2861,
+      "step": 3490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.509e-05,
+      "loss": 0.1158,
+      "step": 3491
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.508000000000001e-05,
+      "loss": 0.0756,
+      "step": 3492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.507e-05,
+      "loss": 1.0189,
+      "step": 3493
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.506e-05,
+      "loss": 1.2276,
+      "step": 3494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.505e-05,
+      "loss": 1.3296,
+      "step": 3495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.504e-05,
+      "loss": 1.0462,
+      "step": 3496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.503e-05,
+      "loss": 0.9952,
+      "step": 3497
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.502e-05,
+      "loss": 1.0868,
+      "step": 3498
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.501e-05,
+      "loss": 1.1822,
+      "step": 3499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 0.9473,
+      "step": 3500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.499000000000001e-05,
+      "loss": 1.3541,
+      "step": 3501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.498e-05,
+      "loss": 0.1696,
+      "step": 3502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.497000000000001e-05,
+      "loss": 0.0836,
+      "step": 3503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.496e-05,
+      "loss": 0.0053,
+      "step": 3504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.494999999999999e-05,
+      "loss": 0.4306,
+      "step": 3505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.494e-05,
+      "loss": 1.2768,
+      "step": 3506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.493e-05,
+      "loss": 1.5631,
+      "step": 3507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.492000000000001e-05,
+      "loss": 1.2194,
+      "step": 3508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.491e-05,
+      "loss": 1.4095,
+      "step": 3509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.49e-05,
+      "loss": 0.9931,
+      "step": 3510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.489000000000001e-05,
+      "loss": 1.3399,
+      "step": 3511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.488e-05,
+      "loss": 0.8445,
+      "step": 3512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.487000000000001e-05,
+      "loss": 1.2266,
+      "step": 3513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.486e-05,
+      "loss": 1.4,
+      "step": 3514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.485e-05,
+      "loss": 0.1863,
+      "step": 3515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.484e-05,
+      "loss": 0.2732,
+      "step": 3516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.483e-05,
+      "loss": 0.1995,
+      "step": 3517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.482e-05,
+      "loss": 0.3362,
+      "step": 3518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.481e-05,
+      "loss": 0.8904,
+      "step": 3519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.48e-05,
+      "loss": 0.9472,
+      "step": 3520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.479000000000001e-05,
+      "loss": 1.5247,
+      "step": 3521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.478000000000001e-05,
+      "loss": 1.2174,
+      "step": 3522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.477e-05,
+      "loss": 1.3539,
+      "step": 3523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.476e-05,
+      "loss": 1.3042,
+      "step": 3524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.475e-05,
+      "loss": 1.3719,
+      "step": 3525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.474000000000001e-05,
+      "loss": 1.4039,
+      "step": 3526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.473e-05,
+      "loss": 0.5607,
+      "step": 3527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.472e-05,
+      "loss": 0.1029,
+      "step": 3528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.471000000000001e-05,
+      "loss": 0.1468,
+      "step": 3529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.47e-05,
+      "loss": 0.7811,
+      "step": 3530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.469e-05,
+      "loss": 1.1627,
+      "step": 3531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.468000000000001e-05,
+      "loss": 1.5466,
+      "step": 3532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.467e-05,
+      "loss": 1.1335,
+      "step": 3533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.466e-05,
+      "loss": 1.3004,
+      "step": 3534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.465e-05,
+      "loss": 1.0343,
+      "step": 3535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.464e-05,
+      "loss": 1.0537,
+      "step": 3536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.463e-05,
+      "loss": 1.4037,
+      "step": 3537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.462e-05,
+      "loss": 1.2121,
+      "step": 3538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.461e-05,
+      "loss": 0.8325,
+      "step": 3539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.460000000000001e-05,
+      "loss": 0.2786,
+      "step": 3540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.459e-05,
+      "loss": 0.0983,
+      "step": 3541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.458000000000001e-05,
+      "loss": 0.5863,
+      "step": 3542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.457000000000001e-05,
+      "loss": 0.3927,
+      "step": 3543
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.455999999999999e-05,
+      "loss": 0.9454,
+      "step": 3544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.455e-05,
+      "loss": 1.3066,
+      "step": 3545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.454e-05,
+      "loss": 1.2367,
+      "step": 3546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.453000000000001e-05,
+      "loss": 1.1842,
+      "step": 3547
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.452e-05,
+      "loss": 0.8698,
+      "step": 3548
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.451e-05,
+      "loss": 1.2614,
+      "step": 3549
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.450000000000001e-05,
+      "loss": 0.9401,
+      "step": 3550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.449e-05,
+      "loss": 0.9313,
+      "step": 3551
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.448e-05,
+      "loss": 0.0182,
+      "step": 3552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.447000000000001e-05,
+      "loss": 0.1448,
+      "step": 3553
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.446e-05,
+      "loss": 0.0802,
+      "step": 3554
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.445e-05,
+      "loss": 0.7923,
+      "step": 3555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.444e-05,
+      "loss": 1.4918,
+      "step": 3556
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.443e-05,
+      "loss": 1.0994,
+      "step": 3557
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.442e-05,
+      "loss": 1.0066,
+      "step": 3558
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.441e-05,
+      "loss": 1.7885,
+      "step": 3559
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.440000000000001e-05,
+      "loss": 1.1661,
+      "step": 3560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.439000000000001e-05,
+      "loss": 1.021,
+      "step": 3561
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.438e-05,
+      "loss": 1.373,
+      "step": 3562
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.437000000000001e-05,
+      "loss": 0.8453,
+      "step": 3563
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.436e-05,
+      "loss": 1.489,
+      "step": 3564
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.435e-05,
+      "loss": 0.0841,
+      "step": 3565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.434e-05,
+      "loss": 0.1217,
+      "step": 3566
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.433e-05,
+      "loss": 0.1896,
+      "step": 3567
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.432000000000001e-05,
+      "loss": 0.6952,
+      "step": 3568
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.431e-05,
+      "loss": 0.6789,
+      "step": 3569
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.43e-05,
+      "loss": 1.3319,
+      "step": 3570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.429000000000001e-05,
+      "loss": 1.5875,
+      "step": 3571
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.428e-05,
+      "loss": 1.1925,
+      "step": 3572
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.427e-05,
+      "loss": 1.1774,
+      "step": 3573
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.426e-05,
+      "loss": 1.525,
+      "step": 3574
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.425e-05,
+      "loss": 1.2968,
+      "step": 3575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.424e-05,
+      "loss": 1.2222,
+      "step": 3576
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.423e-05,
+      "loss": 0.4704,
+      "step": 3577
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.422e-05,
+      "loss": 0.1343,
+      "step": 3578
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.421e-05,
+      "loss": 0.1837,
+      "step": 3579
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.42e-05,
+      "loss": 0.3015,
+      "step": 3580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.419000000000001e-05,
+      "loss": 0.0737,
+      "step": 3581
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.418000000000001e-05,
+      "loss": 1.8763,
+      "step": 3582
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.417e-05,
+      "loss": 1.1679,
+      "step": 3583
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.416e-05,
+      "loss": 0.8651,
+      "step": 3584
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.415e-05,
+      "loss": 1.1584,
+      "step": 3585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.414e-05,
+      "loss": 0.9639,
+      "step": 3586
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.413e-05,
+      "loss": 1.3775,
+      "step": 3587
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.412e-05,
+      "loss": 1.0509,
+      "step": 3588
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.411000000000001e-05,
+      "loss": 0.2809,
+      "step": 3589
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.41e-05,
+      "loss": 0.0856,
+      "step": 3590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.409e-05,
+      "loss": 0.5683,
+      "step": 3591
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.408000000000001e-05,
+      "loss": 0.0045,
+      "step": 3592
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.407e-05,
+      "loss": 0.8884,
+      "step": 3593
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.405999999999999e-05,
+      "loss": 0.8244,
+      "step": 3594
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.405e-05,
+      "loss": 1.0474,
+      "step": 3595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.404e-05,
+      "loss": 1.0945,
+      "step": 3596
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.403e-05,
+      "loss": 1.1497,
+      "step": 3597
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.402e-05,
+      "loss": 1.1328,
+      "step": 3598
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.401e-05,
+      "loss": 0.8845,
+      "step": 3599
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.7419,
+      "step": 3600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.399e-05,
+      "loss": 1.4309,
+      "step": 3601
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.398000000000001e-05,
+      "loss": 0.71,
+      "step": 3602
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.397000000000001e-05,
+      "loss": 0.1875,
+      "step": 3603
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.396e-05,
+      "loss": 0.0621,
+      "step": 3604
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.395e-05,
+      "loss": 0.2399,
+      "step": 3605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.394e-05,
+      "loss": 1.0146,
+      "step": 3606
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.393e-05,
+      "loss": 1.4173,
+      "step": 3607
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.392e-05,
+      "loss": 1.4262,
+      "step": 3608
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.391e-05,
+      "loss": 1.1641,
+      "step": 3609
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.390000000000001e-05,
+      "loss": 1.1796,
+      "step": 3610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.389e-05,
+      "loss": 1.0936,
+      "step": 3611
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.388e-05,
+      "loss": 1.5727,
+      "step": 3612
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.387000000000001e-05,
+      "loss": 1.1912,
+      "step": 3613
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.386e-05,
+      "loss": 0.7132,
+      "step": 3614
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.385e-05,
+      "loss": 0.1379,
+      "step": 3615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.384e-05,
+      "loss": 0.2062,
+      "step": 3616
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.383e-05,
+      "loss": 0.1975,
+      "step": 3617
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.382e-05,
+      "loss": 1.0783,
+      "step": 3618
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.381e-05,
+      "loss": 1.4427,
+      "step": 3619
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.38e-05,
+      "loss": 0.6163,
+      "step": 3620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.379000000000001e-05,
+      "loss": 1.3478,
+      "step": 3621
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.378e-05,
+      "loss": 0.995,
+      "step": 3622
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.377000000000001e-05,
+      "loss": 1.3199,
+      "step": 3623
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.376e-05,
+      "loss": 1.4731,
+      "step": 3624
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.375e-05,
+      "loss": 1.1689,
+      "step": 3625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.374e-05,
+      "loss": 1.0085,
+      "step": 3626
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.373e-05,
+      "loss": 0.375,
+      "step": 3627
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.372e-05,
+      "loss": 0.1638,
+      "step": 3628
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.371e-05,
+      "loss": 0.266,
+      "step": 3629
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.37e-05,
+      "loss": 0.2435,
+      "step": 3630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.369000000000001e-05,
+      "loss": 1.0441,
+      "step": 3631
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.368e-05,
+      "loss": 0.9916,
+      "step": 3632
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.367e-05,
+      "loss": 0.8618,
+      "step": 3633
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.366000000000001e-05,
+      "loss": 0.9792,
+      "step": 3634
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.365e-05,
+      "loss": 1.2272,
+      "step": 3635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.364e-05,
+      "loss": 1.5251,
+      "step": 3636
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.363e-05,
+      "loss": 0.9692,
+      "step": 3637
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.362e-05,
+      "loss": 1.609,
+      "step": 3638
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.361000000000001e-05,
+      "loss": 0.8376,
+      "step": 3639
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.36e-05,
+      "loss": 0.0877,
+      "step": 3640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.359e-05,
+      "loss": 0.1497,
+      "step": 3641
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.358000000000001e-05,
+      "loss": 0.0039,
+      "step": 3642
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.357e-05,
+      "loss": 0.3759,
+      "step": 3643
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.356000000000001e-05,
+      "loss": 1.2787,
+      "step": 3644
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.355e-05,
+      "loss": 1.2589,
+      "step": 3645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.354e-05,
+      "loss": 1.3431,
+      "step": 3646
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.353e-05,
+      "loss": 1.3333,
+      "step": 3647
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.352e-05,
+      "loss": 0.9036,
+      "step": 3648
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.351000000000001e-05,
+      "loss": 1.3342,
+      "step": 3649
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.35e-05,
+      "loss": 1.1478,
+      "step": 3650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.349e-05,
+      "loss": 1.3648,
+      "step": 3651
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.348000000000001e-05,
+      "loss": 0.2596,
+      "step": 3652
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.347e-05,
+      "loss": 0.1961,
+      "step": 3653
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.346e-05,
+      "loss": 0.0501,
+      "step": 3654
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.345e-05,
+      "loss": 0.242,
+      "step": 3655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.344e-05,
+      "loss": 0.8883,
+      "step": 3656
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.343e-05,
+      "loss": 1.0629,
+      "step": 3657
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.342e-05,
+      "loss": 0.965,
+      "step": 3658
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.341e-05,
+      "loss": 1.1069,
+      "step": 3659
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.340000000000001e-05,
+      "loss": 1.1576,
+      "step": 3660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.339e-05,
+      "loss": 1.1801,
+      "step": 3661
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.338e-05,
+      "loss": 1.1377,
+      "step": 3662
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.337000000000001e-05,
+      "loss": 1.6395,
+      "step": 3663
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.336e-05,
+      "loss": 1.1909,
+      "step": 3664
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.335e-05,
+      "loss": 0.3068,
+      "step": 3665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.334e-05,
+      "loss": 0.2166,
+      "step": 3666
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.333e-05,
+      "loss": 0.3322,
+      "step": 3667
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.332e-05,
+      "loss": 0.5493,
+      "step": 3668
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.331e-05,
+      "loss": 1.2172,
+      "step": 3669
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.330000000000001e-05,
+      "loss": 1.1029,
+      "step": 3670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.329e-05,
+      "loss": 1.2211,
+      "step": 3671
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.328e-05,
+      "loss": 1.5467,
+      "step": 3672
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.327000000000001e-05,
+      "loss": 1.0429,
+      "step": 3673
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.326000000000001e-05,
+      "loss": 1.0611,
+      "step": 3674
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.324999999999999e-05,
+      "loss": 1.1217,
+      "step": 3675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.324e-05,
+      "loss": 1.2027,
+      "step": 3676
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.323e-05,
+      "loss": 0.3093,
+      "step": 3677
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.322000000000001e-05,
+      "loss": 0.326,
+      "step": 3678
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.321e-05,
+      "loss": 0.3,
+      "step": 3679
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.32e-05,
+      "loss": 0.8869,
+      "step": 3680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.319000000000001e-05,
+      "loss": 0.5368,
+      "step": 3681
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.318e-05,
+      "loss": 1.1014,
+      "step": 3682
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.317e-05,
+      "loss": 0.9747,
+      "step": 3683
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.316000000000001e-05,
+      "loss": 1.1606,
+      "step": 3684
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.315e-05,
+      "loss": 1.5289,
+      "step": 3685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.314e-05,
+      "loss": 1.3049,
+      "step": 3686
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.313e-05,
+      "loss": 0.9494,
+      "step": 3687
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.312e-05,
+      "loss": 1.1164,
+      "step": 3688
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.311e-05,
+      "loss": 0.2863,
+      "step": 3689
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.31e-05,
+      "loss": 0.1831,
+      "step": 3690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.309000000000001e-05,
+      "loss": 0.1106,
+      "step": 3691
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.308e-05,
+      "loss": 0.5377,
+      "step": 3692
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.307e-05,
+      "loss": 0.9248,
+      "step": 3693
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.306000000000001e-05,
+      "loss": 1.1134,
+      "step": 3694
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.305e-05,
+      "loss": 1.1248,
+      "step": 3695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.303999999999999e-05,
+      "loss": 1.1693,
+      "step": 3696
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.303e-05,
+      "loss": 0.9331,
+      "step": 3697
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.302e-05,
+      "loss": 1.1667,
+      "step": 3698
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.301000000000001e-05,
+      "loss": 0.7,
+      "step": 3699
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.3e-05,
+      "loss": 1.4018,
+      "step": 3700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.299e-05,
+      "loss": 1.0296,
+      "step": 3701
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.298000000000001e-05,
+      "loss": 0.2061,
+      "step": 3702
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.297e-05,
+      "loss": 0.0397,
+      "step": 3703
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.296000000000001e-05,
+      "loss": 0.1979,
+      "step": 3704
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.295e-05,
+      "loss": 0.7708,
+      "step": 3705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.294e-05,
+      "loss": 0.6794,
+      "step": 3706
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.293e-05,
+      "loss": 1.3506,
+      "step": 3707
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.292e-05,
+      "loss": 1.2344,
+      "step": 3708
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.291e-05,
+      "loss": 1.3977,
+      "step": 3709
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.29e-05,
+      "loss": 1.347,
+      "step": 3710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.289e-05,
+      "loss": 1.0231,
+      "step": 3711
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.288000000000001e-05,
+      "loss": 0.8666,
+      "step": 3712
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.287000000000001e-05,
+      "loss": 0.9451,
+      "step": 3713
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.286e-05,
+      "loss": 0.368,
+      "step": 3714
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.285e-05,
+      "loss": 0.171,
+      "step": 3715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.284e-05,
+      "loss": 0.0261,
+      "step": 3716
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.283e-05,
+      "loss": 0.1338,
+      "step": 3717
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.282e-05,
+      "loss": 0.7073,
+      "step": 3718
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.281e-05,
+      "loss": 0.8429,
+      "step": 3719
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.280000000000001e-05,
+      "loss": 1.2418,
+      "step": 3720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.279e-05,
+      "loss": 0.9904,
+      "step": 3721
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.278e-05,
+      "loss": 0.94,
+      "step": 3722
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.277000000000001e-05,
+      "loss": 1.1397,
+      "step": 3723
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.276e-05,
+      "loss": 0.7728,
+      "step": 3724
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.275e-05,
+      "loss": 1.5933,
+      "step": 3725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.274e-05,
+      "loss": 0.7511,
+      "step": 3726
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.273e-05,
+      "loss": 0.3079,
+      "step": 3727
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.272e-05,
+      "loss": 0.0129,
+      "step": 3728
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.271e-05,
+      "loss": 0.2215,
+      "step": 3729
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.27e-05,
+      "loss": 0.4852,
+      "step": 3730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.269e-05,
+      "loss": 1.2252,
+      "step": 3731
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.268e-05,
+      "loss": 0.7033,
+      "step": 3732
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.267000000000001e-05,
+      "loss": 1.1546,
+      "step": 3733
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.266000000000001e-05,
+      "loss": 1.4191,
+      "step": 3734
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.264999999999999e-05,
+      "loss": 1.0767,
+      "step": 3735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.264e-05,
+      "loss": 1.0086,
+      "step": 3736
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.263e-05,
+      "loss": 1.006,
+      "step": 3737
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.262000000000001e-05,
+      "loss": 0.922,
+      "step": 3738
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.261e-05,
+      "loss": 0.666,
+      "step": 3739
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.26e-05,
+      "loss": 0.0408,
+      "step": 3740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.259000000000001e-05,
+      "loss": 0.0759,
+      "step": 3741
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.258e-05,
+      "loss": 0.0036,
+      "step": 3742
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.257e-05,
+      "loss": 0.8046,
+      "step": 3743
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.256000000000001e-05,
+      "loss": 1.2026,
+      "step": 3744
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.255e-05,
+      "loss": 0.8025,
+      "step": 3745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.254e-05,
+      "loss": 0.8494,
+      "step": 3746
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.253e-05,
+      "loss": 0.9993,
+      "step": 3747
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.252e-05,
+      "loss": 1.6772,
+      "step": 3748
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.251e-05,
+      "loss": 1.0593,
+      "step": 3749
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.25e-05,
+      "loss": 1.6379,
+      "step": 3750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.249e-05,
+      "loss": 1.2592,
+      "step": 3751
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.248000000000001e-05,
+      "loss": 0.0919,
+      "step": 3752
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.247e-05,
+      "loss": 0.1232,
+      "step": 3753
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.246000000000001e-05,
+      "loss": 0.247,
+      "step": 3754
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.245000000000001e-05,
+      "loss": 0.2271,
+      "step": 3755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.244e-05,
+      "loss": 0.9477,
+      "step": 3756
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.243e-05,
+      "loss": 0.8559,
+      "step": 3757
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.242e-05,
+      "loss": 1.1591,
+      "step": 3758
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.241000000000001e-05,
+      "loss": 1.2179,
+      "step": 3759
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.24e-05,
+      "loss": 1.1418,
+      "step": 3760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.239e-05,
+      "loss": 0.5332,
+      "step": 3761
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.238000000000001e-05,
+      "loss": 1.3215,
+      "step": 3762
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.237e-05,
+      "loss": 1.4224,
+      "step": 3763
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.236e-05,
+      "loss": 0.4693,
+      "step": 3764
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.235000000000001e-05,
+      "loss": 0.0298,
+      "step": 3765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.234e-05,
+      "loss": 0.0034,
+      "step": 3766
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.233e-05,
+      "loss": 0.3853,
+      "step": 3767
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.232e-05,
+      "loss": 0.8089,
+      "step": 3768
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.231e-05,
+      "loss": 1.004,
+      "step": 3769
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.23e-05,
+      "loss": 1.1187,
+      "step": 3770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.229e-05,
+      "loss": 1.4427,
+      "step": 3771
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.228000000000001e-05,
+      "loss": 1.5467,
+      "step": 3772
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.227000000000001e-05,
+      "loss": 1.2026,
+      "step": 3773
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.226e-05,
+      "loss": 1.6938,
+      "step": 3774
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.225000000000001e-05,
+      "loss": 1.5337,
+      "step": 3775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.224e-05,
+      "loss": 1.1823,
+      "step": 3776
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.223e-05,
+      "loss": 0.0756,
+      "step": 3777
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.222e-05,
+      "loss": 0.2167,
+      "step": 3778
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.221e-05,
+      "loss": 0.0174,
+      "step": 3779
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.220000000000001e-05,
+      "loss": 0.837,
+      "step": 3780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.219e-05,
+      "loss": 1.1461,
+      "step": 3781
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.218e-05,
+      "loss": 1.2588,
+      "step": 3782
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.217000000000001e-05,
+      "loss": 1.1002,
+      "step": 3783
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.216e-05,
+      "loss": 1.2131,
+      "step": 3784
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.215e-05,
+      "loss": 1.0638,
+      "step": 3785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.214e-05,
+      "loss": 1.1598,
+      "step": 3786
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.213e-05,
+      "loss": 1.257,
+      "step": 3787
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.212e-05,
+      "loss": 1.4639,
+      "step": 3788
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.211e-05,
+      "loss": 0.9058,
+      "step": 3789
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.21e-05,
+      "loss": 0.2757,
+      "step": 3790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.209000000000001e-05,
+      "loss": 0.1276,
+      "step": 3791
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.208e-05,
+      "loss": 0.3153,
+      "step": 3792
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.207000000000001e-05,
+      "loss": 1.1104,
+      "step": 3793
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.206000000000001e-05,
+      "loss": 1.2986,
+      "step": 3794
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.205e-05,
+      "loss": 1.257,
+      "step": 3795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.204e-05,
+      "loss": 1.3609,
+      "step": 3796
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.203e-05,
+      "loss": 1.458,
+      "step": 3797
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.202e-05,
+      "loss": 1.0679,
+      "step": 3798
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.201e-05,
+      "loss": 0.9684,
+      "step": 3799
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.2e-05,
+      "loss": 0.7611,
+      "step": 3800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.199000000000001e-05,
+      "loss": 1.4794,
+      "step": 3801
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.198e-05,
+      "loss": 0.0813,
+      "step": 3802
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.197e-05,
+      "loss": 0.1107,
+      "step": 3803
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.196000000000001e-05,
+      "loss": 0.0236,
+      "step": 3804
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.195e-05,
+      "loss": 0.1094,
+      "step": 3805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.193999999999999e-05,
+      "loss": 0.7957,
+      "step": 3806
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.193e-05,
+      "loss": 0.8468,
+      "step": 3807
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.192e-05,
+      "loss": 1.3986,
+      "step": 3808
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.191e-05,
+      "loss": 1.037,
+      "step": 3809
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.19e-05,
+      "loss": 1.2681,
+      "step": 3810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.189e-05,
+      "loss": 1.1757,
+      "step": 3811
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.188000000000001e-05,
+      "loss": 1.0132,
+      "step": 3812
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.187e-05,
+      "loss": 1.2685,
+      "step": 3813
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.186000000000001e-05,
+      "loss": 0.9404,
+      "step": 3814
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.185000000000001e-05,
+      "loss": 0.1328,
+      "step": 3815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.184e-05,
+      "loss": 0.1096,
+      "step": 3816
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.183e-05,
+      "loss": 0.1428,
+      "step": 3817
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.182e-05,
+      "loss": 0.1537,
+      "step": 3818
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.181e-05,
+      "loss": 0.8368,
+      "step": 3819
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.18e-05,
+      "loss": 1.0242,
+      "step": 3820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.179e-05,
+      "loss": 0.9575,
+      "step": 3821
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.178000000000001e-05,
+      "loss": 1.2737,
+      "step": 3822
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.177e-05,
+      "loss": 1.5079,
+      "step": 3823
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.176e-05,
+      "loss": 1.2425,
+      "step": 3824
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.175000000000001e-05,
+      "loss": 1.3658,
+      "step": 3825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.174e-05,
+      "loss": 1.3103,
+      "step": 3826
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.173e-05,
+      "loss": 0.4475,
+      "step": 3827
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.172e-05,
+      "loss": 0.2843,
+      "step": 3828
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.171e-05,
+      "loss": 0.1549,
+      "step": 3829
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.170000000000001e-05,
+      "loss": 0.0357,
+      "step": 3830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.169e-05,
+      "loss": 1.2135,
+      "step": 3831
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.168e-05,
+      "loss": 1.1821,
+      "step": 3832
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.167000000000001e-05,
+      "loss": 1.2789,
+      "step": 3833
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.166e-05,
+      "loss": 1.392,
+      "step": 3834
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.165000000000001e-05,
+      "loss": 1.2906,
+      "step": 3835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.164e-05,
+      "loss": 1.0241,
+      "step": 3836
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.163e-05,
+      "loss": 1.7103,
+      "step": 3837
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.162e-05,
+      "loss": 1.5805,
+      "step": 3838
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.161e-05,
+      "loss": 0.4916,
+      "step": 3839
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.16e-05,
+      "loss": 0.1621,
+      "step": 3840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.159e-05,
+      "loss": 0.2466,
+      "step": 3841
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.158e-05,
+      "loss": 0.0874,
+      "step": 3842
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.157000000000001e-05,
+      "loss": 0.4585,
+      "step": 3843
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.156e-05,
+      "loss": 0.9425,
+      "step": 3844
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.155e-05,
+      "loss": 0.9886,
+      "step": 3845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.154e-05,
+      "loss": 1.0045,
+      "step": 3846
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.153e-05,
+      "loss": 1.2347,
+      "step": 3847
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.152e-05,
+      "loss": 1.1037,
+      "step": 3848
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.151e-05,
+      "loss": 1.1553,
+      "step": 3849
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.15e-05,
+      "loss": 1.0109,
+      "step": 3850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.149000000000001e-05,
+      "loss": 1.1442,
+      "step": 3851
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.148e-05,
+      "loss": 0.3391,
+      "step": 3852
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.147e-05,
+      "loss": 0.2521,
+      "step": 3853
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.146000000000001e-05,
+      "loss": 0.024,
+      "step": 3854
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.145e-05,
+      "loss": 0.0016,
+      "step": 3855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.144e-05,
+      "loss": 1.0956,
+      "step": 3856
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.143e-05,
+      "loss": 1.0079,
+      "step": 3857
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.142e-05,
+      "loss": 1.348,
+      "step": 3858
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.141e-05,
+      "loss": 1.1617,
+      "step": 3859
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.14e-05,
+      "loss": 1.2804,
+      "step": 3860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.139000000000001e-05,
+      "loss": 0.7883,
+      "step": 3861
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.138e-05,
+      "loss": 1.3706,
+      "step": 3862
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.137e-05,
+      "loss": 1.1457,
+      "step": 3863
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.136000000000001e-05,
+      "loss": 0.7798,
+      "step": 3864
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.135000000000001e-05,
+      "loss": 0.0108,
+      "step": 3865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.133999999999999e-05,
+      "loss": 0.0818,
+      "step": 3866
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.133e-05,
+      "loss": 0.2382,
+      "step": 3867
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.132e-05,
+      "loss": 0.7303,
+      "step": 3868
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.131e-05,
+      "loss": 0.9475,
+      "step": 3869
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.13e-05,
+      "loss": 1.3339,
+      "step": 3870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.129e-05,
+      "loss": 1.3081,
+      "step": 3871
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.128000000000001e-05,
+      "loss": 0.8654,
+      "step": 3872
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.127e-05,
+      "loss": 1.0479,
+      "step": 3873
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.126e-05,
+      "loss": 1.1053,
+      "step": 3874
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.125000000000001e-05,
+      "loss": 1.1295,
+      "step": 3875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.124e-05,
+      "loss": 0.8385,
+      "step": 3876
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.123e-05,
+      "loss": 0.4753,
+      "step": 3877
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.122e-05,
+      "loss": 0.0106,
+      "step": 3878
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.121e-05,
+      "loss": 0.2036,
+      "step": 3879
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.12e-05,
+      "loss": 0.2675,
+      "step": 3880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.119e-05,
+      "loss": 1.0531,
+      "step": 3881
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.118000000000001e-05,
+      "loss": 1.1894,
+      "step": 3882
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.117e-05,
+      "loss": 1.6549,
+      "step": 3883
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.116e-05,
+      "loss": 0.9994,
+      "step": 3884
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.115000000000001e-05,
+      "loss": 1.118,
+      "step": 3885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.114000000000001e-05,
+      "loss": 1.3188,
+      "step": 3886
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.112999999999999e-05,
+      "loss": 0.6582,
+      "step": 3887
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.112e-05,
+      "loss": 1.3695,
+      "step": 3888
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.111e-05,
+      "loss": 0.5716,
+      "step": 3889
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.110000000000001e-05,
+      "loss": 0.3067,
+      "step": 3890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.109e-05,
+      "loss": 0.0961,
+      "step": 3891
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.108e-05,
+      "loss": 0.2363,
+      "step": 3892
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.107000000000001e-05,
+      "loss": 0.7301,
+      "step": 3893
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.106e-05,
+      "loss": 0.8832,
+      "step": 3894
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.105e-05,
+      "loss": 1.2265,
+      "step": 3895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.104000000000001e-05,
+      "loss": 1.25,
+      "step": 3896
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.103e-05,
+      "loss": 1.1533,
+      "step": 3897
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.102e-05,
+      "loss": 1.3927,
+      "step": 3898
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.101e-05,
+      "loss": 1.2584,
+      "step": 3899
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.1e-05,
+      "loss": 0.8678,
+      "step": 3900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0990000000000004e-05,
+      "loss": 0.8575,
+      "step": 3901
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.098e-05,
+      "loss": 0.2124,
+      "step": 3902
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.097000000000001e-05,
+      "loss": 0.1289,
+      "step": 3903
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0960000000000006e-05,
+      "loss": 0.0276,
+      "step": 3904
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0950000000000004e-05,
+      "loss": 0.2017,
+      "step": 3905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.094000000000001e-05,
+      "loss": 0.8374,
+      "step": 3906
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.093e-05,
+      "loss": 1.0041,
+      "step": 3907
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.092e-05,
+      "loss": 1.2468,
+      "step": 3908
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.091e-05,
+      "loss": 1.3311,
+      "step": 3909
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.09e-05,
+      "loss": 1.1137,
+      "step": 3910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.089e-05,
+      "loss": 0.6096,
+      "step": 3911
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.088000000000001e-05,
+      "loss": 0.9295,
+      "step": 3912
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0870000000000005e-05,
+      "loss": 1.781,
+      "step": 3913
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0860000000000003e-05,
+      "loss": 0.5326,
+      "step": 3914
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.085000000000001e-05,
+      "loss": 0.0166,
+      "step": 3915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.084000000000001e-05,
+      "loss": 0.1164,
+      "step": 3916
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.083e-05,
+      "loss": 0.0029,
+      "step": 3917
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.082e-05,
+      "loss": 0.4556,
+      "step": 3918
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.081e-05,
+      "loss": 1.3802,
+      "step": 3919
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.08e-05,
+      "loss": 1.0336,
+      "step": 3920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0790000000000006e-05,
+      "loss": 1.3606,
+      "step": 3921
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0780000000000004e-05,
+      "loss": 1.1066,
+      "step": 3922
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.077e-05,
+      "loss": 1.1619,
+      "step": 3923
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.076000000000001e-05,
+      "loss": 0.9714,
+      "step": 3924
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0750000000000006e-05,
+      "loss": 0.8233,
+      "step": 3925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.074000000000001e-05,
+      "loss": 1.3809,
+      "step": 3926
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0729999999999996e-05,
+      "loss": 0.7206,
+      "step": 3927
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.072e-05,
+      "loss": 0.2009,
+      "step": 3928
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.071e-05,
+      "loss": 0.115,
+      "step": 3929
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.07e-05,
+      "loss": 0.3322,
+      "step": 3930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.069e-05,
+      "loss": 0.6388,
+      "step": 3931
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.068e-05,
+      "loss": 1.2627,
+      "step": 3932
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.067000000000001e-05,
+      "loss": 1.0867,
+      "step": 3933
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0660000000000005e-05,
+      "loss": 1.3567,
+      "step": 3934
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0650000000000004e-05,
+      "loss": 1.0921,
+      "step": 3935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.064000000000001e-05,
+      "loss": 0.9487,
+      "step": 3936
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.063e-05,
+      "loss": 1.421,
+      "step": 3937
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.062e-05,
+      "loss": 1.9115,
+      "step": 3938
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.061e-05,
+      "loss": 1.1316,
+      "step": 3939
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.06e-05,
+      "loss": 0.2643,
+      "step": 3940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.059e-05,
+      "loss": 0.0699,
+      "step": 3941
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0580000000000006e-05,
+      "loss": 0.0069,
+      "step": 3942
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0570000000000005e-05,
+      "loss": 0.6185,
+      "step": 3943
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.056e-05,
+      "loss": 1.2884,
+      "step": 3944
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.055000000000001e-05,
+      "loss": 1.6046,
+      "step": 3945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0540000000000007e-05,
+      "loss": 1.5091,
+      "step": 3946
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.053e-05,
+      "loss": 0.984,
+      "step": 3947
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0519999999999997e-05,
+      "loss": 1.2452,
+      "step": 3948
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.051e-05,
+      "loss": 1.1376,
+      "step": 3949
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.05e-05,
+      "loss": 1.2762,
+      "step": 3950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0490000000000005e-05,
+      "loss": 0.7034,
+      "step": 3951
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0480000000000004e-05,
+      "loss": 0.7298,
+      "step": 3952
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.047e-05,
+      "loss": 0.2169,
+      "step": 3953
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.046000000000001e-05,
+      "loss": 0.0399,
+      "step": 3954
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0450000000000006e-05,
+      "loss": 1.2571,
+      "step": 3955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.044000000000001e-05,
+      "loss": 1.2939,
+      "step": 3956
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0429999999999996e-05,
+      "loss": 1.0578,
+      "step": 3957
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.042e-05,
+      "loss": 1.0356,
+      "step": 3958
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.041e-05,
+      "loss": 1.1808,
+      "step": 3959
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.04e-05,
+      "loss": 1.3709,
+      "step": 3960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.039e-05,
+      "loss": 1.2226,
+      "step": 3961
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.038e-05,
+      "loss": 1.0924,
+      "step": 3962
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0370000000000006e-05,
+      "loss": 1.7082,
+      "step": 3963
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0360000000000005e-05,
+      "loss": 0.3354,
+      "step": 3964
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.035e-05,
+      "loss": 0.0775,
+      "step": 3965
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.034000000000001e-05,
+      "loss": 0.3785,
+      "step": 3966
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.033e-05,
+      "loss": 0.6085,
+      "step": 3967
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.032e-05,
+      "loss": 1.0021,
+      "step": 3968
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.031e-05,
+      "loss": 1.3564,
+      "step": 3969
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.03e-05,
+      "loss": 0.8435,
+      "step": 3970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.029e-05,
+      "loss": 1.5109,
+      "step": 3971
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0280000000000006e-05,
+      "loss": 1.8771,
+      "step": 3972
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0270000000000004e-05,
+      "loss": 1.3009,
+      "step": 3973
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.026e-05,
+      "loss": 1.0811,
+      "step": 3974
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.025000000000001e-05,
+      "loss": 0.786,
+      "step": 3975
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0240000000000006e-05,
+      "loss": 1.4237,
+      "step": 3976
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.023e-05,
+      "loss": 0.111,
+      "step": 3977
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0219999999999996e-05,
+      "loss": 0.3114,
+      "step": 3978
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.021e-05,
+      "loss": 0.115,
+      "step": 3979
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.02e-05,
+      "loss": 0.6197,
+      "step": 3980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0190000000000005e-05,
+      "loss": 0.5066,
+      "step": 3981
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.018e-05,
+      "loss": 1.0946,
+      "step": 3982
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.017e-05,
+      "loss": 1.2428,
+      "step": 3983
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.016000000000001e-05,
+      "loss": 0.9774,
+      "step": 3984
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0150000000000005e-05,
+      "loss": 1.3974,
+      "step": 3985
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.014000000000001e-05,
+      "loss": 0.9295,
+      "step": 3986
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0129999999999995e-05,
+      "loss": 0.5945,
+      "step": 3987
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.012e-05,
+      "loss": 1.3168,
+      "step": 3988
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.011e-05,
+      "loss": 0.5308,
+      "step": 3989
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0100000000000004e-05,
+      "loss": 0.0958,
+      "step": 3990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.009e-05,
+      "loss": 0.1678,
+      "step": 3991
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.008e-05,
+      "loss": 0.1119,
+      "step": 3992
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0070000000000006e-05,
+      "loss": 0.9108,
+      "step": 3993
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0060000000000004e-05,
+      "loss": 0.9745,
+      "step": 3994
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.005000000000001e-05,
+      "loss": 1.3849,
+      "step": 3995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.004000000000001e-05,
+      "loss": 0.8781,
+      "step": 3996
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.003e-05,
+      "loss": 1.2002,
+      "step": 3997
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.002e-05,
+      "loss": 1.0854,
+      "step": 3998
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0009999999999996e-05,
+      "loss": 1.1983,
+      "step": 3999
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6e-05,
+      "loss": 0.8793,
+      "step": 4000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999e-05,
+      "loss": 1.2847,
+      "step": 4001
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9980000000000005e-05,
+      "loss": 0.5857,
+      "step": 4002
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9970000000000004e-05,
+      "loss": 0.1242,
+      "step": 4003
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996e-05,
+      "loss": 0.0853,
+      "step": 4004
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995000000000001e-05,
+      "loss": 0.2132,
+      "step": 4005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9940000000000005e-05,
+      "loss": 0.8098,
+      "step": 4006
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.993000000000001e-05,
+      "loss": 1.2405,
+      "step": 4007
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9919999999999996e-05,
+      "loss": 1.5634,
+      "step": 4008
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.991e-05,
+      "loss": 1.2727,
+      "step": 4009
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99e-05,
+      "loss": 1.0105,
+      "step": 4010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9890000000000004e-05,
+      "loss": 1.0629,
+      "step": 4011
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.988e-05,
+      "loss": 0.8102,
+      "step": 4012
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.987e-05,
+      "loss": 1.2647,
+      "step": 4013
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9860000000000006e-05,
+      "loss": 0.3986,
+      "step": 4014
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9850000000000005e-05,
+      "loss": 0.2435,
+      "step": 4015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.984000000000001e-05,
+      "loss": 0.118,
+      "step": 4016
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.983000000000001e-05,
+      "loss": 0.8535,
+      "step": 4017
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.982e-05,
+      "loss": 1.0061,
+      "step": 4018
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.981e-05,
+      "loss": 1.3011,
+      "step": 4019
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 1.0284,
+      "step": 4020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.979e-05,
+      "loss": 1.0179,
+      "step": 4021
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.978e-05,
+      "loss": 1.3644,
+      "step": 4022
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9770000000000005e-05,
+      "loss": 0.8888,
+      "step": 4023
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9760000000000004e-05,
+      "loss": 1.0345,
+      "step": 4024
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.975000000000001e-05,
+      "loss": 1.1955,
+      "step": 4025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.974000000000001e-05,
+      "loss": 1.0274,
+      "step": 4026
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9730000000000006e-05,
+      "loss": 0.2506,
+      "step": 4027
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.972e-05,
+      "loss": 0.0049,
+      "step": 4028
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.971e-05,
+      "loss": 0.2492,
+      "step": 4029
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.97e-05,
+      "loss": 0.4992,
+      "step": 4030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.969e-05,
+      "loss": 1.3016,
+      "step": 4031
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9680000000000005e-05,
+      "loss": 1.1039,
+      "step": 4032
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.967e-05,
+      "loss": 0.7312,
+      "step": 4033
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.966000000000001e-05,
+      "loss": 0.9687,
+      "step": 4034
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9650000000000007e-05,
+      "loss": 1.2256,
+      "step": 4035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9640000000000005e-05,
+      "loss": 0.9835,
+      "step": 4036
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.963000000000001e-05,
+      "loss": 0.8848,
+      "step": 4037
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9619999999999995e-05,
+      "loss": 1.3106,
+      "step": 4038
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.961e-05,
+      "loss": 0.5548,
+      "step": 4039
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.96e-05,
+      "loss": 0.2246,
+      "step": 4040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9590000000000004e-05,
+      "loss": 0.3017,
+      "step": 4041
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.958e-05,
+      "loss": 0.2609,
+      "step": 4042
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.957e-05,
+      "loss": 0.3171,
+      "step": 4043
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9560000000000006e-05,
+      "loss": 1.3641,
+      "step": 4044
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9550000000000004e-05,
+      "loss": 1.4966,
+      "step": 4045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.954000000000001e-05,
+      "loss": 1.163,
+      "step": 4046
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.953000000000001e-05,
+      "loss": 1.125,
+      "step": 4047
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.952e-05,
+      "loss": 0.9979,
+      "step": 4048
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.951e-05,
+      "loss": 0.7684,
+      "step": 4049
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.95e-05,
+      "loss": 1.2416,
+      "step": 4050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.949e-05,
+      "loss": 0.665,
+      "step": 4051
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.948e-05,
+      "loss": 0.047,
+      "step": 4052
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9470000000000005e-05,
+      "loss": 0.031,
+      "step": 4053
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.946e-05,
+      "loss": 0.1029,
+      "step": 4054
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.945000000000001e-05,
+      "loss": 0.1246,
+      "step": 4055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.944000000000001e-05,
+      "loss": 0.8633,
+      "step": 4056
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9430000000000005e-05,
+      "loss": 0.7577,
+      "step": 4057
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.942e-05,
+      "loss": 1.4039,
+      "step": 4058
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.941e-05,
+      "loss": 0.7579,
+      "step": 4059
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.94e-05,
+      "loss": 1.1441,
+      "step": 4060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.939e-05,
+      "loss": 1.0297,
+      "step": 4061
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9380000000000004e-05,
+      "loss": 1.0432,
+      "step": 4062
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.937e-05,
+      "loss": 1.0489,
+      "step": 4063
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.936000000000001e-05,
+      "loss": 0.6182,
+      "step": 4064
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9350000000000006e-05,
+      "loss": 0.02,
+      "step": 4065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9340000000000004e-05,
+      "loss": 0.0062,
+      "step": 4066
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.933000000000001e-05,
+      "loss": 0.1104,
+      "step": 4067
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9319999999999994e-05,
+      "loss": 0.8156,
+      "step": 4068
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.931e-05,
+      "loss": 1.1232,
+      "step": 4069
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.93e-05,
+      "loss": 1.4147,
+      "step": 4070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.929e-05,
+      "loss": 1.1906,
+      "step": 4071
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.928e-05,
+      "loss": 0.8381,
+      "step": 4072
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.927e-05,
+      "loss": 0.7068,
+      "step": 4073
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9260000000000005e-05,
+      "loss": 1.0638,
+      "step": 4074
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9250000000000004e-05,
+      "loss": 1.0283,
+      "step": 4075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.924000000000001e-05,
+      "loss": 1.0987,
+      "step": 4076
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.923000000000001e-05,
+      "loss": 0.4677,
+      "step": 4077
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.922e-05,
+      "loss": 0.0012,
+      "step": 4078
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.921e-05,
+      "loss": 0.1784,
+      "step": 4079
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.92e-05,
+      "loss": 0.13,
+      "step": 4080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.919e-05,
+      "loss": 0.266,
+      "step": 4081
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.918e-05,
+      "loss": 1.011,
+      "step": 4082
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9170000000000004e-05,
+      "loss": 1.159,
+      "step": 4083
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.916e-05,
+      "loss": 1.3838,
+      "step": 4084
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.915000000000001e-05,
+      "loss": 1.3423,
+      "step": 4085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9140000000000006e-05,
+      "loss": 1.2035,
+      "step": 4086
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9130000000000005e-05,
+      "loss": 1.3098,
+      "step": 4087
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9119999999999996e-05,
+      "loss": 1.6871,
+      "step": 4088
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.911e-05,
+      "loss": 0.9511,
+      "step": 4089
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.91e-05,
+      "loss": 0.3325,
+      "step": 4090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.909e-05,
+      "loss": 0.4698,
+      "step": 4091
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9080000000000004e-05,
+      "loss": 0.1826,
+      "step": 4092
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.907e-05,
+      "loss": 0.2797,
+      "step": 4093
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.906000000000001e-05,
+      "loss": 1.1108,
+      "step": 4094
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9050000000000006e-05,
+      "loss": 1.2174,
+      "step": 4095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9040000000000004e-05,
+      "loss": 1.1533,
+      "step": 4096
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.903000000000001e-05,
+      "loss": 1.1485,
+      "step": 4097
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.902e-05,
+      "loss": 1.0232,
+      "step": 4098
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.901e-05,
+      "loss": 0.9945,
+      "step": 4099
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9e-05,
+      "loss": 1.4329,
+      "step": 4100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.899e-05,
+      "loss": 1.3065,
+      "step": 4101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.898e-05,
+      "loss": 0.4074,
+      "step": 4102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8970000000000006e-05,
+      "loss": 0.2408,
+      "step": 4103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8960000000000005e-05,
+      "loss": 0.189,
+      "step": 4104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.895e-05,
+      "loss": 0.3017,
+      "step": 4105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.894000000000001e-05,
+      "loss": 1.0103,
+      "step": 4106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.893000000000001e-05,
+      "loss": 0.793,
+      "step": 4107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.892e-05,
+      "loss": 1.4944,
+      "step": 4108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.891e-05,
+      "loss": 1.1618,
+      "step": 4109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.89e-05,
+      "loss": 0.8751,
+      "step": 4110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.889e-05,
+      "loss": 1.3073,
+      "step": 4111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.888e-05,
+      "loss": 1.46,
+      "step": 4112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8870000000000004e-05,
+      "loss": 1.5125,
+      "step": 4113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.886e-05,
+      "loss": 1.0385,
+      "step": 4114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.885000000000001e-05,
+      "loss": 0.4426,
+      "step": 4115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8840000000000006e-05,
+      "loss": 0.0025,
+      "step": 4116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8830000000000004e-05,
+      "loss": 0.2026,
+      "step": 4117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8819999999999996e-05,
+      "loss": 0.411,
+      "step": 4118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.881e-05,
+      "loss": 1.2474,
+      "step": 4119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.88e-05,
+      "loss": 1.3536,
+      "step": 4120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.879e-05,
+      "loss": 0.9629,
+      "step": 4121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.878e-05,
+      "loss": 1.0765,
+      "step": 4122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.877e-05,
+      "loss": 1.1492,
+      "step": 4123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.876000000000001e-05,
+      "loss": 1.1196,
+      "step": 4124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8750000000000005e-05,
+      "loss": 1.0482,
+      "step": 4125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8740000000000003e-05,
+      "loss": 1.0325,
+      "step": 4126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.873000000000001e-05,
+      "loss": 0.1455,
+      "step": 4127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.872000000000001e-05,
+      "loss": 0.0283,
+      "step": 4128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.871e-05,
+      "loss": 0.4652,
+      "step": 4129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.87e-05,
+      "loss": 0.5121,
+      "step": 4130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.869e-05,
+      "loss": 1.3449,
+      "step": 4131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.868e-05,
+      "loss": 1.2444,
+      "step": 4132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8670000000000006e-05,
+      "loss": 1.2041,
+      "step": 4133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8660000000000004e-05,
+      "loss": 1.0922,
+      "step": 4134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.865e-05,
+      "loss": 1.1885,
+      "step": 4135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.864000000000001e-05,
+      "loss": 0.9894,
+      "step": 4136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8630000000000006e-05,
+      "loss": 1.2051,
+      "step": 4137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.862000000000001e-05,
+      "loss": 1.5188,
+      "step": 4138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8609999999999996e-05,
+      "loss": 0.7965,
+      "step": 4139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.86e-05,
+      "loss": 0.0979,
+      "step": 4140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.859e-05,
+      "loss": 0.0586,
+      "step": 4141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.858e-05,
+      "loss": 0.0706,
+      "step": 4142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.857e-05,
+      "loss": 0.229,
+      "step": 4143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.856e-05,
+      "loss": 0.9332,
+      "step": 4144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.855000000000001e-05,
+      "loss": 1.4694,
+      "step": 4145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8540000000000005e-05,
+      "loss": 1.3409,
+      "step": 4146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8530000000000004e-05,
+      "loss": 1.3444,
+      "step": 4147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.852000000000001e-05,
+      "loss": 0.9702,
+      "step": 4148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.851e-05,
+      "loss": 1.2079,
+      "step": 4149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.85e-05,
+      "loss": 0.8163,
+      "step": 4150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.849e-05,
+      "loss": 0.724,
+      "step": 4151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.848e-05,
+      "loss": 0.0769,
+      "step": 4152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.847e-05,
+      "loss": 0.2164,
+      "step": 4153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8460000000000006e-05,
+      "loss": 0.3677,
+      "step": 4154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8450000000000005e-05,
+      "loss": 0.7713,
+      "step": 4155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.844e-05,
+      "loss": 0.9995,
+      "step": 4156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.843000000000001e-05,
+      "loss": 1.3182,
+      "step": 4157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8420000000000006e-05,
+      "loss": 1.0238,
+      "step": 4158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.841e-05,
+      "loss": 0.9577,
+      "step": 4159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8399999999999997e-05,
+      "loss": 1.2535,
+      "step": 4160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.839e-05,
+      "loss": 1.1286,
+      "step": 4161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.838e-05,
+      "loss": 0.7372,
+      "step": 4162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8370000000000005e-05,
+      "loss": 1.8265,
+      "step": 4163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8360000000000004e-05,
+      "loss": 0.6631,
+      "step": 4164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.835e-05,
+      "loss": 0.2275,
+      "step": 4165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.834000000000001e-05,
+      "loss": 0.0029,
+      "step": 4166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8330000000000006e-05,
+      "loss": 0.0977,
+      "step": 4167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.832000000000001e-05,
+      "loss": 0.2503,
+      "step": 4168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8309999999999996e-05,
+      "loss": 0.7873,
+      "step": 4169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.83e-05,
+      "loss": 1.4768,
+      "step": 4170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.829e-05,
+      "loss": 1.2479,
+      "step": 4171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8280000000000004e-05,
+      "loss": 0.9996,
+      "step": 4172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.827e-05,
+      "loss": 0.9842,
+      "step": 4173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.826e-05,
+      "loss": 0.9094,
+      "step": 4174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8250000000000006e-05,
+      "loss": 1.142,
+      "step": 4175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8240000000000005e-05,
+      "loss": 1.3361,
+      "step": 4176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.823000000000001e-05,
+      "loss": 0.4197,
+      "step": 4177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.822000000000001e-05,
+      "loss": 0.3696,
+      "step": 4178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.821e-05,
+      "loss": 0.2174,
+      "step": 4179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.82e-05,
+      "loss": 0.2963,
+      "step": 4180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.819e-05,
+      "loss": 0.5906,
+      "step": 4181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.818e-05,
+      "loss": 1.4083,
+      "step": 4182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.817e-05,
+      "loss": 1.1404,
+      "step": 4183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8160000000000006e-05,
+      "loss": 0.9672,
+      "step": 4184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8150000000000004e-05,
+      "loss": 1.3335,
+      "step": 4185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.814e-05,
+      "loss": 1.1201,
+      "step": 4186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.813000000000001e-05,
+      "loss": 1.4262,
+      "step": 4187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8120000000000006e-05,
+      "loss": 1.187,
+      "step": 4188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.811e-05,
+      "loss": 0.5144,
+      "step": 4189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8099999999999996e-05,
+      "loss": 0.1599,
+      "step": 4190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.809e-05,
+      "loss": 0.213,
+      "step": 4191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.808e-05,
+      "loss": 0.064,
+      "step": 4192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8070000000000005e-05,
+      "loss": 0.7307,
+      "step": 4193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.806e-05,
+      "loss": 1.2113,
+      "step": 4194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.805e-05,
+      "loss": 0.8799,
+      "step": 4195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.804000000000001e-05,
+      "loss": 1.5926,
+      "step": 4196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8030000000000005e-05,
+      "loss": 0.8218,
+      "step": 4197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.802000000000001e-05,
+      "loss": 0.9744,
+      "step": 4198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8009999999999995e-05,
+      "loss": 0.8346,
+      "step": 4199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8e-05,
+      "loss": 0.6636,
+      "step": 4200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.799e-05,
+      "loss": 1.0095,
+      "step": 4201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7980000000000004e-05,
+      "loss": 0.1176,
+      "step": 4202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.797e-05,
+      "loss": 0.0066,
+      "step": 4203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.796e-05,
+      "loss": 0.029,
+      "step": 4204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7950000000000006e-05,
+      "loss": 0.0131,
+      "step": 4205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7940000000000004e-05,
+      "loss": 0.7694,
+      "step": 4206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.793000000000001e-05,
+      "loss": 1.1576,
+      "step": 4207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.792000000000001e-05,
+      "loss": 1.4184,
+      "step": 4208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.791e-05,
+      "loss": 0.8623,
+      "step": 4209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.79e-05,
+      "loss": 1.2141,
+      "step": 4210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.789e-05,
+      "loss": 1.1098,
+      "step": 4211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.788e-05,
+      "loss": 1.0423,
+      "step": 4212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.787e-05,
+      "loss": 1.3573,
+      "step": 4213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7860000000000005e-05,
+      "loss": 0.6457,
+      "step": 4214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7850000000000003e-05,
+      "loss": 0.0667,
+      "step": 4215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.784000000000001e-05,
+      "loss": 0.0819,
+      "step": 4216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.783000000000001e-05,
+      "loss": 0.368,
+      "step": 4217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7820000000000005e-05,
+      "loss": 0.9988,
+      "step": 4218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.781e-05,
+      "loss": 1.2473,
+      "step": 4219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 0.8852,
+      "step": 4220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.779e-05,
+      "loss": 1.4437,
+      "step": 4221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.778e-05,
+      "loss": 1.3349,
+      "step": 4222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7770000000000004e-05,
+      "loss": 1.5383,
+      "step": 4223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.776e-05,
+      "loss": 1.2611,
+      "step": 4224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.775e-05,
+      "loss": 0.8499,
+      "step": 4225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7740000000000006e-05,
+      "loss": 1.4862,
+      "step": 4226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7730000000000005e-05,
+      "loss": 0.1859,
+      "step": 4227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.772000000000001e-05,
+      "loss": 0.2648,
+      "step": 4228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7709999999999995e-05,
+      "loss": 0.0305,
+      "step": 4229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.77e-05,
+      "loss": 0.163,
+      "step": 4230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.769e-05,
+      "loss": 0.3477,
+      "step": 4231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7680000000000003e-05,
+      "loss": 1.3372,
+      "step": 4232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.767e-05,
+      "loss": 1.3574,
+      "step": 4233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.766e-05,
+      "loss": 0.9717,
+      "step": 4234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7650000000000005e-05,
+      "loss": 0.9409,
+      "step": 4235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7640000000000004e-05,
+      "loss": 1.159,
+      "step": 4236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.763000000000001e-05,
+      "loss": 1.0958,
+      "step": 4237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.762000000000001e-05,
+      "loss": 0.9424,
+      "step": 4238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.761e-05,
+      "loss": 1.2662,
+      "step": 4239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.76e-05,
+      "loss": 0.6171,
+      "step": 4240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.759e-05,
+      "loss": 0.1188,
+      "step": 4241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.758e-05,
+      "loss": 0.4739,
+      "step": 4242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.757e-05,
+      "loss": 0.8284,
+      "step": 4243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7560000000000005e-05,
+      "loss": 0.5589,
+      "step": 4244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.755e-05,
+      "loss": 1.4124,
+      "step": 4245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.754000000000001e-05,
+      "loss": 1.0501,
+      "step": 4246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7530000000000007e-05,
+      "loss": 1.1786,
+      "step": 4247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7520000000000005e-05,
+      "loss": 1.1951,
+      "step": 4248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7509999999999997e-05,
+      "loss": 1.1439,
+      "step": 4249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7499999999999995e-05,
+      "loss": 1.472,
+      "step": 4250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.749e-05,
+      "loss": 0.9978,
+      "step": 4251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.748e-05,
+      "loss": 0.4311,
+      "step": 4252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7470000000000004e-05,
+      "loss": 0.1329,
+      "step": 4253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.746e-05,
+      "loss": 0.022,
+      "step": 4254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.745e-05,
+      "loss": 0.1628,
+      "step": 4255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7440000000000006e-05,
+      "loss": 0.0695,
+      "step": 4256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7430000000000004e-05,
+      "loss": 1.2309,
+      "step": 4257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.742000000000001e-05,
+      "loss": 1.0178,
+      "step": 4258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.741000000000001e-05,
+      "loss": 1.2729,
+      "step": 4259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.74e-05,
+      "loss": 1.0491,
+      "step": 4260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.739e-05,
+      "loss": 1.1371,
+      "step": 4261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.738e-05,
+      "loss": 1.3564,
+      "step": 4262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.737e-05,
+      "loss": 1.0715,
+      "step": 4263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.736e-05,
+      "loss": 0.8373,
+      "step": 4264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7350000000000005e-05,
+      "loss": 0.2592,
+      "step": 4265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.734e-05,
+      "loss": 0.3993,
+      "step": 4266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.733000000000001e-05,
+      "loss": 0.2959,
+      "step": 4267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.732000000000001e-05,
+      "loss": 0.8193,
+      "step": 4268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7310000000000005e-05,
+      "loss": 1.1585,
+      "step": 4269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.73e-05,
+      "loss": 1.1719,
+      "step": 4270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.729e-05,
+      "loss": 1.1998,
+      "step": 4271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.728e-05,
+      "loss": 0.7377,
+      "step": 4272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.727e-05,
+      "loss": 1.0583,
+      "step": 4273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7260000000000004e-05,
+      "loss": 1.3092,
+      "step": 4274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.725e-05,
+      "loss": 0.9968,
+      "step": 4275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.724000000000001e-05,
+      "loss": 1.3934,
+      "step": 4276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7230000000000006e-05,
+      "loss": 0.222,
+      "step": 4277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7220000000000004e-05,
+      "loss": 0.1292,
+      "step": 4278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.721000000000001e-05,
+      "loss": 0.0089,
+      "step": 4279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.72e-05,
+      "loss": 0.2192,
+      "step": 4280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.719e-05,
+      "loss": 0.6676,
+      "step": 4281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.718e-05,
+      "loss": 1.0659,
+      "step": 4282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.717e-05,
+      "loss": 1.2036,
+      "step": 4283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.716e-05,
+      "loss": 1.252,
+      "step": 4284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.715000000000001e-05,
+      "loss": 1.1757,
+      "step": 4285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7140000000000005e-05,
+      "loss": 1.128,
+      "step": 4286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7130000000000004e-05,
+      "loss": 1.5458,
+      "step": 4287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.712000000000001e-05,
+      "loss": 1.3047,
+      "step": 4288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.711000000000001e-05,
+      "loss": 0.6545,
+      "step": 4289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.71e-05,
+      "loss": 0.0979,
+      "step": 4290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.709e-05,
+      "loss": 0.0119,
+      "step": 4291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.708e-05,
+      "loss": 0.0433,
+      "step": 4292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.707e-05,
+      "loss": 0.095,
+      "step": 4293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.706e-05,
+      "loss": 0.3483,
+      "step": 4294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7050000000000004e-05,
+      "loss": 1.081,
+      "step": 4295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.704e-05,
+      "loss": 1.2409,
+      "step": 4296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.703000000000001e-05,
+      "loss": 1.4155,
+      "step": 4297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7020000000000006e-05,
+      "loss": 1.3319,
+      "step": 4298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7010000000000005e-05,
+      "loss": 1.2246,
+      "step": 4299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 0.8664,
+      "step": 4300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.699e-05,
+      "loss": 1.3551,
+      "step": 4301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.698e-05,
+      "loss": 0.6906,
+      "step": 4302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.697e-05,
+      "loss": 0.1404,
+      "step": 4303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6960000000000004e-05,
+      "loss": 0.1309,
+      "step": 4304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.695e-05,
+      "loss": 0.2955,
+      "step": 4305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.694000000000001e-05,
+      "loss": 0.3065,
+      "step": 4306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6930000000000006e-05,
+      "loss": 0.9079,
+      "step": 4307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6920000000000004e-05,
+      "loss": 0.9474,
+      "step": 4308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.691000000000001e-05,
+      "loss": 1.2398,
+      "step": 4309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.69e-05,
+      "loss": 1.3639,
+      "step": 4310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.689e-05,
+      "loss": 1.4144,
+      "step": 4311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.688e-05,
+      "loss": 1.6582,
+      "step": 4312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.687e-05,
+      "loss": 1.0001,
+      "step": 4313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.686e-05,
+      "loss": 0.8458,
+      "step": 4314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6850000000000006e-05,
+      "loss": 0.2664,
+      "step": 4315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6840000000000005e-05,
+      "loss": 0.1492,
+      "step": 4316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.683e-05,
+      "loss": 0.4694,
+      "step": 4317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.682000000000001e-05,
+      "loss": 0.5274,
+      "step": 4318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.681000000000001e-05,
+      "loss": 1.8039,
+      "step": 4319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.68e-05,
+      "loss": 0.8598,
+      "step": 4320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.679e-05,
+      "loss": 1.3019,
+      "step": 4321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.678e-05,
+      "loss": 1.3756,
+      "step": 4322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.677e-05,
+      "loss": 1.2511,
+      "step": 4323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6760000000000005e-05,
+      "loss": 0.7606,
+      "step": 4324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6750000000000004e-05,
+      "loss": 0.6912,
+      "step": 4325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.674e-05,
+      "loss": 1.3899,
+      "step": 4326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.673000000000001e-05,
+      "loss": 0.215,
+      "step": 4327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6720000000000006e-05,
+      "loss": 0.3035,
+      "step": 4328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6710000000000004e-05,
+      "loss": 0.2567,
+      "step": 4329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6699999999999996e-05,
+      "loss": 0.1772,
+      "step": 4330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.669e-05,
+      "loss": 1.1863,
+      "step": 4331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.668e-05,
+      "loss": 1.2685,
+      "step": 4332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.667e-05,
+      "loss": 1.2599,
+      "step": 4333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.666e-05,
+      "loss": 1.4332,
+      "step": 4334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.665e-05,
+      "loss": 1.2402,
+      "step": 4335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6640000000000007e-05,
+      "loss": 1.4687,
+      "step": 4336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6630000000000005e-05,
+      "loss": 1.4354,
+      "step": 4337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6620000000000003e-05,
+      "loss": 1.5845,
+      "step": 4338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.661000000000001e-05,
+      "loss": 0.7969,
+      "step": 4339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.66e-05,
+      "loss": 0.1122,
+      "step": 4340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.659e-05,
+      "loss": 0.1978,
+      "step": 4341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.658e-05,
+      "loss": 0.0815,
+      "step": 4342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.657e-05,
+      "loss": 0.4913,
+      "step": 4343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.656e-05,
+      "loss": 1.0386,
+      "step": 4344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6550000000000006e-05,
+      "loss": 0.9332,
+      "step": 4345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6540000000000004e-05,
+      "loss": 1.3993,
+      "step": 4346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.653e-05,
+      "loss": 0.9921,
+      "step": 4347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.652000000000001e-05,
+      "loss": 0.8374,
+      "step": 4348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6510000000000006e-05,
+      "loss": 1.2739,
+      "step": 4349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.65e-05,
+      "loss": 0.7792,
+      "step": 4350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6489999999999996e-05,
+      "loss": 1.3231,
+      "step": 4351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.648e-05,
+      "loss": 0.5801,
+      "step": 4352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.647e-05,
+      "loss": 0.1381,
+      "step": 4353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6460000000000005e-05,
+      "loss": 0.1003,
+      "step": 4354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.645e-05,
+      "loss": 0.4791,
+      "step": 4355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.644e-05,
+      "loss": 1.1628,
+      "step": 4356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.643000000000001e-05,
+      "loss": 1.2177,
+      "step": 4357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6420000000000005e-05,
+      "loss": 1.3754,
+      "step": 4358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.641000000000001e-05,
+      "loss": 0.9202,
+      "step": 4359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 1.1671,
+      "step": 4360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.639e-05,
+      "loss": 1.1539,
+      "step": 4361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.638e-05,
+      "loss": 1.2948,
+      "step": 4362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.637e-05,
+      "loss": 0.8928,
+      "step": 4363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.636e-05,
+      "loss": 0.3468,
+      "step": 4364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.635e-05,
+      "loss": 0.1095,
+      "step": 4365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6340000000000006e-05,
+      "loss": 0.213,
+      "step": 4366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6330000000000004e-05,
+      "loss": 0.2273,
+      "step": 4367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.632e-05,
+      "loss": 1.0017,
+      "step": 4368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.631000000000001e-05,
+      "loss": 1.0519,
+      "step": 4369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.63e-05,
+      "loss": 1.3194,
+      "step": 4370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.629e-05,
+      "loss": 1.3233,
+      "step": 4371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6279999999999996e-05,
+      "loss": 1.2469,
+      "step": 4372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.627e-05,
+      "loss": 1.3465,
+      "step": 4373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.626e-05,
+      "loss": 1.2265,
+      "step": 4374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 1.2585,
+      "step": 4375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6240000000000004e-05,
+      "loss": 1.2695,
+      "step": 4376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.623e-05,
+      "loss": 0.1985,
+      "step": 4377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.622000000000001e-05,
+      "loss": 0.0029,
+      "step": 4378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6210000000000006e-05,
+      "loss": 0.0147,
+      "step": 4379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.620000000000001e-05,
+      "loss": 0.6874,
+      "step": 4380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6189999999999996e-05,
+      "loss": 0.848,
+      "step": 4381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.618e-05,
+      "loss": 1.1815,
+      "step": 4382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.617e-05,
+      "loss": 1.5649,
+      "step": 4383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6160000000000004e-05,
+      "loss": 0.9539,
+      "step": 4384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.615e-05,
+      "loss": 1.1659,
+      "step": 4385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.614e-05,
+      "loss": 1.0586,
+      "step": 4386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6130000000000006e-05,
+      "loss": 0.7797,
+      "step": 4387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6120000000000005e-05,
+      "loss": 1.0499,
+      "step": 4388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.611000000000001e-05,
+      "loss": 0.3436,
+      "step": 4389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.610000000000001e-05,
+      "loss": 0.1691,
+      "step": 4390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.609e-05,
+      "loss": 0.2428,
+      "step": 4391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.608e-05,
+      "loss": 0.0495,
+      "step": 4392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6070000000000004e-05,
+      "loss": 1.2734,
+      "step": 4393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.606e-05,
+      "loss": 1.2423,
+      "step": 4394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.605e-05,
+      "loss": 1.096,
+      "step": 4395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6040000000000006e-05,
+      "loss": 1.183,
+      "step": 4396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6030000000000004e-05,
+      "loss": 1.2253,
+      "step": 4397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.602000000000001e-05,
+      "loss": 0.8216,
+      "step": 4398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.601000000000001e-05,
+      "loss": 1.2454,
+      "step": 4399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.0674,
+      "step": 4400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.599e-05,
+      "loss": 1.0297,
+      "step": 4401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5979999999999996e-05,
+      "loss": 0.6479,
+      "step": 4402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.597e-05,
+      "loss": 0.0899,
+      "step": 4403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.596e-05,
+      "loss": 0.3607,
+      "step": 4404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5950000000000005e-05,
+      "loss": 0.5539,
+      "step": 4405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.594e-05,
+      "loss": 1.1256,
+      "step": 4406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.593e-05,
+      "loss": 1.0396,
+      "step": 4407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.592000000000001e-05,
+      "loss": 1.2599,
+      "step": 4408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5910000000000005e-05,
+      "loss": 1.2551,
+      "step": 4409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.590000000000001e-05,
+      "loss": 1.3375,
+      "step": 4410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5889999999999995e-05,
+      "loss": 1.5548,
+      "step": 4411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.588e-05,
+      "loss": 1.0313,
+      "step": 4412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.587e-05,
+      "loss": 1.1027,
+      "step": 4413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5860000000000004e-05,
+      "loss": 0.1417,
+      "step": 4414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.585e-05,
+      "loss": 0.1208,
+      "step": 4415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.584e-05,
+      "loss": 0.1489,
+      "step": 4416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5830000000000006e-05,
+      "loss": 0.2122,
+      "step": 4417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5820000000000004e-05,
+      "loss": 0.4975,
+      "step": 4418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.581000000000001e-05,
+      "loss": 1.1949,
+      "step": 4419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 1.2673,
+      "step": 4420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.579e-05,
+      "loss": 1.0937,
+      "step": 4421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.578e-05,
+      "loss": 1.3795,
+      "step": 4422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.577e-05,
+      "loss": 1.3032,
+      "step": 4423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.576e-05,
+      "loss": 1.2967,
+      "step": 4424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.575e-05,
+      "loss": 0.5771,
+      "step": 4425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5740000000000005e-05,
+      "loss": 0.8668,
+      "step": 4426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5730000000000003e-05,
+      "loss": 0.0867,
+      "step": 4427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.572000000000001e-05,
+      "loss": 0.0894,
+      "step": 4428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.571000000000001e-05,
+      "loss": 0.0314,
+      "step": 4429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5700000000000005e-05,
+      "loss": 0.8967,
+      "step": 4430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.569e-05,
+      "loss": 0.9903,
+      "step": 4431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5679999999999995e-05,
+      "loss": 1.3406,
+      "step": 4432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.567e-05,
+      "loss": 1.2284,
+      "step": 4433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.566e-05,
+      "loss": 1.1508,
+      "step": 4434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5650000000000004e-05,
+      "loss": 0.7044,
+      "step": 4435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.564e-05,
+      "loss": 1.5626,
+      "step": 4436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.563e-05,
+      "loss": 1.1906,
+      "step": 4437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5620000000000006e-05,
+      "loss": 1.3314,
+      "step": 4438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5610000000000005e-05,
+      "loss": 0.8006,
+      "step": 4439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.560000000000001e-05,
+      "loss": 0.082,
+      "step": 4440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5589999999999995e-05,
+      "loss": 0.0161,
+      "step": 4441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.558e-05,
+      "loss": 0.1851,
+      "step": 4442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.557e-05,
+      "loss": 0.7412,
+      "step": 4443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.556e-05,
+      "loss": 0.8748,
+      "step": 4444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.555e-05,
+      "loss": 0.9239,
+      "step": 4445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.554e-05,
+      "loss": 1.3474,
+      "step": 4446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5530000000000005e-05,
+      "loss": 0.9286,
+      "step": 4447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5520000000000004e-05,
+      "loss": 1.4323,
+      "step": 4448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.551000000000001e-05,
+      "loss": 1.3288,
+      "step": 4449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.550000000000001e-05,
+      "loss": 0.929,
+      "step": 4450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.549e-05,
+      "loss": 0.9306,
+      "step": 4451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.548e-05,
+      "loss": 0.0879,
+      "step": 4452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.547e-05,
+      "loss": 0.1708,
+      "step": 4453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.546e-05,
+      "loss": 0.0618,
+      "step": 4454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.545e-05,
+      "loss": 0.1452,
+      "step": 4455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5440000000000005e-05,
+      "loss": 0.5876,
+      "step": 4456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.543e-05,
+      "loss": 0.7892,
+      "step": 4457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.542000000000001e-05,
+      "loss": 1.1435,
+      "step": 4458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5410000000000007e-05,
+      "loss": 1.5332,
+      "step": 4459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5400000000000005e-05,
+      "loss": 0.8889,
+      "step": 4460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5389999999999997e-05,
+      "loss": 0.9442,
+      "step": 4461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.538e-05,
+      "loss": 0.8787,
+      "step": 4462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.537e-05,
+      "loss": 1.0338,
+      "step": 4463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.536e-05,
+      "loss": 0.6364,
+      "step": 4464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5350000000000004e-05,
+      "loss": 0.1215,
+      "step": 4465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.534e-05,
+      "loss": 0.0042,
+      "step": 4466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.533000000000001e-05,
+      "loss": 0.0699,
+      "step": 4467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5320000000000006e-05,
+      "loss": 0.1649,
+      "step": 4468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5310000000000004e-05,
+      "loss": 0.7461,
+      "step": 4469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.530000000000001e-05,
+      "loss": 1.0718,
+      "step": 4470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5289999999999994e-05,
+      "loss": 1.1746,
+      "step": 4471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.528e-05,
+      "loss": 0.9237,
+      "step": 4472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.527e-05,
+      "loss": 1.9542,
+      "step": 4473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.526e-05,
+      "loss": 1.2371,
+      "step": 4474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.525e-05,
+      "loss": 1.1396,
+      "step": 4475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.524e-05,
+      "loss": 1.2125,
+      "step": 4476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5230000000000005e-05,
+      "loss": 0.1872,
+      "step": 4477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.522e-05,
+      "loss": 0.2699,
+      "step": 4478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.521000000000001e-05,
+      "loss": 0.0741,
+      "step": 4479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 0.5228,
+      "step": 4480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.519e-05,
+      "loss": 0.9801,
+      "step": 4481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.518e-05,
+      "loss": 1.2376,
+      "step": 4482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.517e-05,
+      "loss": 1.0867,
+      "step": 4483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.516e-05,
+      "loss": 1.2045,
+      "step": 4484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.515e-05,
+      "loss": 1.1804,
+      "step": 4485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5140000000000004e-05,
+      "loss": 1.0163,
+      "step": 4486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.513e-05,
+      "loss": 1.9487,
+      "step": 4487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.512000000000001e-05,
+      "loss": 1.0931,
+      "step": 4488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5110000000000006e-05,
+      "loss": 1.0527,
+      "step": 4489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5100000000000004e-05,
+      "loss": 0.2114,
+      "step": 4490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5089999999999996e-05,
+      "loss": 0.2216,
+      "step": 4491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.508e-05,
+      "loss": 0.4262,
+      "step": 4492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.507e-05,
+      "loss": 0.3414,
+      "step": 4493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.506e-05,
+      "loss": 0.81,
+      "step": 4494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.505e-05,
+      "loss": 1.4315,
+      "step": 4495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.504e-05,
+      "loss": 1.1491,
+      "step": 4496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.503000000000001e-05,
+      "loss": 1.1162,
+      "step": 4497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5020000000000005e-05,
+      "loss": 1.2213,
+      "step": 4498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5010000000000004e-05,
+      "loss": 1.1529,
+      "step": 4499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 1.3533,
+      "step": 4500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.499000000000001e-05,
+      "loss": 1.5256,
+      "step": 4501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.498e-05,
+      "loss": 0.3017,
+      "step": 4502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.497e-05,
+      "loss": 0.2175,
+      "step": 4503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.496e-05,
+      "loss": 0.137,
+      "step": 4504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.495e-05,
+      "loss": 0.2238,
+      "step": 4505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4940000000000006e-05,
+      "loss": 0.7961,
+      "step": 4506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4930000000000004e-05,
+      "loss": 1.1653,
+      "step": 4507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.492e-05,
+      "loss": 1.2326,
+      "step": 4508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.491000000000001e-05,
+      "loss": 1.0216,
+      "step": 4509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4900000000000006e-05,
+      "loss": 1.433,
+      "step": 4510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4890000000000005e-05,
+      "loss": 0.8539,
+      "step": 4511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4879999999999996e-05,
+      "loss": 1.3122,
+      "step": 4512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.487e-05,
+      "loss": 0.8548,
+      "step": 4513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.486e-05,
+      "loss": 0.4416,
+      "step": 4514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.485e-05,
+      "loss": 0.084,
+      "step": 4515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4840000000000003e-05,
+      "loss": 0.2268,
+      "step": 4516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.483e-05,
+      "loss": 0.2222,
+      "step": 4517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.482000000000001e-05,
+      "loss": 1.0624,
+      "step": 4518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4810000000000005e-05,
+      "loss": 1.1193,
+      "step": 4519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4800000000000004e-05,
+      "loss": 1.4091,
+      "step": 4520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.479000000000001e-05,
+      "loss": 1.2611,
+      "step": 4521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.478e-05,
+      "loss": 1.0108,
+      "step": 4522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.477e-05,
+      "loss": 1.2684,
+      "step": 4523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.476e-05,
+      "loss": 1.1487,
+      "step": 4524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.475e-05,
+      "loss": 1.6349,
+      "step": 4525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.474e-05,
+      "loss": 1.1804,
+      "step": 4526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4730000000000006e-05,
+      "loss": 0.0803,
+      "step": 4527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4720000000000005e-05,
+      "loss": 0.2293,
+      "step": 4528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.471e-05,
+      "loss": 0.1194,
+      "step": 4529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.470000000000001e-05,
+      "loss": 0.1831,
+      "step": 4530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.469000000000001e-05,
+      "loss": 1.4535,
+      "step": 4531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.468e-05,
+      "loss": 1.1952,
+      "step": 4532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.467e-05,
+      "loss": 1.2624,
+      "step": 4533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.466e-05,
+      "loss": 1.3873,
+      "step": 4534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.465e-05,
+      "loss": 1.2666,
+      "step": 4535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4640000000000005e-05,
+      "loss": 1.1111,
+      "step": 4536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4630000000000004e-05,
+      "loss": 1.4112,
+      "step": 4537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.462e-05,
+      "loss": 0.7794,
+      "step": 4538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.461000000000001e-05,
+      "loss": 1.2795,
+      "step": 4539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4600000000000006e-05,
+      "loss": 0.2353,
+      "step": 4540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.459000000000001e-05,
+      "loss": 0.1048,
+      "step": 4541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4579999999999996e-05,
+      "loss": 0.0496,
+      "step": 4542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.457e-05,
+      "loss": 0.4388,
+      "step": 4543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.456e-05,
+      "loss": 1.1803,
+      "step": 4544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.455e-05,
+      "loss": 1.19,
+      "step": 4545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.454e-05,
+      "loss": 1.7621,
+      "step": 4546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.453e-05,
+      "loss": 1.1684,
+      "step": 4547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4520000000000007e-05,
+      "loss": 1.5187,
+      "step": 4548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4510000000000005e-05,
+      "loss": 1.1858,
+      "step": 4549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.45e-05,
+      "loss": 1.0673,
+      "step": 4550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.449000000000001e-05,
+      "loss": 1.2112,
+      "step": 4551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.448e-05,
+      "loss": 0.4071,
+      "step": 4552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.447e-05,
+      "loss": 0.0353,
+      "step": 4553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.446e-05,
+      "loss": 0.2704,
+      "step": 4554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.445e-05,
+      "loss": 0.7123,
+      "step": 4555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.444e-05,
+      "loss": 0.9059,
+      "step": 4556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4430000000000006e-05,
+      "loss": 0.9972,
+      "step": 4557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4420000000000004e-05,
+      "loss": 1.0092,
+      "step": 4558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.441e-05,
+      "loss": 0.9248,
+      "step": 4559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.440000000000001e-05,
+      "loss": 1.0897,
+      "step": 4560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4390000000000006e-05,
+      "loss": 0.7007,
+      "step": 4561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.438e-05,
+      "loss": 1.0261,
+      "step": 4562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4369999999999996e-05,
+      "loss": 1.4548,
+      "step": 4563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.436e-05,
+      "loss": 0.562,
+      "step": 4564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.435e-05,
+      "loss": 0.1692,
+      "step": 4565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4340000000000005e-05,
+      "loss": 0.1941,
+      "step": 4566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.433e-05,
+      "loss": 0.0186,
+      "step": 4567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.432e-05,
+      "loss": 0.5741,
+      "step": 4568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.431000000000001e-05,
+      "loss": 1.0978,
+      "step": 4569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4300000000000005e-05,
+      "loss": 1.0297,
+      "step": 4570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.429000000000001e-05,
+      "loss": 1.2685,
+      "step": 4571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4279999999999995e-05,
+      "loss": 1.1071,
+      "step": 4572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.427e-05,
+      "loss": 1.1081,
+      "step": 4573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.426e-05,
+      "loss": 1.4909,
+      "step": 4574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4250000000000004e-05,
+      "loss": 0.9368,
+      "step": 4575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.424e-05,
+      "loss": 1.2721,
+      "step": 4576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.423e-05,
+      "loss": 0.2267,
+      "step": 4577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4220000000000006e-05,
+      "loss": 0.0984,
+      "step": 4578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4210000000000004e-05,
+      "loss": 0.1111,
+      "step": 4579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.420000000000001e-05,
+      "loss": 0.199,
+      "step": 4580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.419000000000001e-05,
+      "loss": 0.4932,
+      "step": 4581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.418e-05,
+      "loss": 1.1482,
+      "step": 4582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.417e-05,
+      "loss": 1.0765,
+      "step": 4583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4159999999999996e-05,
+      "loss": 1.2128,
+      "step": 4584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.415e-05,
+      "loss": 1.1996,
+      "step": 4585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.414e-05,
+      "loss": 1.033,
+      "step": 4586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4130000000000005e-05,
+      "loss": 2.1093,
+      "step": 4587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4120000000000004e-05,
+      "loss": 1.7111,
+      "step": 4588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.411e-05,
+      "loss": 0.6274,
+      "step": 4589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.410000000000001e-05,
+      "loss": 0.0946,
+      "step": 4590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4090000000000006e-05,
+      "loss": 0.0714,
+      "step": 4591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.408e-05,
+      "loss": 0.4548,
+      "step": 4592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4069999999999996e-05,
+      "loss": 0.7626,
+      "step": 4593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.406e-05,
+      "loss": 1.4338,
+      "step": 4594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.405e-05,
+      "loss": 0.8456,
+      "step": 4595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4040000000000004e-05,
+      "loss": 1.0525,
+      "step": 4596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.403e-05,
+      "loss": 1.4742,
+      "step": 4597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.402e-05,
+      "loss": 1.063,
+      "step": 4598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4010000000000006e-05,
+      "loss": 1.0359,
+      "step": 4599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 0.9041,
+      "step": 4600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.399000000000001e-05,
+      "loss": 1.0751,
+      "step": 4601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3979999999999995e-05,
+      "loss": 0.1403,
+      "step": 4602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.397e-05,
+      "loss": 0.5093,
+      "step": 4603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.396e-05,
+      "loss": 0.5707,
+      "step": 4604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3950000000000004e-05,
+      "loss": 0.4657,
+      "step": 4605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.394e-05,
+      "loss": 1.1763,
+      "step": 4606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.393e-05,
+      "loss": 1.5585,
+      "step": 4607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3920000000000006e-05,
+      "loss": 0.8756,
+      "step": 4608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3910000000000004e-05,
+      "loss": 1.3331,
+      "step": 4609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.390000000000001e-05,
+      "loss": 0.879,
+      "step": 4610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.389000000000001e-05,
+      "loss": 1.074,
+      "step": 4611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.388e-05,
+      "loss": 1.7171,
+      "step": 4612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.387e-05,
+      "loss": 1.1048,
+      "step": 4613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.386e-05,
+      "loss": 0.5405,
+      "step": 4614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.385e-05,
+      "loss": 0.1639,
+      "step": 4615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.384e-05,
+      "loss": 0.1972,
+      "step": 4616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3830000000000005e-05,
+      "loss": 0.0859,
+      "step": 4617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.382e-05,
+      "loss": 0.8068,
+      "step": 4618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.381e-05,
+      "loss": 0.9658,
+      "step": 4619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 0.7887,
+      "step": 4620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3790000000000005e-05,
+      "loss": 0.8277,
+      "step": 4621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.378e-05,
+      "loss": 0.968,
+      "step": 4622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3769999999999995e-05,
+      "loss": 1.1188,
+      "step": 4623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.376e-05,
+      "loss": 1.4055,
+      "step": 4624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.375e-05,
+      "loss": 1.612,
+      "step": 4625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3740000000000004e-05,
+      "loss": 1.0851,
+      "step": 4626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.373e-05,
+      "loss": 0.0437,
+      "step": 4627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.372e-05,
+      "loss": 0.0738,
+      "step": 4628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3710000000000006e-05,
+      "loss": 0.1586,
+      "step": 4629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3700000000000004e-05,
+      "loss": 0.6203,
+      "step": 4630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.369000000000001e-05,
+      "loss": 1.4661,
+      "step": 4631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.368000000000001e-05,
+      "loss": 1.2689,
+      "step": 4632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.367e-05,
+      "loss": 1.3535,
+      "step": 4633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.366e-05,
+      "loss": 1.3205,
+      "step": 4634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.365e-05,
+      "loss": 1.3294,
+      "step": 4635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.364e-05,
+      "loss": 1.0237,
+      "step": 4636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.363e-05,
+      "loss": 1.0707,
+      "step": 4637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3620000000000005e-05,
+      "loss": 1.5703,
+      "step": 4638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3610000000000003e-05,
+      "loss": 0.4707,
+      "step": 4639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.360000000000001e-05,
+      "loss": 0.2006,
+      "step": 4640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.359000000000001e-05,
+      "loss": 0.4082,
+      "step": 4641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3580000000000005e-05,
+      "loss": 0.0502,
+      "step": 4642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.357e-05,
+      "loss": 0.397,
+      "step": 4643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.356e-05,
+      "loss": 1.0887,
+      "step": 4644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.355e-05,
+      "loss": 0.7363,
+      "step": 4645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.354e-05,
+      "loss": 1.2081,
+      "step": 4646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3530000000000004e-05,
+      "loss": 1.1342,
+      "step": 4647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.352e-05,
+      "loss": 0.8895,
+      "step": 4648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.351000000000001e-05,
+      "loss": 1.1242,
+      "step": 4649
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3500000000000006e-05,
+      "loss": 0.8047,
+      "step": 4650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3490000000000005e-05,
+      "loss": 1.3153,
+      "step": 4651
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.348000000000001e-05,
+      "loss": 0.4728,
+      "step": 4652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3469999999999995e-05,
+      "loss": 0.0427,
+      "step": 4653
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.346e-05,
+      "loss": 0.0146,
+      "step": 4654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.345e-05,
+      "loss": 0.3386,
+      "step": 4655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.344e-05,
+      "loss": 0.7286,
+      "step": 4656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.343e-05,
+      "loss": 0.7204,
+      "step": 4657
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.342e-05,
+      "loss": 0.9346,
+      "step": 4658
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3410000000000005e-05,
+      "loss": 1.2416,
+      "step": 4659
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3400000000000004e-05,
+      "loss": 1.1074,
+      "step": 4660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.339000000000001e-05,
+      "loss": 0.7304,
+      "step": 4661
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.338000000000001e-05,
+      "loss": 1.1794,
+      "step": 4662
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.337e-05,
+      "loss": 1.5685,
+      "step": 4663
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.336e-05,
+      "loss": 0.9742,
+      "step": 4664
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.335e-05,
+      "loss": 0.1456,
+      "step": 4665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.334e-05,
+      "loss": 0.2793,
+      "step": 4666
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.333e-05,
+      "loss": 0.1072,
+      "step": 4667
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3320000000000004e-05,
+      "loss": 0.4414,
+      "step": 4668
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.331e-05,
+      "loss": 1.0788,
+      "step": 4669
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.330000000000001e-05,
+      "loss": 1.2248,
+      "step": 4670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3290000000000006e-05,
+      "loss": 1.1859,
+      "step": 4671
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3280000000000005e-05,
+      "loss": 0.9413,
+      "step": 4672
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3269999999999996e-05,
+      "loss": 1.1434,
+      "step": 4673
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.326e-05,
+      "loss": 1.4349,
+      "step": 4674
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.325e-05,
+      "loss": 0.9099,
+      "step": 4675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.324e-05,
+      "loss": 1.2181,
+      "step": 4676
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3230000000000004e-05,
+      "loss": 0.2455,
+      "step": 4677
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.322e-05,
+      "loss": 0.1773,
+      "step": 4678
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.321000000000001e-05,
+      "loss": 0.3024,
+      "step": 4679
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3200000000000006e-05,
+      "loss": 0.0427,
+      "step": 4680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3190000000000004e-05,
+      "loss": 1.0202,
+      "step": 4681
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.318000000000001e-05,
+      "loss": 1.3395,
+      "step": 4682
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.317e-05,
+      "loss": 1.1686,
+      "step": 4683
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.316e-05,
+      "loss": 1.241,
+      "step": 4684
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.315e-05,
+      "loss": 1.2053,
+      "step": 4685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.314e-05,
+      "loss": 0.9355,
+      "step": 4686
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.313e-05,
+      "loss": 1.0455,
+      "step": 4687
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3120000000000006e-05,
+      "loss": 1.5722,
+      "step": 4688
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3110000000000005e-05,
+      "loss": 0.9002,
+      "step": 4689
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.31e-05,
+      "loss": 0.3744,
+      "step": 4690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.309000000000001e-05,
+      "loss": 0.2693,
+      "step": 4691
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.308000000000001e-05,
+      "loss": 0.1285,
+      "step": 4692
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.307e-05,
+      "loss": 0.2624,
+      "step": 4693
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.306e-05,
+      "loss": 0.9147,
+      "step": 4694
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.305e-05,
+      "loss": 1.0315,
+      "step": 4695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.304e-05,
+      "loss": 1.252,
+      "step": 4696
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.303e-05,
+      "loss": 1.2026,
+      "step": 4697
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3020000000000004e-05,
+      "loss": 0.9875,
+      "step": 4698
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.301e-05,
+      "loss": 0.8243,
+      "step": 4699
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 1.484,
+      "step": 4700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2990000000000006e-05,
+      "loss": 1.2465,
+      "step": 4701
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2980000000000004e-05,
+      "loss": 0.1054,
+      "step": 4702
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2969999999999996e-05,
+      "loss": 0.193,
+      "step": 4703
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.296e-05,
+      "loss": 0.0958,
+      "step": 4704
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.295e-05,
+      "loss": 0.475,
+      "step": 4705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.294e-05,
+      "loss": 1.0428,
+      "step": 4706
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.293e-05,
+      "loss": 1.4631,
+      "step": 4707
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.292e-05,
+      "loss": 1.0762,
+      "step": 4708
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.291000000000001e-05,
+      "loss": 1.2891,
+      "step": 4709
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2900000000000005e-05,
+      "loss": 1.278,
+      "step": 4710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2890000000000004e-05,
+      "loss": 1.6713,
+      "step": 4711
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.288000000000001e-05,
+      "loss": 1.1034,
+      "step": 4712
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.287e-05,
+      "loss": 1.6439,
+      "step": 4713
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.286e-05,
+      "loss": 0.708,
+      "step": 4714
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.285e-05,
+      "loss": 0.2703,
+      "step": 4715
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.284e-05,
+      "loss": 0.1287,
+      "step": 4716
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.283e-05,
+      "loss": 0.0136,
+      "step": 4717
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2820000000000006e-05,
+      "loss": 0.7527,
+      "step": 4718
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2810000000000004e-05,
+      "loss": 0.7789,
+      "step": 4719
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.28e-05,
+      "loss": 1.2139,
+      "step": 4720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.279000000000001e-05,
+      "loss": 1.6716,
+      "step": 4721
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2780000000000006e-05,
+      "loss": 1.1456,
+      "step": 4722
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.277e-05,
+      "loss": 0.8074,
+      "step": 4723
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2759999999999996e-05,
+      "loss": 1.0649,
+      "step": 4724
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.275e-05,
+      "loss": 1.4338,
+      "step": 4725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.274e-05,
+      "loss": 1.2881,
+      "step": 4726
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.273e-05,
+      "loss": 0.6193,
+      "step": 4727
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2720000000000003e-05,
+      "loss": 0.1865,
+      "step": 4728
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.271e-05,
+      "loss": 0.1035,
+      "step": 4729
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.270000000000001e-05,
+      "loss": 0.5214,
+      "step": 4730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2690000000000005e-05,
+      "loss": 0.5741,
+      "step": 4731
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2680000000000004e-05,
+      "loss": 1.0209,
+      "step": 4732
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2669999999999995e-05,
+      "loss": 1.1187,
+      "step": 4733
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.266e-05,
+      "loss": 0.9532,
+      "step": 4734
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.265e-05,
+      "loss": 1.0021,
+      "step": 4735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.264e-05,
+      "loss": 0.8301,
+      "step": 4736
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.263e-05,
+      "loss": 0.9424,
+      "step": 4737
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.262e-05,
+      "loss": 1.0457,
+      "step": 4738
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2610000000000006e-05,
+      "loss": 0.9009,
+      "step": 4739
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2600000000000005e-05,
+      "loss": 0.1132,
+      "step": 4740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.259e-05,
+      "loss": 0.472,
+      "step": 4741
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.258000000000001e-05,
+      "loss": 0.1089,
+      "step": 4742
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.257e-05,
+      "loss": 0.3879,
+      "step": 4743
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.256e-05,
+      "loss": 1.1742,
+      "step": 4744
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.255e-05,
+      "loss": 1.2531,
+      "step": 4745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.254e-05,
+      "loss": 1.0967,
+      "step": 4746
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.253e-05,
+      "loss": 1.0915,
+      "step": 4747
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2520000000000005e-05,
+      "loss": 1.1078,
+      "step": 4748
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2510000000000004e-05,
+      "loss": 1.2779,
+      "step": 4749
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.25e-05,
+      "loss": 1.2162,
+      "step": 4750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.249000000000001e-05,
+      "loss": 1.2489,
+      "step": 4751
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2480000000000006e-05,
+      "loss": 0.4834,
+      "step": 4752
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.247000000000001e-05,
+      "loss": 0.0915,
+      "step": 4753
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2459999999999996e-05,
+      "loss": 0.4018,
+      "step": 4754
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.245e-05,
+      "loss": 0.6921,
+      "step": 4755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.244e-05,
+      "loss": 0.756,
+      "step": 4756
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2430000000000005e-05,
+      "loss": 1.1045,
+      "step": 4757
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.242e-05,
+      "loss": 0.9178,
+      "step": 4758
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.241e-05,
+      "loss": 1.2015,
+      "step": 4759
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2400000000000007e-05,
+      "loss": 1.3868,
+      "step": 4760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2390000000000005e-05,
+      "loss": 0.9739,
+      "step": 4761
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.238000000000001e-05,
+      "loss": 1.1316,
+      "step": 4762
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.237000000000001e-05,
+      "loss": 1.405,
+      "step": 4763
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.236e-05,
+      "loss": 0.6092,
+      "step": 4764
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.235e-05,
+      "loss": 0.3554,
+      "step": 4765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.234e-05,
+      "loss": 0.0721,
+      "step": 4766
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.233e-05,
+      "loss": 0.2083,
+      "step": 4767
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.232e-05,
+      "loss": 0.397,
+      "step": 4768
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2310000000000006e-05,
+      "loss": 0.8134,
+      "step": 4769
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2300000000000004e-05,
+      "loss": 1.2894,
+      "step": 4770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.229e-05,
+      "loss": 1.2084,
+      "step": 4771
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.228000000000001e-05,
+      "loss": 1.3696,
+      "step": 4772
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2270000000000006e-05,
+      "loss": 1.0964,
+      "step": 4773
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.226e-05,
+      "loss": 1.1805,
+      "step": 4774
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2249999999999996e-05,
+      "loss": 0.714,
+      "step": 4775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.224e-05,
+      "loss": 1.0558,
+      "step": 4776
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.223e-05,
+      "loss": 0.2496,
+      "step": 4777
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2220000000000005e-05,
+      "loss": 0.0965,
+      "step": 4778
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.221e-05,
+      "loss": 0.068,
+      "step": 4779
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.22e-05,
+      "loss": 0.3739,
+      "step": 4780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.219000000000001e-05,
+      "loss": 0.4184,
+      "step": 4781
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2180000000000005e-05,
+      "loss": 0.3448,
+      "step": 4782
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.217000000000001e-05,
+      "loss": 1.1472,
+      "step": 4783
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2159999999999995e-05,
+      "loss": 1.3336,
+      "step": 4784
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.215e-05,
+      "loss": 0.9527,
+      "step": 4785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.214e-05,
+      "loss": 1.0608,
+      "step": 4786
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2130000000000004e-05,
+      "loss": 1.3808,
+      "step": 4787
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.212e-05,
+      "loss": 1.4152,
+      "step": 4788
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.211e-05,
+      "loss": 0.3849,
+      "step": 4789
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2100000000000006e-05,
+      "loss": 0.1858,
+      "step": 4790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2090000000000004e-05,
+      "loss": 0.4059,
+      "step": 4791
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.208000000000001e-05,
+      "loss": 0.372,
+      "step": 4792
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.207000000000001e-05,
+      "loss": 0.5909,
+      "step": 4793
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.206e-05,
+      "loss": 0.6349,
+      "step": 4794
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.205e-05,
+      "loss": 1.4524,
+      "step": 4795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.204e-05,
+      "loss": 0.9256,
+      "step": 4796
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.203e-05,
+      "loss": 1.0498,
+      "step": 4797
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.202e-05,
+      "loss": 1.051,
+      "step": 4798
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2010000000000005e-05,
+      "loss": 1.2831,
+      "step": 4799
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.199,
+      "step": 4800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.199000000000001e-05,
+      "loss": 1.2292,
+      "step": 4801
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.198000000000001e-05,
+      "loss": 0.1946,
+      "step": 4802
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1970000000000006e-05,
+      "loss": 0.129,
+      "step": 4803
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.196e-05,
+      "loss": 0.2556,
+      "step": 4804
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1949999999999996e-05,
+      "loss": 0.6237,
+      "step": 4805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.194e-05,
+      "loss": 1.139,
+      "step": 4806
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.193e-05,
+      "loss": 1.0744,
+      "step": 4807
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1920000000000004e-05,
+      "loss": 1.0348,
+      "step": 4808
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.191e-05,
+      "loss": 1.4221,
+      "step": 4809
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.19e-05,
+      "loss": 0.8476,
+      "step": 4810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1890000000000006e-05,
+      "loss": 0.9434,
+      "step": 4811
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1880000000000005e-05,
+      "loss": 0.7929,
+      "step": 4812
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.187000000000001e-05,
+      "loss": 1.3373,
+      "step": 4813
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1859999999999995e-05,
+      "loss": 0.6147,
+      "step": 4814
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.185e-05,
+      "loss": 0.0848,
+      "step": 4815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.184e-05,
+      "loss": 0.1903,
+      "step": 4816
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1830000000000004e-05,
+      "loss": 0.3614,
+      "step": 4817
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.182e-05,
+      "loss": 0.4913,
+      "step": 4818
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.181e-05,
+      "loss": 1.216,
+      "step": 4819
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 1.2748,
+      "step": 4820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1790000000000004e-05,
+      "loss": 1.1406,
+      "step": 4821
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.178000000000001e-05,
+      "loss": 0.8339,
+      "step": 4822
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.177000000000001e-05,
+      "loss": 0.9663,
+      "step": 4823
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.176e-05,
+      "loss": 1.2225,
+      "step": 4824
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.175e-05,
+      "loss": 1.5125,
+      "step": 4825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.174e-05,
+      "loss": 1.4218,
+      "step": 4826
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.173e-05,
+      "loss": 0.6951,
+      "step": 4827
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.172e-05,
+      "loss": 0.2585,
+      "step": 4828
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1710000000000005e-05,
+      "loss": 0.1457,
+      "step": 4829
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.17e-05,
+      "loss": 0.1398,
+      "step": 4830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.169000000000001e-05,
+      "loss": 0.6599,
+      "step": 4831
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.168000000000001e-05,
+      "loss": 1.0073,
+      "step": 4832
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1670000000000005e-05,
+      "loss": 1.2546,
+      "step": 4833
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.166e-05,
+      "loss": 1.596,
+      "step": 4834
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1649999999999995e-05,
+      "loss": 1.061,
+      "step": 4835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.164e-05,
+      "loss": 0.8361,
+      "step": 4836
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.163e-05,
+      "loss": 1.2961,
+      "step": 4837
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1620000000000004e-05,
+      "loss": 1.5309,
+      "step": 4838
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.161e-05,
+      "loss": 0.6208,
+      "step": 4839
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.16e-05,
+      "loss": 0.1266,
+      "step": 4840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1590000000000006e-05,
+      "loss": 0.1771,
+      "step": 4841
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1580000000000004e-05,
+      "loss": 0.1734,
+      "step": 4842
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.157000000000001e-05,
+      "loss": 0.3635,
+      "step": 4843
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1559999999999994e-05,
+      "loss": 0.9581,
+      "step": 4844
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.155e-05,
+      "loss": 0.9311,
+      "step": 4845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.154e-05,
+      "loss": 1.331,
+      "step": 4846
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.153e-05,
+      "loss": 1.2882,
+      "step": 4847
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.152e-05,
+      "loss": 1.0864,
+      "step": 4848
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.151e-05,
+      "loss": 1.3499,
+      "step": 4849
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1500000000000005e-05,
+      "loss": 1.2443,
+      "step": 4850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.149e-05,
+      "loss": 1.4724,
+      "step": 4851
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.148000000000001e-05,
+      "loss": 0.4474,
+      "step": 4852
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.147000000000001e-05,
+      "loss": 0.0935,
+      "step": 4853
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.146e-05,
+      "loss": 0.0092,
+      "step": 4854
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.145e-05,
+      "loss": 0.1824,
+      "step": 4855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.144e-05,
+      "loss": 1.1736,
+      "step": 4856
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.143e-05,
+      "loss": 0.7984,
+      "step": 4857
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.142e-05,
+      "loss": 1.1486,
+      "step": 4858
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1410000000000004e-05,
+      "loss": 1.2085,
+      "step": 4859
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.14e-05,
+      "loss": 1.5415,
+      "step": 4860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.139000000000001e-05,
+      "loss": 1.4861,
+      "step": 4861
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1380000000000006e-05,
+      "loss": 1.0136,
+      "step": 4862
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1370000000000005e-05,
+      "loss": 1.3687,
+      "step": 4863
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1359999999999996e-05,
+      "loss": 0.9792,
+      "step": 4864
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.135e-05,
+      "loss": 0.3643,
+      "step": 4865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.134e-05,
+      "loss": 0.2627,
+      "step": 4866
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.133e-05,
+      "loss": 0.1231,
+      "step": 4867
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.132e-05,
+      "loss": 0.1254,
+      "step": 4868
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.131e-05,
+      "loss": 1.4296,
+      "step": 4869
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.130000000000001e-05,
+      "loss": 1.1243,
+      "step": 4870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1290000000000005e-05,
+      "loss": 1.3342,
+      "step": 4871
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1280000000000004e-05,
+      "loss": 1.0978,
+      "step": 4872
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.127000000000001e-05,
+      "loss": 0.8462,
+      "step": 4873
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1259999999999994e-05,
+      "loss": 0.7282,
+      "step": 4874
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.125e-05,
+      "loss": 1.3527,
+      "step": 4875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.124e-05,
+      "loss": 1.028,
+      "step": 4876
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.123e-05,
+      "loss": 0.4464,
+      "step": 4877
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.122e-05,
+      "loss": 0.1641,
+      "step": 4878
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.121e-05,
+      "loss": 0.0867,
+      "step": 4879
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 0.1559,
+      "step": 4880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.119e-05,
+      "loss": 0.3545,
+      "step": 4881
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.118000000000001e-05,
+      "loss": 0.8449,
+      "step": 4882
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1170000000000006e-05,
+      "loss": 1.4368,
+      "step": 4883
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1160000000000005e-05,
+      "loss": 1.3338,
+      "step": 4884
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1149999999999996e-05,
+      "loss": 1.1761,
+      "step": 4885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.114e-05,
+      "loss": 1.0585,
+      "step": 4886
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.113e-05,
+      "loss": 0.8157,
+      "step": 4887
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.112e-05,
+      "loss": 0.9055,
+      "step": 4888
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1110000000000004e-05,
+      "loss": 0.7097,
+      "step": 4889
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.11e-05,
+      "loss": 0.2326,
+      "step": 4890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.109000000000001e-05,
+      "loss": 0.1,
+      "step": 4891
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1080000000000006e-05,
+      "loss": 0.018,
+      "step": 4892
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1070000000000004e-05,
+      "loss": 0.1528,
+      "step": 4893
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.106000000000001e-05,
+      "loss": 0.6922,
+      "step": 4894
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.105e-05,
+      "loss": 1.596,
+      "step": 4895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.104e-05,
+      "loss": 1.5371,
+      "step": 4896
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.103e-05,
+      "loss": 1.1073,
+      "step": 4897
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.102e-05,
+      "loss": 1.1359,
+      "step": 4898
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.101e-05,
+      "loss": 0.9074,
+      "step": 4899
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 1.9817,
+      "step": 4900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0990000000000005e-05,
+      "loss": 1.2221,
+      "step": 4901
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.098e-05,
+      "loss": 0.0767,
+      "step": 4902
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.097000000000001e-05,
+      "loss": 0.04,
+      "step": 4903
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.096000000000001e-05,
+      "loss": 0.163,
+      "step": 4904
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.095e-05,
+      "loss": 0.1066,
+      "step": 4905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.094e-05,
+      "loss": 0.8798,
+      "step": 4906
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.093e-05,
+      "loss": 1.4208,
+      "step": 4907
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.092e-05,
+      "loss": 1.1519,
+      "step": 4908
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.091e-05,
+      "loss": 1.3203,
+      "step": 4909
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0900000000000004e-05,
+      "loss": 0.7495,
+      "step": 4910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.089e-05,
+      "loss": 1.2199,
+      "step": 4911
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.088000000000001e-05,
+      "loss": 0.9075,
+      "step": 4912
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0870000000000006e-05,
+      "loss": 1.5021,
+      "step": 4913
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0860000000000004e-05,
+      "loss": 0.6761,
+      "step": 4914
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0849999999999996e-05,
+      "loss": 0.0825,
+      "step": 4915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.084e-05,
+      "loss": 0.1723,
+      "step": 4916
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.083e-05,
+      "loss": 0.2775,
+      "step": 4917
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.082e-05,
+      "loss": 0.5031,
+      "step": 4918
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.081e-05,
+      "loss": 1.2509,
+      "step": 4919
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.08e-05,
+      "loss": 0.9666,
+      "step": 4920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.079000000000001e-05,
+      "loss": 0.9843,
+      "step": 4921
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0780000000000005e-05,
+      "loss": 1.12,
+      "step": 4922
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0770000000000003e-05,
+      "loss": 1.3003,
+      "step": 4923
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.076000000000001e-05,
+      "loss": 1.1031,
+      "step": 4924
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.075e-05,
+      "loss": 1.3152,
+      "step": 4925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.074e-05,
+      "loss": 1.2996,
+      "step": 4926
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.073e-05,
+      "loss": 0.2358,
+      "step": 4927
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.072e-05,
+      "loss": 0.0779,
+      "step": 4928
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.071e-05,
+      "loss": 0.1792,
+      "step": 4929
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0700000000000006e-05,
+      "loss": 0.4429,
+      "step": 4930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0690000000000004e-05,
+      "loss": 0.5056,
+      "step": 4931
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.068e-05,
+      "loss": 1.0488,
+      "step": 4932
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.067000000000001e-05,
+      "loss": 1.0314,
+      "step": 4933
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0660000000000006e-05,
+      "loss": 1.2258,
+      "step": 4934
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.065e-05,
+      "loss": 1.001,
+      "step": 4935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0639999999999996e-05,
+      "loss": 1.5661,
+      "step": 4936
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.063e-05,
+      "loss": 1.0309,
+      "step": 4937
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.062e-05,
+      "loss": 1.0813,
+      "step": 4938
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0610000000000005e-05,
+      "loss": 0.8254,
+      "step": 4939
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0600000000000003e-05,
+      "loss": 0.1882,
+      "step": 4940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.059e-05,
+      "loss": 0.0027,
+      "step": 4941
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.058000000000001e-05,
+      "loss": 0.1009,
+      "step": 4942
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0570000000000005e-05,
+      "loss": 0.7077,
+      "step": 4943
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.056000000000001e-05,
+      "loss": 0.871,
+      "step": 4944
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0549999999999995e-05,
+      "loss": 1.1983,
+      "step": 4945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.054e-05,
+      "loss": 1.327,
+      "step": 4946
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.053e-05,
+      "loss": 0.9945,
+      "step": 4947
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.052e-05,
+      "loss": 1.2771,
+      "step": 4948
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.051e-05,
+      "loss": 0.9138,
+      "step": 4949
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.05e-05,
+      "loss": 1.4649,
+      "step": 4950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0490000000000006e-05,
+      "loss": 1.5474,
+      "step": 4951
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0480000000000005e-05,
+      "loss": 0.8336,
+      "step": 4952
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.047e-05,
+      "loss": 0.2497,
+      "step": 4953
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.046000000000001e-05,
+      "loss": 0.1647,
+      "step": 4954
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.045e-05,
+      "loss": 0.1042,
+      "step": 4955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.044e-05,
+      "loss": 1.2198,
+      "step": 4956
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0429999999999997e-05,
+      "loss": 1.0904,
+      "step": 4957
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.042e-05,
+      "loss": 0.9643,
+      "step": 4958
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.041e-05,
+      "loss": 0.9612,
+      "step": 4959
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0400000000000005e-05,
+      "loss": 1.0452,
+      "step": 4960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0390000000000004e-05,
+      "loss": 1.4047,
+      "step": 4961
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.038e-05,
+      "loss": 0.159,
+      "step": 4962
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.037000000000001e-05,
+      "loss": 1.8504,
+      "step": 4963
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0360000000000006e-05,
+      "loss": 0.9272,
+      "step": 4964
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.035e-05,
+      "loss": 0.143,
+      "step": 4965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0339999999999996e-05,
+      "loss": 0.0514,
+      "step": 4966
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.033e-05,
+      "loss": 0.0037,
+      "step": 4967
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.032e-05,
+      "loss": 0.4036,
+      "step": 4968
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0310000000000005e-05,
+      "loss": 1.189,
+      "step": 4969
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.03e-05,
+      "loss": 0.9268,
+      "step": 4970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.029e-05,
+      "loss": 1.1877,
+      "step": 4971
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0280000000000006e-05,
+      "loss": 0.8382,
+      "step": 4972
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0270000000000005e-05,
+      "loss": 1.2508,
+      "step": 4973
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.026000000000001e-05,
+      "loss": 0.8647,
+      "step": 4974
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0249999999999995e-05,
+      "loss": 0.847,
+      "step": 4975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.024e-05,
+      "loss": 0.929,
+      "step": 4976
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.023e-05,
+      "loss": 0.1799,
+      "step": 4977
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0220000000000004e-05,
+      "loss": 0.0579,
+      "step": 4978
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.021e-05,
+      "loss": 0.0645,
+      "step": 4979
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.02e-05,
+      "loss": 0.2197,
+      "step": 4980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0190000000000006e-05,
+      "loss": 1.0321,
+      "step": 4981
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0180000000000004e-05,
+      "loss": 1.0254,
+      "step": 4982
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.017000000000001e-05,
+      "loss": 1.3243,
+      "step": 4983
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.016000000000001e-05,
+      "loss": 1.1572,
+      "step": 4984
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.015e-05,
+      "loss": 0.9187,
+      "step": 4985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.014e-05,
+      "loss": 1.2477,
+      "step": 4986
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0129999999999996e-05,
+      "loss": 1.1285,
+      "step": 4987
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.012e-05,
+      "loss": 1.2836,
+      "step": 4988
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.011e-05,
+      "loss": 0.5915,
+      "step": 4989
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0100000000000005e-05,
+      "loss": 0.2136,
+      "step": 4990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.009e-05,
+      "loss": 0.1602,
+      "step": 4991
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.008e-05,
+      "loss": 0.6643,
+      "step": 4992
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.007000000000001e-05,
+      "loss": 0.5937,
+      "step": 4993
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0060000000000005e-05,
+      "loss": 1.1112,
+      "step": 4994
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.005e-05,
+      "loss": 1.0424,
+      "step": 4995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0039999999999995e-05,
+      "loss": 1.5094,
+      "step": 4996
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.003e-05,
+      "loss": 0.8934,
+      "step": 4997
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.002e-05,
+      "loss": 1.0421,
+      "step": 4998
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0010000000000004e-05,
+      "loss": 1.1192,
+      "step": 4999
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5e-05,
+      "loss": 1.3408,
+      "step": 5000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.999e-05,
+      "loss": 0.994,
+      "step": 5001
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9980000000000006e-05,
+      "loss": 0.0799,
+      "step": 5002
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997e-05,
+      "loss": 0.1149,
+      "step": 5003
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996e-05,
+      "loss": 0.1541,
+      "step": 5004
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995e-05,
+      "loss": 1.0334,
+      "step": 5005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9940000000000006e-05,
+      "loss": 0.4112,
+      "step": 5006
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9930000000000005e-05,
+      "loss": 1.0117,
+      "step": 5007
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992e-05,
+      "loss": 0.9565,
+      "step": 5008
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991e-05,
+      "loss": 1.2377,
+      "step": 5009
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99e-05,
+      "loss": 1.2083,
+      "step": 5010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9890000000000005e-05,
+      "loss": 1.3301,
+      "step": 5011
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9880000000000004e-05,
+      "loss": 1.0742,
+      "step": 5012
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987e-05,
+      "loss": 0.9404,
+      "step": 5013
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986e-05,
+      "loss": 0.7442,
+      "step": 5014
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9850000000000006e-05,
+      "loss": 0.0803,
+      "step": 5015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9840000000000004e-05,
+      "loss": 0.2842,
+      "step": 5016
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983e-05,
+      "loss": 0.4265,
+      "step": 5017
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982e-05,
+      "loss": 0.3014,
+      "step": 5018
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981e-05,
+      "loss": 1.0004,
+      "step": 5019
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 1.2855,
+      "step": 5020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979e-05,
+      "loss": 1.3504,
+      "step": 5021
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978e-05,
+      "loss": 1.2315,
+      "step": 5022
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977e-05,
+      "loss": 1.155,
+      "step": 5023
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.976e-05,
+      "loss": 0.916,
+      "step": 5024
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975e-05,
+      "loss": 1.242,
+      "step": 5025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974e-05,
+      "loss": 0.8666,
+      "step": 5026
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973000000000001e-05,
+      "loss": 0.3631,
+      "step": 5027
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972e-05,
+      "loss": 0.0057,
+      "step": 5028
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9710000000000003e-05,
+      "loss": 0.0109,
+      "step": 5029
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97e-05,
+      "loss": 0.5833,
+      "step": 5030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969e-05,
+      "loss": 1.4127,
+      "step": 5031
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9680000000000005e-05,
+      "loss": 1.2113,
+      "step": 5032
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967e-05,
+      "loss": 0.909,
+      "step": 5033
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966e-05,
+      "loss": 1.2507,
+      "step": 5034
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965e-05,
+      "loss": 1.5647,
+      "step": 5035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9640000000000006e-05,
+      "loss": 1.0317,
+      "step": 5036
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9630000000000004e-05,
+      "loss": 1.107,
+      "step": 5037
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.962e-05,
+      "loss": 0.9232,
+      "step": 5038
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.961e-05,
+      "loss": 1.0007,
+      "step": 5039
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96e-05,
+      "loss": 0.2055,
+      "step": 5040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9590000000000005e-05,
+      "loss": 0.2051,
+      "step": 5041
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.958e-05,
+      "loss": 0.2775,
+      "step": 5042
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.957e-05,
+      "loss": 0.4331,
+      "step": 5043
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.956e-05,
+      "loss": 0.8317,
+      "step": 5044
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9550000000000005e-05,
+      "loss": 0.8989,
+      "step": 5045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9540000000000003e-05,
+      "loss": 1.2401,
+      "step": 5046
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.953e-05,
+      "loss": 1.2457,
+      "step": 5047
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.952e-05,
+      "loss": 1.3551,
+      "step": 5048
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.951e-05,
+      "loss": 1.129,
+      "step": 5049
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 1.7558,
+      "step": 5050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.949e-05,
+      "loss": 1.3939,
+      "step": 5051
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.948000000000001e-05,
+      "loss": 0.2434,
+      "step": 5052
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.947e-05,
+      "loss": 0.0027,
+      "step": 5053
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.946e-05,
+      "loss": 0.2224,
+      "step": 5054
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.945e-05,
+      "loss": 0.1851,
+      "step": 5055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.944e-05,
+      "loss": 1.1255,
+      "step": 5056
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9430000000000006e-05,
+      "loss": 0.7815,
+      "step": 5057
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.942e-05,
+      "loss": 1.0168,
+      "step": 5058
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.941e-05,
+      "loss": 0.9712,
+      "step": 5059
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.94e-05,
+      "loss": 1.137,
+      "step": 5060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.939e-05,
+      "loss": 1.4675,
+      "step": 5061
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9380000000000005e-05,
+      "loss": 1.4289,
+      "step": 5062
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.937e-05,
+      "loss": 1.7096,
+      "step": 5063
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.936e-05,
+      "loss": 1.0544,
+      "step": 5064
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.935e-05,
+      "loss": 0.1184,
+      "step": 5065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9340000000000005e-05,
+      "loss": 0.2966,
+      "step": 5066
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9330000000000004e-05,
+      "loss": 0.0506,
+      "step": 5067
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.932e-05,
+      "loss": 0.3967,
+      "step": 5068
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.931e-05,
+      "loss": 0.7312,
+      "step": 5069
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.93e-05,
+      "loss": 1.3111,
+      "step": 5070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9290000000000004e-05,
+      "loss": 1.0185,
+      "step": 5071
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.928e-05,
+      "loss": 1.532,
+      "step": 5072
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927000000000001e-05,
+      "loss": 0.9353,
+      "step": 5073
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926e-05,
+      "loss": 1.0265,
+      "step": 5074
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9250000000000004e-05,
+      "loss": 1.5257,
+      "step": 5075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924e-05,
+      "loss": 1.345,
+      "step": 5076
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923e-05,
+      "loss": 0.1752,
+      "step": 5077
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9220000000000006e-05,
+      "loss": 0.0531,
+      "step": 5078
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921e-05,
+      "loss": 0.502,
+      "step": 5079
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92e-05,
+      "loss": 0.1396,
+      "step": 5080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919e-05,
+      "loss": 0.5108,
+      "step": 5081
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.918000000000001e-05,
+      "loss": 1.0206,
+      "step": 5082
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9170000000000005e-05,
+      "loss": 0.9268,
+      "step": 5083
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9160000000000004e-05,
+      "loss": 1.1958,
+      "step": 5084
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915e-05,
+      "loss": 1.3957,
+      "step": 5085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914e-05,
+      "loss": 1.0273,
+      "step": 5086
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9130000000000006e-05,
+      "loss": 1.2386,
+      "step": 5087
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9120000000000004e-05,
+      "loss": 0.9853,
+      "step": 5088
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.911e-05,
+      "loss": 0.7067,
+      "step": 5089
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.91e-05,
+      "loss": 0.2627,
+      "step": 5090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9090000000000006e-05,
+      "loss": 0.2452,
+      "step": 5091
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9080000000000004e-05,
+      "loss": 0.2746,
+      "step": 5092
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.907e-05,
+      "loss": 0.8475,
+      "step": 5093
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.906e-05,
+      "loss": 1.3742,
+      "step": 5094
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.905e-05,
+      "loss": 1.4148,
+      "step": 5095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9040000000000005e-05,
+      "loss": 1.4212,
+      "step": 5096
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.903e-05,
+      "loss": 0.9229,
+      "step": 5097
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.902e-05,
+      "loss": 1.302,
+      "step": 5098
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.901e-05,
+      "loss": 1.5382,
+      "step": 5099
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9e-05,
+      "loss": 1.1848,
+      "step": 5100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8990000000000004e-05,
+      "loss": 1.4986,
+      "step": 5101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.898e-05,
+      "loss": 0.1655,
+      "step": 5102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.897000000000001e-05,
+      "loss": 0.0599,
+      "step": 5103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.896e-05,
+      "loss": 0.1263,
+      "step": 5104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8950000000000004e-05,
+      "loss": 0.0707,
+      "step": 5105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.894e-05,
+      "loss": 0.9319,
+      "step": 5106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.893e-05,
+      "loss": 0.6741,
+      "step": 5107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8920000000000006e-05,
+      "loss": 1.1837,
+      "step": 5108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.891e-05,
+      "loss": 1.0505,
+      "step": 5109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.89e-05,
+      "loss": 0.8714,
+      "step": 5110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.889e-05,
+      "loss": 1.3938,
+      "step": 5111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8880000000000006e-05,
+      "loss": 1.3789,
+      "step": 5112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8870000000000005e-05,
+      "loss": 0.9414,
+      "step": 5113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.886e-05,
+      "loss": 1.263,
+      "step": 5114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.885e-05,
+      "loss": 0.3389,
+      "step": 5115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.884e-05,
+      "loss": 0.0531,
+      "step": 5116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8830000000000005e-05,
+      "loss": 0.3226,
+      "step": 5117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8820000000000004e-05,
+      "loss": 0.1753,
+      "step": 5118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.881e-05,
+      "loss": 0.2483,
+      "step": 5119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.88e-05,
+      "loss": 1.4776,
+      "step": 5120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8790000000000006e-05,
+      "loss": 1.2681,
+      "step": 5121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8780000000000004e-05,
+      "loss": 1.2942,
+      "step": 5122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.877e-05,
+      "loss": 0.9061,
+      "step": 5123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.876e-05,
+      "loss": 0.8663,
+      "step": 5124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.875e-05,
+      "loss": 1.1093,
+      "step": 5125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8740000000000004e-05,
+      "loss": 1.3073,
+      "step": 5126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.873e-05,
+      "loss": 0.449,
+      "step": 5127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.872000000000001e-05,
+      "loss": 0.1919,
+      "step": 5128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.871e-05,
+      "loss": 0.1586,
+      "step": 5129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.87e-05,
+      "loss": 0.7261,
+      "step": 5130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.869e-05,
+      "loss": 1.2038,
+      "step": 5131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.868e-05,
+      "loss": 0.9278,
+      "step": 5132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.867000000000001e-05,
+      "loss": 1.3883,
+      "step": 5133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.866e-05,
+      "loss": 1.1129,
+      "step": 5134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8650000000000003e-05,
+      "loss": 1.3896,
+      "step": 5135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.864e-05,
+      "loss": 1.2744,
+      "step": 5136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.863e-05,
+      "loss": 1.2391,
+      "step": 5137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8620000000000005e-05,
+      "loss": 1.3317,
+      "step": 5138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.861e-05,
+      "loss": 0.2724,
+      "step": 5139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.86e-05,
+      "loss": 0.2714,
+      "step": 5140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.859e-05,
+      "loss": 0.7441,
+      "step": 5141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8580000000000006e-05,
+      "loss": 0.2683,
+      "step": 5142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8570000000000004e-05,
+      "loss": 1.3454,
+      "step": 5143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.856e-05,
+      "loss": 1.1572,
+      "step": 5144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.855e-05,
+      "loss": 1.1381,
+      "step": 5145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.854e-05,
+      "loss": 1.2218,
+      "step": 5146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8530000000000005e-05,
+      "loss": 0.9484,
+      "step": 5147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.852e-05,
+      "loss": 1.421,
+      "step": 5148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.851e-05,
+      "loss": 1.4673,
+      "step": 5149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.85e-05,
+      "loss": 1.1087,
+      "step": 5150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8490000000000005e-05,
+      "loss": 1.1649,
+      "step": 5151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8480000000000003e-05,
+      "loss": 0.2006,
+      "step": 5152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.847e-05,
+      "loss": 0.1644,
+      "step": 5153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.846e-05,
+      "loss": 0.0298,
+      "step": 5154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.845e-05,
+      "loss": 0.7378,
+      "step": 5155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8440000000000004e-05,
+      "loss": 1.0901,
+      "step": 5156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.843e-05,
+      "loss": 1.1002,
+      "step": 5157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.842000000000001e-05,
+      "loss": 1.3172,
+      "step": 5158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.841e-05,
+      "loss": 1.2433,
+      "step": 5159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 1.2839,
+      "step": 5160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.839e-05,
+      "loss": 0.9295,
+      "step": 5161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.838e-05,
+      "loss": 1.4927,
+      "step": 5162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8370000000000006e-05,
+      "loss": 1.1498,
+      "step": 5163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.836e-05,
+      "loss": 1.1543,
+      "step": 5164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.835e-05,
+      "loss": 0.3848,
+      "step": 5165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.834e-05,
+      "loss": 0.1265,
+      "step": 5166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.833e-05,
+      "loss": 0.0949,
+      "step": 5167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8320000000000005e-05,
+      "loss": 0.2608,
+      "step": 5168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8309999999999997e-05,
+      "loss": 1.1343,
+      "step": 5169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.83e-05,
+      "loss": 1.4185,
+      "step": 5170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.829e-05,
+      "loss": 0.9977,
+      "step": 5171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8280000000000005e-05,
+      "loss": 1.2686,
+      "step": 5172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8270000000000004e-05,
+      "loss": 1.2234,
+      "step": 5173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.826e-05,
+      "loss": 0.8872,
+      "step": 5174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.825e-05,
+      "loss": 1.3574,
+      "step": 5175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.824e-05,
+      "loss": 1.3756,
+      "step": 5176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8230000000000004e-05,
+      "loss": 0.4277,
+      "step": 5177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.822e-05,
+      "loss": 0.0341,
+      "step": 5178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.821e-05,
+      "loss": 0.4089,
+      "step": 5179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.82e-05,
+      "loss": 0.592,
+      "step": 5180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8190000000000004e-05,
+      "loss": 1.1691,
+      "step": 5181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.818e-05,
+      "loss": 1.1604,
+      "step": 5182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.817e-05,
+      "loss": 0.9411,
+      "step": 5183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.816e-05,
+      "loss": 1.3187,
+      "step": 5184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.815e-05,
+      "loss": 1.0404,
+      "step": 5185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.814e-05,
+      "loss": 0.7663,
+      "step": 5186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.813e-05,
+      "loss": 1.307,
+      "step": 5187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.812000000000001e-05,
+      "loss": 1.858,
+      "step": 5188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8110000000000005e-05,
+      "loss": 0.18,
+      "step": 5189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8100000000000004e-05,
+      "loss": 0.095,
+      "step": 5190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.809e-05,
+      "loss": 0.0226,
+      "step": 5191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.808e-05,
+      "loss": 0.2075,
+      "step": 5192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8070000000000006e-05,
+      "loss": 1.1933,
+      "step": 5193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8060000000000004e-05,
+      "loss": 1.0614,
+      "step": 5194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.805e-05,
+      "loss": 0.9759,
+      "step": 5195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.804e-05,
+      "loss": 0.9737,
+      "step": 5196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8030000000000006e-05,
+      "loss": 1.0782,
+      "step": 5197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8020000000000004e-05,
+      "loss": 0.9587,
+      "step": 5198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.801e-05,
+      "loss": 1.2892,
+      "step": 5199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8e-05,
+      "loss": 1.0182,
+      "step": 5200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.799e-05,
+      "loss": 0.9413,
+      "step": 5201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7980000000000005e-05,
+      "loss": 0.1158,
+      "step": 5202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.797e-05,
+      "loss": 0.215,
+      "step": 5203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.796e-05,
+      "loss": 0.2579,
+      "step": 5204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.795e-05,
+      "loss": 0.374,
+      "step": 5205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.794e-05,
+      "loss": 0.997,
+      "step": 5206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7930000000000004e-05,
+      "loss": 1.5091,
+      "step": 5207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.792e-05,
+      "loss": 1.1466,
+      "step": 5208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.791000000000001e-05,
+      "loss": 0.9316,
+      "step": 5209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.79e-05,
+      "loss": 1.1593,
+      "step": 5210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7890000000000004e-05,
+      "loss": 1.1862,
+      "step": 5211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.788e-05,
+      "loss": 1.345,
+      "step": 5212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.787e-05,
+      "loss": 1.1107,
+      "step": 5213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7860000000000006e-05,
+      "loss": 0.7068,
+      "step": 5214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.785e-05,
+      "loss": 0.0846,
+      "step": 5215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.784e-05,
+      "loss": 0.3709,
+      "step": 5216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.783e-05,
+      "loss": 0.142,
+      "step": 5217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7820000000000006e-05,
+      "loss": 0.7196,
+      "step": 5218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7810000000000005e-05,
+      "loss": 1.4617,
+      "step": 5219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.78e-05,
+      "loss": 1.0763,
+      "step": 5220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.779e-05,
+      "loss": 0.8081,
+      "step": 5221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.778e-05,
+      "loss": 0.7642,
+      "step": 5222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7770000000000005e-05,
+      "loss": 0.8714,
+      "step": 5223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7760000000000004e-05,
+      "loss": 1.0166,
+      "step": 5224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.775e-05,
+      "loss": 1.181,
+      "step": 5225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.774e-05,
+      "loss": 1.134,
+      "step": 5226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7730000000000005e-05,
+      "loss": 0.2919,
+      "step": 5227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7720000000000004e-05,
+      "loss": 0.1087,
+      "step": 5228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.771e-05,
+      "loss": 0.0266,
+      "step": 5229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.77e-05,
+      "loss": 0.1847,
+      "step": 5230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.769e-05,
+      "loss": 0.636,
+      "step": 5231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7680000000000004e-05,
+      "loss": 1.2235,
+      "step": 5232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.767e-05,
+      "loss": 1.404,
+      "step": 5233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.766000000000001e-05,
+      "loss": 1.2421,
+      "step": 5234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.765e-05,
+      "loss": 0.9679,
+      "step": 5235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7640000000000005e-05,
+      "loss": 1.3817,
+      "step": 5236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.763e-05,
+      "loss": 0.9418,
+      "step": 5237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.762e-05,
+      "loss": 1.681,
+      "step": 5238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.761000000000001e-05,
+      "loss": 0.6486,
+      "step": 5239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76e-05,
+      "loss": 0.035,
+      "step": 5240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7590000000000003e-05,
+      "loss": 0.0511,
+      "step": 5241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.758e-05,
+      "loss": 0.9126,
+      "step": 5242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.757e-05,
+      "loss": 0.2491,
+      "step": 5243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7560000000000005e-05,
+      "loss": 1.1171,
+      "step": 5244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.755e-05,
+      "loss": 1.0594,
+      "step": 5245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.754e-05,
+      "loss": 1.6866,
+      "step": 5246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.753e-05,
+      "loss": 1.2107,
+      "step": 5247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7520000000000006e-05,
+      "loss": 1.0424,
+      "step": 5248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7510000000000004e-05,
+      "loss": 0.8017,
+      "step": 5249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.75e-05,
+      "loss": 0.9915,
+      "step": 5250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.749e-05,
+      "loss": 1.0775,
+      "step": 5251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.748e-05,
+      "loss": 0.2862,
+      "step": 5252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7470000000000005e-05,
+      "loss": 0.0033,
+      "step": 5253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.746e-05,
+      "loss": 0.0808,
+      "step": 5254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.745e-05,
+      "loss": 0.332,
+      "step": 5255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.744e-05,
+      "loss": 0.5204,
+      "step": 5256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7430000000000005e-05,
+      "loss": 1.2542,
+      "step": 5257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.742e-05,
+      "loss": 1.2115,
+      "step": 5258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.741e-05,
+      "loss": 1.0806,
+      "step": 5259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74e-05,
+      "loss": 1.1604,
+      "step": 5260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.739e-05,
+      "loss": 1.302,
+      "step": 5261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7380000000000004e-05,
+      "loss": 1.086,
+      "step": 5262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.737e-05,
+      "loss": 1.2698,
+      "step": 5263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.736000000000001e-05,
+      "loss": 0.1371,
+      "step": 5264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.735e-05,
+      "loss": 0.0586,
+      "step": 5265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7340000000000004e-05,
+      "loss": 0.0549,
+      "step": 5266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.733e-05,
+      "loss": 0.2943,
+      "step": 5267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.732e-05,
+      "loss": 0.9059,
+      "step": 5268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7310000000000006e-05,
+      "loss": 1.1613,
+      "step": 5269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.73e-05,
+      "loss": 0.8838,
+      "step": 5270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.729e-05,
+      "loss": 1.2833,
+      "step": 5271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.728e-05,
+      "loss": 0.9268,
+      "step": 5272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7270000000000007e-05,
+      "loss": 1.5614,
+      "step": 5273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7260000000000005e-05,
+      "loss": 0.7221,
+      "step": 5274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7249999999999997e-05,
+      "loss": 0.9321,
+      "step": 5275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.724e-05,
+      "loss": 1.1584,
+      "step": 5276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.723e-05,
+      "loss": 0.521,
+      "step": 5277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7220000000000005e-05,
+      "loss": 0.0802,
+      "step": 5278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7210000000000004e-05,
+      "loss": 0.039,
+      "step": 5279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.72e-05,
+      "loss": 0.5944,
+      "step": 5280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.719e-05,
+      "loss": 1.2013,
+      "step": 5281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.718e-05,
+      "loss": 1.3906,
+      "step": 5282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7170000000000004e-05,
+      "loss": 1.1541,
+      "step": 5283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.716e-05,
+      "loss": 1.1124,
+      "step": 5284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.715e-05,
+      "loss": 1.5288,
+      "step": 5285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.714e-05,
+      "loss": 1.4348,
+      "step": 5286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7130000000000004e-05,
+      "loss": 1.3746,
+      "step": 5287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.712e-05,
+      "loss": 1.2824,
+      "step": 5288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.711e-05,
+      "loss": 0.2904,
+      "step": 5289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.71e-05,
+      "loss": 0.1313,
+      "step": 5290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.709e-05,
+      "loss": 0.156,
+      "step": 5291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.708e-05,
+      "loss": 0.0007,
+      "step": 5292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.707e-05,
+      "loss": 0.5227,
+      "step": 5293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.706000000000001e-05,
+      "loss": 1.3004,
+      "step": 5294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.705e-05,
+      "loss": 1.65,
+      "step": 5295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7040000000000004e-05,
+      "loss": 1.1047,
+      "step": 5296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.703e-05,
+      "loss": 0.8658,
+      "step": 5297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.702e-05,
+      "loss": 1.1213,
+      "step": 5298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7010000000000006e-05,
+      "loss": 0.9512,
+      "step": 5299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7e-05,
+      "loss": 1.4775,
+      "step": 5300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.699e-05,
+      "loss": 1.3675,
+      "step": 5301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.698e-05,
+      "loss": 0.1714,
+      "step": 5302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6970000000000006e-05,
+      "loss": 0.0555,
+      "step": 5303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6960000000000004e-05,
+      "loss": 0.3008,
+      "step": 5304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.695e-05,
+      "loss": 1.1718,
+      "step": 5305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.694e-05,
+      "loss": 1.1623,
+      "step": 5306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.693e-05,
+      "loss": 1.2564,
+      "step": 5307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6920000000000005e-05,
+      "loss": 1.086,
+      "step": 5308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.691e-05,
+      "loss": 0.8592,
+      "step": 5309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.69e-05,
+      "loss": 1.547,
+      "step": 5310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.689e-05,
+      "loss": 1.4224,
+      "step": 5311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.688e-05,
+      "loss": 0.4253,
+      "step": 5312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6870000000000004e-05,
+      "loss": 0.9677,
+      "step": 5313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.686e-05,
+      "loss": 0.7753,
+      "step": 5314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.685000000000001e-05,
+      "loss": 0.145,
+      "step": 5315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.684e-05,
+      "loss": 0.1642,
+      "step": 5316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6830000000000004e-05,
+      "loss": 0.4301,
+      "step": 5317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.682e-05,
+      "loss": 0.797,
+      "step": 5318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.681e-05,
+      "loss": 1.2373,
+      "step": 5319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 1.1598,
+      "step": 5320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.679e-05,
+      "loss": 1.3574,
+      "step": 5321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.678e-05,
+      "loss": 1.2468,
+      "step": 5322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.677e-05,
+      "loss": 0.9431,
+      "step": 5323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6760000000000006e-05,
+      "loss": 0.8772,
+      "step": 5324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6750000000000005e-05,
+      "loss": 1.1057,
+      "step": 5325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.674e-05,
+      "loss": 1.1939,
+      "step": 5326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.673e-05,
+      "loss": 0.7336,
+      "step": 5327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.672e-05,
+      "loss": 0.0943,
+      "step": 5328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6710000000000005e-05,
+      "loss": 0.1596,
+      "step": 5329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6700000000000003e-05,
+      "loss": 0.0245,
+      "step": 5330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.669e-05,
+      "loss": 0.7655,
+      "step": 5331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.668e-05,
+      "loss": 0.6516,
+      "step": 5332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6670000000000005e-05,
+      "loss": 1.1821,
+      "step": 5333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6660000000000004e-05,
+      "loss": 1.0415,
+      "step": 5334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.665e-05,
+      "loss": 1.1262,
+      "step": 5335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.664e-05,
+      "loss": 0.9718,
+      "step": 5336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.663e-05,
+      "loss": 1.1799,
+      "step": 5337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6620000000000004e-05,
+      "loss": 1.4894,
+      "step": 5338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.661e-05,
+      "loss": 0.7662,
+      "step": 5339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 0.6275,
+      "step": 5340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.659e-05,
+      "loss": 0.2112,
+      "step": 5341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6580000000000005e-05,
+      "loss": 0.5108,
+      "step": 5342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.657e-05,
+      "loss": 1.0344,
+      "step": 5343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.656e-05,
+      "loss": 1.4305,
+      "step": 5344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.655000000000001e-05,
+      "loss": 0.7711,
+      "step": 5345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.654e-05,
+      "loss": 1.3726,
+      "step": 5346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6530000000000003e-05,
+      "loss": 1.2449,
+      "step": 5347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.652e-05,
+      "loss": 1.2257,
+      "step": 5348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.651e-05,
+      "loss": 1.0802,
+      "step": 5349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 1.2183,
+      "step": 5350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.649e-05,
+      "loss": 1.0666,
+      "step": 5351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.648e-05,
+      "loss": 0.3286,
+      "step": 5352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.647e-05,
+      "loss": 0.2917,
+      "step": 5353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6460000000000006e-05,
+      "loss": 0.4343,
+      "step": 5354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6450000000000004e-05,
+      "loss": 0.0287,
+      "step": 5355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.644e-05,
+      "loss": 1.2737,
+      "step": 5356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.643e-05,
+      "loss": 1.1861,
+      "step": 5357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.642e-05,
+      "loss": 1.4535,
+      "step": 5358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6410000000000005e-05,
+      "loss": 1.2016,
+      "step": 5359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.64e-05,
+      "loss": 1.0965,
+      "step": 5360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.639e-05,
+      "loss": 1.2501,
+      "step": 5361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.638e-05,
+      "loss": 0.8766,
+      "step": 5362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6370000000000005e-05,
+      "loss": 1.0993,
+      "step": 5363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.636e-05,
+      "loss": 0.4152,
+      "step": 5364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.635e-05,
+      "loss": 0.179,
+      "step": 5365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.634e-05,
+      "loss": 0.386,
+      "step": 5366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.633e-05,
+      "loss": 0.0034,
+      "step": 5367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6320000000000004e-05,
+      "loss": 0.7866,
+      "step": 5368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.631e-05,
+      "loss": 1.436,
+      "step": 5369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.630000000000001e-05,
+      "loss": 1.0678,
+      "step": 5370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.629e-05,
+      "loss": 1.169,
+      "step": 5371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6280000000000004e-05,
+      "loss": 1.1469,
+      "step": 5372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.627e-05,
+      "loss": 1.2942,
+      "step": 5373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.626e-05,
+      "loss": 1.1234,
+      "step": 5374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6250000000000006e-05,
+      "loss": 1.4663,
+      "step": 5375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.624e-05,
+      "loss": 1.332,
+      "step": 5376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.623e-05,
+      "loss": 0.437,
+      "step": 5377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.622e-05,
+      "loss": 0.0993,
+      "step": 5378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6210000000000006e-05,
+      "loss": 0.1839,
+      "step": 5379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 0.246,
+      "step": 5380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.619e-05,
+      "loss": 1.4356,
+      "step": 5381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.618e-05,
+      "loss": 0.6228,
+      "step": 5382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.617e-05,
+      "loss": 1.1851,
+      "step": 5383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6160000000000005e-05,
+      "loss": 1.1277,
+      "step": 5384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6150000000000004e-05,
+      "loss": 1.232,
+      "step": 5385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.614e-05,
+      "loss": 1.3863,
+      "step": 5386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.613e-05,
+      "loss": 0.8053,
+      "step": 5387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.612e-05,
+      "loss": 1.802,
+      "step": 5388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6110000000000004e-05,
+      "loss": 0.6733,
+      "step": 5389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.61e-05,
+      "loss": 0.0101,
+      "step": 5390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.609e-05,
+      "loss": 0.0011,
+      "step": 5391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.608e-05,
+      "loss": 0.4216,
+      "step": 5392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6070000000000004e-05,
+      "loss": 0.2901,
+      "step": 5393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.606e-05,
+      "loss": 0.7388,
+      "step": 5394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.605e-05,
+      "loss": 1.0958,
+      "step": 5395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.604e-05,
+      "loss": 1.7693,
+      "step": 5396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.603e-05,
+      "loss": 1.0034,
+      "step": 5397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.602e-05,
+      "loss": 1.1567,
+      "step": 5398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.601e-05,
+      "loss": 1.3603,
+      "step": 5399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.0939,
+      "step": 5400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.599e-05,
+      "loss": 1.2424,
+      "step": 5401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5980000000000004e-05,
+      "loss": 0.1751,
+      "step": 5402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.597e-05,
+      "loss": 0.1497,
+      "step": 5403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.596e-05,
+      "loss": 0.3046,
+      "step": 5404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5950000000000006e-05,
+      "loss": 0.0715,
+      "step": 5405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.594e-05,
+      "loss": 0.789,
+      "step": 5406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.593e-05,
+      "loss": 1.239,
+      "step": 5407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.592e-05,
+      "loss": 0.9539,
+      "step": 5408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5910000000000006e-05,
+      "loss": 0.9281,
+      "step": 5409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5900000000000004e-05,
+      "loss": 0.9367,
+      "step": 5410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.589e-05,
+      "loss": 0.6931,
+      "step": 5411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.588e-05,
+      "loss": 1.5151,
+      "step": 5412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.587e-05,
+      "loss": 1.0214,
+      "step": 5413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5860000000000005e-05,
+      "loss": 0.3651,
+      "step": 5414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.585e-05,
+      "loss": 0.0963,
+      "step": 5415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.584e-05,
+      "loss": 0.4033,
+      "step": 5416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.583e-05,
+      "loss": 0.741,
+      "step": 5417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5820000000000005e-05,
+      "loss": 0.3992,
+      "step": 5418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5810000000000004e-05,
+      "loss": 1.0753,
+      "step": 5419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.58e-05,
+      "loss": 1.2357,
+      "step": 5420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.579e-05,
+      "loss": 1.2455,
+      "step": 5421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.578e-05,
+      "loss": 1.2005,
+      "step": 5422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5770000000000004e-05,
+      "loss": 1.1112,
+      "step": 5423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.576e-05,
+      "loss": 0.7221,
+      "step": 5424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.575e-05,
+      "loss": 1.3978,
+      "step": 5425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.574e-05,
+      "loss": 1.4117,
+      "step": 5426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.573e-05,
+      "loss": 0.4681,
+      "step": 5427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.572e-05,
+      "loss": 0.19,
+      "step": 5428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.571e-05,
+      "loss": 0.4231,
+      "step": 5429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5700000000000006e-05,
+      "loss": 0.4289,
+      "step": 5430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.569e-05,
+      "loss": 0.4951,
+      "step": 5431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.568e-05,
+      "loss": 1.4177,
+      "step": 5432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.567e-05,
+      "loss": 1.2655,
+      "step": 5433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.566e-05,
+      "loss": 0.9674,
+      "step": 5434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5650000000000005e-05,
+      "loss": 1.3555,
+      "step": 5435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.564e-05,
+      "loss": 1.0008,
+      "step": 5436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.563e-05,
+      "loss": 0.7633,
+      "step": 5437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.562e-05,
+      "loss": 1.1232,
+      "step": 5438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5610000000000005e-05,
+      "loss": 1.0092,
+      "step": 5439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 0.2051,
+      "step": 5440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.559e-05,
+      "loss": 0.0115,
+      "step": 5441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.558e-05,
+      "loss": 0.0772,
+      "step": 5442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.557e-05,
+      "loss": 0.811,
+      "step": 5443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5560000000000004e-05,
+      "loss": 1.1785,
+      "step": 5444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.555e-05,
+      "loss": 1.2437,
+      "step": 5445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.554000000000001e-05,
+      "loss": 1.0193,
+      "step": 5446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.553e-05,
+      "loss": 1.4297,
+      "step": 5447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5520000000000005e-05,
+      "loss": 1.2881,
+      "step": 5448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.551e-05,
+      "loss": 0.9401,
+      "step": 5449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.55e-05,
+      "loss": 1.0405,
+      "step": 5450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.549000000000001e-05,
+      "loss": 1.6392,
+      "step": 5451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.548e-05,
+      "loss": 0.2597,
+      "step": 5452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5470000000000003e-05,
+      "loss": 0.1616,
+      "step": 5453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.546e-05,
+      "loss": 0.1594,
+      "step": 5454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.545000000000001e-05,
+      "loss": 0.4451,
+      "step": 5455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5440000000000005e-05,
+      "loss": 0.5526,
+      "step": 5456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.543e-05,
+      "loss": 1.2674,
+      "step": 5457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.542e-05,
+      "loss": 1.2664,
+      "step": 5458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.541e-05,
+      "loss": 1.2997,
+      "step": 5459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 1.1401,
+      "step": 5460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5390000000000004e-05,
+      "loss": 1.3102,
+      "step": 5461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.538e-05,
+      "loss": 0.8156,
+      "step": 5462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.537e-05,
+      "loss": 0.899,
+      "step": 5463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.536e-05,
+      "loss": 0.4293,
+      "step": 5464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5350000000000005e-05,
+      "loss": 0.049,
+      "step": 5465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.534e-05,
+      "loss": 0.2279,
+      "step": 5466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.533e-05,
+      "loss": 0.3139,
+      "step": 5467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.532e-05,
+      "loss": 0.592,
+      "step": 5468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5310000000000005e-05,
+      "loss": 1.0257,
+      "step": 5469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.53e-05,
+      "loss": 1.2893,
+      "step": 5470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.529e-05,
+      "loss": 1.204,
+      "step": 5471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.528e-05,
+      "loss": 1.1393,
+      "step": 5472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.527e-05,
+      "loss": 1.3538,
+      "step": 5473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5260000000000004e-05,
+      "loss": 1.8312,
+      "step": 5474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.525e-05,
+      "loss": 1.0689,
+      "step": 5475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.524000000000001e-05,
+      "loss": 1.258,
+      "step": 5476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.523e-05,
+      "loss": 0.1044,
+      "step": 5477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5220000000000004e-05,
+      "loss": 0.044,
+      "step": 5478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.521e-05,
+      "loss": 0.1535,
+      "step": 5479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.52e-05,
+      "loss": 0.1994,
+      "step": 5480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5190000000000006e-05,
+      "loss": 0.8848,
+      "step": 5481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.518e-05,
+      "loss": 1.3138,
+      "step": 5482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.517e-05,
+      "loss": 1.238,
+      "step": 5483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.516e-05,
+      "loss": 1.1105,
+      "step": 5484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5150000000000006e-05,
+      "loss": 1.3165,
+      "step": 5485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5140000000000005e-05,
+      "loss": 0.9832,
+      "step": 5486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.513e-05,
+      "loss": 0.9333,
+      "step": 5487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.512e-05,
+      "loss": 0.9978,
+      "step": 5488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.511e-05,
+      "loss": 0.6356,
+      "step": 5489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5100000000000005e-05,
+      "loss": 0.1006,
+      "step": 5490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5090000000000004e-05,
+      "loss": 0.1256,
+      "step": 5491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.508e-05,
+      "loss": 0.3011,
+      "step": 5492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.507e-05,
+      "loss": 1.0827,
+      "step": 5493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.506e-05,
+      "loss": 1.0142,
+      "step": 5494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5050000000000004e-05,
+      "loss": 1.2071,
+      "step": 5495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.504e-05,
+      "loss": 1.1201,
+      "step": 5496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.503e-05,
+      "loss": 0.914,
+      "step": 5497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.502e-05,
+      "loss": 1.0027,
+      "step": 5498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5010000000000004e-05,
+      "loss": 1.1077,
+      "step": 5499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.5e-05,
+      "loss": 1.6071,
+      "step": 5500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.499e-05,
+      "loss": 1.4356,
+      "step": 5501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.498e-05,
+      "loss": 0.6409,
+      "step": 5502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.497e-05,
+      "loss": 0.311,
+      "step": 5503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.496e-05,
+      "loss": 0.0196,
+      "step": 5504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.495e-05,
+      "loss": 0.239,
+      "step": 5505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.494000000000001e-05,
+      "loss": 0.4802,
+      "step": 5506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.493e-05,
+      "loss": 0.9787,
+      "step": 5507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4920000000000004e-05,
+      "loss": 1.321,
+      "step": 5508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.491e-05,
+      "loss": 1.168,
+      "step": 5509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.49e-05,
+      "loss": 1.0323,
+      "step": 5510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4890000000000006e-05,
+      "loss": 0.9566,
+      "step": 5511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.488e-05,
+      "loss": 1.2673,
+      "step": 5512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.487e-05,
+      "loss": 1.0354,
+      "step": 5513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.486e-05,
+      "loss": 1.2549,
+      "step": 5514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4850000000000006e-05,
+      "loss": 0.1761,
+      "step": 5515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4840000000000004e-05,
+      "loss": 0.2805,
+      "step": 5516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.483e-05,
+      "loss": 0.1873,
+      "step": 5517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.482e-05,
+      "loss": 0.9654,
+      "step": 5518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.481e-05,
+      "loss": 0.3823,
+      "step": 5519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 0.8886,
+      "step": 5520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.479e-05,
+      "loss": 1.4743,
+      "step": 5521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.478e-05,
+      "loss": 0.8587,
+      "step": 5522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.477e-05,
+      "loss": 1.3786,
+      "step": 5523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4760000000000005e-05,
+      "loss": 1.047,
+      "step": 5524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4750000000000004e-05,
+      "loss": 0.9838,
+      "step": 5525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.474e-05,
+      "loss": 0.7146,
+      "step": 5526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.473e-05,
+      "loss": 0.005,
+      "step": 5527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.472e-05,
+      "loss": 0.3591,
+      "step": 5528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4710000000000004e-05,
+      "loss": 0.0337,
+      "step": 5529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.47e-05,
+      "loss": 0.2197,
+      "step": 5530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.469e-05,
+      "loss": 0.9197,
+      "step": 5531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.468e-05,
+      "loss": 1.3237,
+      "step": 5532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.467e-05,
+      "loss": 1.4721,
+      "step": 5533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.466e-05,
+      "loss": 1.2469,
+      "step": 5534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.465e-05,
+      "loss": 1.0519,
+      "step": 5535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4640000000000006e-05,
+      "loss": 1.1348,
+      "step": 5536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.463e-05,
+      "loss": 0.9167,
+      "step": 5537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.462e-05,
+      "loss": 1.4815,
+      "step": 5538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.461e-05,
+      "loss": 0.6434,
+      "step": 5539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.46e-05,
+      "loss": 0.2443,
+      "step": 5540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4590000000000005e-05,
+      "loss": 0.0314,
+      "step": 5541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.458e-05,
+      "loss": 0.1274,
+      "step": 5542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.457e-05,
+      "loss": 0.5201,
+      "step": 5543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.456e-05,
+      "loss": 1.1387,
+      "step": 5544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4550000000000005e-05,
+      "loss": 1.2466,
+      "step": 5545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4540000000000004e-05,
+      "loss": 1.0096,
+      "step": 5546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.453e-05,
+      "loss": 0.8595,
+      "step": 5547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.452e-05,
+      "loss": 1.195,
+      "step": 5548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.451e-05,
+      "loss": 0.7539,
+      "step": 5549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 1.7382,
+      "step": 5550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.449e-05,
+      "loss": 1.2585,
+      "step": 5551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.448e-05,
+      "loss": 0.1336,
+      "step": 5552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.447e-05,
+      "loss": 0.0678,
+      "step": 5553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4460000000000005e-05,
+      "loss": 0.1729,
+      "step": 5554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.445e-05,
+      "loss": 0.0819,
+      "step": 5555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.444e-05,
+      "loss": 1.0276,
+      "step": 5556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.443e-05,
+      "loss": 1.1599,
+      "step": 5557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.442e-05,
+      "loss": 1.2711,
+      "step": 5558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4410000000000003e-05,
+      "loss": 1.2408,
+      "step": 5559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.44e-05,
+      "loss": 0.776,
+      "step": 5560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.439000000000001e-05,
+      "loss": 0.9413,
+      "step": 5561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.438e-05,
+      "loss": 1.4296,
+      "step": 5562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4370000000000004e-05,
+      "loss": 1.1572,
+      "step": 5563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.436e-05,
+      "loss": 0.5419,
+      "step": 5564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.435e-05,
+      "loss": 0.231,
+      "step": 5565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4340000000000006e-05,
+      "loss": 0.0089,
+      "step": 5566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4330000000000004e-05,
+      "loss": 0.0588,
+      "step": 5567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.432e-05,
+      "loss": 0.9789,
+      "step": 5568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.431e-05,
+      "loss": 0.9693,
+      "step": 5569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.43e-05,
+      "loss": 1.1566,
+      "step": 5570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4290000000000005e-05,
+      "loss": 1.1923,
+      "step": 5571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.428e-05,
+      "loss": 1.2605,
+      "step": 5572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.427e-05,
+      "loss": 0.6582,
+      "step": 5573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.426e-05,
+      "loss": 0.863,
+      "step": 5574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4250000000000005e-05,
+      "loss": 1.2449,
+      "step": 5575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.424e-05,
+      "loss": 0.8193,
+      "step": 5576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.423e-05,
+      "loss": 0.1503,
+      "step": 5577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.422e-05,
+      "loss": 0.1715,
+      "step": 5578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.421e-05,
+      "loss": 0.3577,
+      "step": 5579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 0.4225,
+      "step": 5580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.419e-05,
+      "loss": 1.0551,
+      "step": 5581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.418000000000001e-05,
+      "loss": 1.319,
+      "step": 5582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.417e-05,
+      "loss": 1.1854,
+      "step": 5583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4160000000000004e-05,
+      "loss": 0.9067,
+      "step": 5584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.415e-05,
+      "loss": 1.1854,
+      "step": 5585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.414e-05,
+      "loss": 0.8176,
+      "step": 5586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4130000000000006e-05,
+      "loss": 1.0761,
+      "step": 5587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.412e-05,
+      "loss": 1.0395,
+      "step": 5588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.411e-05,
+      "loss": 0.7157,
+      "step": 5589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.41e-05,
+      "loss": 0.0911,
+      "step": 5590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4090000000000006e-05,
+      "loss": 0.1715,
+      "step": 5591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4080000000000005e-05,
+      "loss": 0.3237,
+      "step": 5592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.407e-05,
+      "loss": 1.1655,
+      "step": 5593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.406e-05,
+      "loss": 0.8944,
+      "step": 5594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.405e-05,
+      "loss": 1.3424,
+      "step": 5595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4040000000000005e-05,
+      "loss": 1.3017,
+      "step": 5596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4030000000000004e-05,
+      "loss": 1.1716,
+      "step": 5597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.402e-05,
+      "loss": 1.406,
+      "step": 5598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.401e-05,
+      "loss": 1.2272,
+      "step": 5599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.5965,
+      "step": 5600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3990000000000004e-05,
+      "loss": 1.1656,
+      "step": 5601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.398e-05,
+      "loss": 0.2158,
+      "step": 5602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.397e-05,
+      "loss": 0.1524,
+      "step": 5603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.396e-05,
+      "loss": 0.0411,
+      "step": 5604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3950000000000004e-05,
+      "loss": 0.1131,
+      "step": 5605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.394e-05,
+      "loss": 0.6383,
+      "step": 5606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.393e-05,
+      "loss": 1.2152,
+      "step": 5607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.392e-05,
+      "loss": 1.026,
+      "step": 5608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.391e-05,
+      "loss": 1.2714,
+      "step": 5609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.39e-05,
+      "loss": 1.0481,
+      "step": 5610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.389e-05,
+      "loss": 1.2006,
+      "step": 5611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.388000000000001e-05,
+      "loss": 1.1971,
+      "step": 5612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.387e-05,
+      "loss": 1.2608,
+      "step": 5613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3860000000000004e-05,
+      "loss": 1.0125,
+      "step": 5614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.385e-05,
+      "loss": 0.1555,
+      "step": 5615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.384e-05,
+      "loss": 0.0797,
+      "step": 5616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3830000000000006e-05,
+      "loss": 0.259,
+      "step": 5617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.382e-05,
+      "loss": 0.4794,
+      "step": 5618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.381e-05,
+      "loss": 1.2446,
+      "step": 5619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.38e-05,
+      "loss": 0.895,
+      "step": 5620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3790000000000006e-05,
+      "loss": 1.2044,
+      "step": 5621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3780000000000004e-05,
+      "loss": 1.4715,
+      "step": 5622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.377e-05,
+      "loss": 0.8243,
+      "step": 5623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.376e-05,
+      "loss": 1.4164,
+      "step": 5624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.375e-05,
+      "loss": 0.7601,
+      "step": 5625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3740000000000005e-05,
+      "loss": 1.0358,
+      "step": 5626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.373e-05,
+      "loss": 0.6893,
+      "step": 5627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.372e-05,
+      "loss": 0.267,
+      "step": 5628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.371e-05,
+      "loss": 0.2661,
+      "step": 5629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3700000000000005e-05,
+      "loss": 0.3972,
+      "step": 5630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3690000000000004e-05,
+      "loss": 0.9343,
+      "step": 5631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.368e-05,
+      "loss": 1.0746,
+      "step": 5632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.367e-05,
+      "loss": 1.4983,
+      "step": 5633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.366e-05,
+      "loss": 1.1817,
+      "step": 5634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3650000000000004e-05,
+      "loss": 1.2269,
+      "step": 5635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.364e-05,
+      "loss": 1.076,
+      "step": 5636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.363000000000001e-05,
+      "loss": 1.4039,
+      "step": 5637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.362e-05,
+      "loss": 1.6912,
+      "step": 5638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.361e-05,
+      "loss": 1.0565,
+      "step": 5639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.36e-05,
+      "loss": 0.0849,
+      "step": 5640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.359e-05,
+      "loss": 0.0441,
+      "step": 5641
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3580000000000006e-05,
+      "loss": 0.7099,
+      "step": 5642
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.357e-05,
+      "loss": 0.3215,
+      "step": 5643
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.356e-05,
+      "loss": 0.7425,
+      "step": 5644
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.355e-05,
+      "loss": 0.866,
+      "step": 5645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.354e-05,
+      "loss": 1.3652,
+      "step": 5646
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3530000000000005e-05,
+      "loss": 1.1971,
+      "step": 5647
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.352e-05,
+      "loss": 1.0441,
+      "step": 5648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.351e-05,
+      "loss": 1.4314,
+      "step": 5649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.35e-05,
+      "loss": 1.2621,
+      "step": 5650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3490000000000005e-05,
+      "loss": 1.1055,
+      "step": 5651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3480000000000004e-05,
+      "loss": 0.188,
+      "step": 5652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.347e-05,
+      "loss": 0.0022,
+      "step": 5653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.346e-05,
+      "loss": 0.1301,
+      "step": 5654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.345e-05,
+      "loss": 0.849,
+      "step": 5655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3440000000000004e-05,
+      "loss": 1.1812,
+      "step": 5656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.343e-05,
+      "loss": 1.2206,
+      "step": 5657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.342e-05,
+      "loss": 1.0233,
+      "step": 5658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.341e-05,
+      "loss": 1.2254,
+      "step": 5659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 1.0815,
+      "step": 5660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.339e-05,
+      "loss": 1.0836,
+      "step": 5661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.338e-05,
+      "loss": 1.587,
+      "step": 5662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.337e-05,
+      "loss": 1.0722,
+      "step": 5663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.336e-05,
+      "loss": 0.7884,
+      "step": 5664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.335e-05,
+      "loss": 0.6483,
+      "step": 5665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.334e-05,
+      "loss": 0.1972,
+      "step": 5666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.333000000000001e-05,
+      "loss": 0.0422,
+      "step": 5667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.332e-05,
+      "loss": 0.8971,
+      "step": 5668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3310000000000004e-05,
+      "loss": 1.2238,
+      "step": 5669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.33e-05,
+      "loss": 1.2836,
+      "step": 5670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.329e-05,
+      "loss": 1.041,
+      "step": 5671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3280000000000006e-05,
+      "loss": 0.9578,
+      "step": 5672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.327e-05,
+      "loss": 0.9216,
+      "step": 5673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.326e-05,
+      "loss": 1.2303,
+      "step": 5674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.325e-05,
+      "loss": 0.8069,
+      "step": 5675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.324e-05,
+      "loss": 0.9544,
+      "step": 5676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3230000000000005e-05,
+      "loss": 0.278,
+      "step": 5677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3219999999999996e-05,
+      "loss": 0.0669,
+      "step": 5678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.321e-05,
+      "loss": 0.131,
+      "step": 5679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.32e-05,
+      "loss": 0.6999,
+      "step": 5680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3190000000000005e-05,
+      "loss": 1.0213,
+      "step": 5681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.318e-05,
+      "loss": 1.1845,
+      "step": 5682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.317e-05,
+      "loss": 0.9584,
+      "step": 5683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.316e-05,
+      "loss": 1.1005,
+      "step": 5684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.315e-05,
+      "loss": 1.6594,
+      "step": 5685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3140000000000004e-05,
+      "loss": 1.4479,
+      "step": 5686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.313e-05,
+      "loss": 1.1086,
+      "step": 5687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.312000000000001e-05,
+      "loss": 1.3777,
+      "step": 5688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.311e-05,
+      "loss": 1.0902,
+      "step": 5689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3100000000000004e-05,
+      "loss": 0.2912,
+      "step": 5690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.309e-05,
+      "loss": 0.1029,
+      "step": 5691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.308e-05,
+      "loss": 0.2421,
+      "step": 5692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3070000000000006e-05,
+      "loss": 0.3285,
+      "step": 5693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.306e-05,
+      "loss": 0.3455,
+      "step": 5694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.305e-05,
+      "loss": 0.8913,
+      "step": 5695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.304e-05,
+      "loss": 1.1104,
+      "step": 5696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3030000000000006e-05,
+      "loss": 1.025,
+      "step": 5697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3020000000000005e-05,
+      "loss": 0.8433,
+      "step": 5698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.301e-05,
+      "loss": 1.0391,
+      "step": 5699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.3e-05,
+      "loss": 1.0803,
+      "step": 5700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.299e-05,
+      "loss": 1.6139,
+      "step": 5701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2980000000000005e-05,
+      "loss": 0.4952,
+      "step": 5702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2970000000000004e-05,
+      "loss": 0.1671,
+      "step": 5703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.296e-05,
+      "loss": 0.2032,
+      "step": 5704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.295e-05,
+      "loss": 0.0739,
+      "step": 5705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2940000000000006e-05,
+      "loss": 0.9447,
+      "step": 5706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2930000000000004e-05,
+      "loss": 1.3584,
+      "step": 5707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.292e-05,
+      "loss": 0.9854,
+      "step": 5708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.291e-05,
+      "loss": 0.9654,
+      "step": 5709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.29e-05,
+      "loss": 1.552,
+      "step": 5710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2890000000000004e-05,
+      "loss": 0.8661,
+      "step": 5711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.288e-05,
+      "loss": 1.0202,
+      "step": 5712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.287000000000001e-05,
+      "loss": 1.2952,
+      "step": 5713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.286e-05,
+      "loss": 0.1654,
+      "step": 5714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.285e-05,
+      "loss": 0.0589,
+      "step": 5715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.284e-05,
+      "loss": 0.1026,
+      "step": 5716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.283e-05,
+      "loss": 0.2426,
+      "step": 5717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.282000000000001e-05,
+      "loss": 0.8315,
+      "step": 5718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.281e-05,
+      "loss": 0.873,
+      "step": 5719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 1.3271,
+      "step": 5720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.279e-05,
+      "loss": 1.3121,
+      "step": 5721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.278e-05,
+      "loss": 1.0631,
+      "step": 5722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2770000000000006e-05,
+      "loss": 1.277,
+      "step": 5723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.276e-05,
+      "loss": 1.4512,
+      "step": 5724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.275e-05,
+      "loss": 0.8374,
+      "step": 5725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.274e-05,
+      "loss": 1.029,
+      "step": 5726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2730000000000006e-05,
+      "loss": 0.0891,
+      "step": 5727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2720000000000004e-05,
+      "loss": 0.1111,
+      "step": 5728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.271e-05,
+      "loss": 0.0756,
+      "step": 5729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.27e-05,
+      "loss": 0.6329,
+      "step": 5730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.269e-05,
+      "loss": 1.4259,
+      "step": 5731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2680000000000005e-05,
+      "loss": 1.1072,
+      "step": 5732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.267e-05,
+      "loss": 0.9667,
+      "step": 5733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.266e-05,
+      "loss": 1.1032,
+      "step": 5734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.265e-05,
+      "loss": 1.2125,
+      "step": 5735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2640000000000005e-05,
+      "loss": 1.5111,
+      "step": 5736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2630000000000004e-05,
+      "loss": 1.2816,
+      "step": 5737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.262e-05,
+      "loss": 1.2431,
+      "step": 5738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.261e-05,
+      "loss": 0.9837,
+      "step": 5739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.26e-05,
+      "loss": 0.2996,
+      "step": 5740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2590000000000004e-05,
+      "loss": 0.0134,
+      "step": 5741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.258e-05,
+      "loss": 0.0741,
+      "step": 5742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.257000000000001e-05,
+      "loss": 0.4496,
+      "step": 5743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.256e-05,
+      "loss": 1.428,
+      "step": 5744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2550000000000004e-05,
+      "loss": 1.1785,
+      "step": 5745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.254e-05,
+      "loss": 1.2677,
+      "step": 5746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.253e-05,
+      "loss": 1.1162,
+      "step": 5747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2520000000000006e-05,
+      "loss": 0.7273,
+      "step": 5748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.251e-05,
+      "loss": 1.1077,
+      "step": 5749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.25e-05,
+      "loss": 1.6009,
+      "step": 5750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.249e-05,
+      "loss": 1.0295,
+      "step": 5751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.248e-05,
+      "loss": 0.4661,
+      "step": 5752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2470000000000005e-05,
+      "loss": 0.1387,
+      "step": 5753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.246e-05,
+      "loss": 0.1539,
+      "step": 5754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.245e-05,
+      "loss": 0.075,
+      "step": 5755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.244e-05,
+      "loss": 0.6506,
+      "step": 5756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2430000000000005e-05,
+      "loss": 0.8712,
+      "step": 5757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2420000000000004e-05,
+      "loss": 0.8536,
+      "step": 5758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.241e-05,
+      "loss": 1.3422,
+      "step": 5759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.24e-05,
+      "loss": 0.8392,
+      "step": 5760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.239e-05,
+      "loss": 1.3509,
+      "step": 5761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2380000000000004e-05,
+      "loss": 1.299,
+      "step": 5762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.237e-05,
+      "loss": 1.1823,
+      "step": 5763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.236e-05,
+      "loss": 0.4533,
+      "step": 5764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.235e-05,
+      "loss": 0.0213,
+      "step": 5765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2340000000000005e-05,
+      "loss": 0.2456,
+      "step": 5766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.233e-05,
+      "loss": 0.0957,
+      "step": 5767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.232e-05,
+      "loss": 0.6006,
+      "step": 5768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.231e-05,
+      "loss": 1.4396,
+      "step": 5769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.23e-05,
+      "loss": 1.4178,
+      "step": 5770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.229e-05,
+      "loss": 1.4341,
+      "step": 5771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.228e-05,
+      "loss": 1.5889,
+      "step": 5772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.227000000000001e-05,
+      "loss": 1.3231,
+      "step": 5773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.226e-05,
+      "loss": 0.6914,
+      "step": 5774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2250000000000004e-05,
+      "loss": 1.2528,
+      "step": 5775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.224e-05,
+      "loss": 0.9849,
+      "step": 5776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.223e-05,
+      "loss": 0.3425,
+      "step": 5777
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2220000000000006e-05,
+      "loss": 0.239,
+      "step": 5778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.221e-05,
+      "loss": 0.0294,
+      "step": 5779
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.22e-05,
+      "loss": 0.0989,
+      "step": 5780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.219e-05,
+      "loss": 1.1178,
+      "step": 5781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2180000000000006e-05,
+      "loss": 1.2588,
+      "step": 5782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2170000000000005e-05,
+      "loss": 1.0414,
+      "step": 5783
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2159999999999996e-05,
+      "loss": 1.4248,
+      "step": 5784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.215e-05,
+      "loss": 1.1105,
+      "step": 5785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.214e-05,
+      "loss": 0.9114,
+      "step": 5786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2130000000000005e-05,
+      "loss": 1.3226,
+      "step": 5787
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.212e-05,
+      "loss": 0.9179,
+      "step": 5788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.211e-05,
+      "loss": 0.5443,
+      "step": 5789
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.21e-05,
+      "loss": 0.3001,
+      "step": 5790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.209e-05,
+      "loss": 0.163,
+      "step": 5791
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2080000000000004e-05,
+      "loss": 0.2072,
+      "step": 5792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.207e-05,
+      "loss": 0.5021,
+      "step": 5793
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.206e-05,
+      "loss": 0.8021,
+      "step": 5794
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.205e-05,
+      "loss": 0.992,
+      "step": 5795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2040000000000004e-05,
+      "loss": 1.2443,
+      "step": 5796
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.203e-05,
+      "loss": 1.2431,
+      "step": 5797
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.202e-05,
+      "loss": 1.4314,
+      "step": 5798
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.201e-05,
+      "loss": 1.1948,
+      "step": 5799
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2e-05,
+      "loss": 0.7622,
+      "step": 5800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.199e-05,
+      "loss": 1.3466,
+      "step": 5801
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.198e-05,
+      "loss": 0.1613,
+      "step": 5802
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1970000000000006e-05,
+      "loss": 0.2097,
+      "step": 5803
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.196e-05,
+      "loss": 0.134,
+      "step": 5804
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.195e-05,
+      "loss": 0.442,
+      "step": 5805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.194e-05,
+      "loss": 0.7073,
+      "step": 5806
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.193e-05,
+      "loss": 1.5959,
+      "step": 5807
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1920000000000005e-05,
+      "loss": 1.0016,
+      "step": 5808
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.191e-05,
+      "loss": 0.9529,
+      "step": 5809
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.19e-05,
+      "loss": 1.4016,
+      "step": 5810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.189e-05,
+      "loss": 1.0579,
+      "step": 5811
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1880000000000006e-05,
+      "loss": 1.3744,
+      "step": 5812
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1870000000000004e-05,
+      "loss": 1.1744,
+      "step": 5813
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.186e-05,
+      "loss": 0.4285,
+      "step": 5814
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.185e-05,
+      "loss": 0.0766,
+      "step": 5815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.184e-05,
+      "loss": 0.067,
+      "step": 5816
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1830000000000004e-05,
+      "loss": 0.4626,
+      "step": 5817
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.182e-05,
+      "loss": 0.3313,
+      "step": 5818
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.181000000000001e-05,
+      "loss": 0.712,
+      "step": 5819
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.18e-05,
+      "loss": 1.4619,
+      "step": 5820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.179e-05,
+      "loss": 1.5789,
+      "step": 5821
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.178e-05,
+      "loss": 0.9179,
+      "step": 5822
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.177e-05,
+      "loss": 1.2043,
+      "step": 5823
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.176000000000001e-05,
+      "loss": 1.3546,
+      "step": 5824
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.175e-05,
+      "loss": 1.3573,
+      "step": 5825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1740000000000004e-05,
+      "loss": 0.8646,
+      "step": 5826
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.173e-05,
+      "loss": 0.3795,
+      "step": 5827
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.172e-05,
+      "loss": 0.0627,
+      "step": 5828
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1710000000000006e-05,
+      "loss": 0.016,
+      "step": 5829
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.17e-05,
+      "loss": 0.4052,
+      "step": 5830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.169e-05,
+      "loss": 0.9754,
+      "step": 5831
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.168e-05,
+      "loss": 1.3161,
+      "step": 5832
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1670000000000006e-05,
+      "loss": 0.8151,
+      "step": 5833
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1660000000000004e-05,
+      "loss": 0.9716,
+      "step": 5834
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.165e-05,
+      "loss": 1.0253,
+      "step": 5835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.164e-05,
+      "loss": 1.4577,
+      "step": 5836
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.163e-05,
+      "loss": 1.2697,
+      "step": 5837
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1620000000000005e-05,
+      "loss": 0.8936,
+      "step": 5838
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.161e-05,
+      "loss": 0.6796,
+      "step": 5839
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.16e-05,
+      "loss": 0.3503,
+      "step": 5840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.159e-05,
+      "loss": 0.1849,
+      "step": 5841
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1580000000000005e-05,
+      "loss": 0.0198,
+      "step": 5842
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1570000000000003e-05,
+      "loss": 0.1802,
+      "step": 5843
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.156e-05,
+      "loss": 0.8585,
+      "step": 5844
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.155e-05,
+      "loss": 1.3334,
+      "step": 5845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.154e-05,
+      "loss": 1.0411,
+      "step": 5846
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1530000000000004e-05,
+      "loss": 1.1519,
+      "step": 5847
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.152e-05,
+      "loss": 1.256,
+      "step": 5848
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.151000000000001e-05,
+      "loss": 1.0403,
+      "step": 5849
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.15e-05,
+      "loss": 1.0887,
+      "step": 5850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1490000000000004e-05,
+      "loss": 1.0646,
+      "step": 5851
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.148e-05,
+      "loss": 0.2511,
+      "step": 5852
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.147e-05,
+      "loss": 0.1002,
+      "step": 5853
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1460000000000006e-05,
+      "loss": 0.0008,
+      "step": 5854
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.145e-05,
+      "loss": 0.104,
+      "step": 5855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.144e-05,
+      "loss": 0.6017,
+      "step": 5856
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.143e-05,
+      "loss": 0.6605,
+      "step": 5857
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.142000000000001e-05,
+      "loss": 1.2057,
+      "step": 5858
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1410000000000005e-05,
+      "loss": 1.2766,
+      "step": 5859
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.14e-05,
+      "loss": 1.5034,
+      "step": 5860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.139e-05,
+      "loss": 1.1689,
+      "step": 5861
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.138e-05,
+      "loss": 0.9028,
+      "step": 5862
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1370000000000005e-05,
+      "loss": 1.0238,
+      "step": 5863
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1360000000000004e-05,
+      "loss": 0.5368,
+      "step": 5864
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.135e-05,
+      "loss": 0.1117,
+      "step": 5865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.134e-05,
+      "loss": 0.2173,
+      "step": 5866
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.133e-05,
+      "loss": 0.1364,
+      "step": 5867
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1320000000000004e-05,
+      "loss": 1.2329,
+      "step": 5868
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.131e-05,
+      "loss": 0.7069,
+      "step": 5869
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.13e-05,
+      "loss": 1.3997,
+      "step": 5870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.129e-05,
+      "loss": 1.0954,
+      "step": 5871
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1280000000000005e-05,
+      "loss": 1.3485,
+      "step": 5872
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.127e-05,
+      "loss": 1.1611,
+      "step": 5873
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.126e-05,
+      "loss": 1.054,
+      "step": 5874
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.125e-05,
+      "loss": 1.939,
+      "step": 5875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.124e-05,
+      "loss": 1.1059,
+      "step": 5876
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.123e-05,
+      "loss": 0.8122,
+      "step": 5877
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.122e-05,
+      "loss": 0.0877,
+      "step": 5878
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.121000000000001e-05,
+      "loss": 0.0585,
+      "step": 5879
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.12e-05,
+      "loss": 0.2834,
+      "step": 5880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1190000000000004e-05,
+      "loss": 0.5312,
+      "step": 5881
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.118e-05,
+      "loss": 1.0555,
+      "step": 5882
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.117e-05,
+      "loss": 1.2094,
+      "step": 5883
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1160000000000006e-05,
+      "loss": 0.974,
+      "step": 5884
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.115e-05,
+      "loss": 1.1209,
+      "step": 5885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.114e-05,
+      "loss": 1.0288,
+      "step": 5886
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.113e-05,
+      "loss": 0.8818,
+      "step": 5887
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1120000000000006e-05,
+      "loss": 1.3665,
+      "step": 5888
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1110000000000005e-05,
+      "loss": 0.6969,
+      "step": 5889
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.11e-05,
+      "loss": 0.3354,
+      "step": 5890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.109e-05,
+      "loss": 0.1811,
+      "step": 5891
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.108e-05,
+      "loss": 0.331,
+      "step": 5892
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1070000000000005e-05,
+      "loss": 0.51,
+      "step": 5893
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.106e-05,
+      "loss": 0.8682,
+      "step": 5894
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.105e-05,
+      "loss": 1.2671,
+      "step": 5895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.104e-05,
+      "loss": 1.2005,
+      "step": 5896
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.103e-05,
+      "loss": 1.0473,
+      "step": 5897
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1020000000000004e-05,
+      "loss": 0.8451,
+      "step": 5898
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.101e-05,
+      "loss": 1.0718,
+      "step": 5899
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1e-05,
+      "loss": 1.2695,
+      "step": 5900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.099e-05,
+      "loss": 1.2728,
+      "step": 5901
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0980000000000004e-05,
+      "loss": 0.0215,
+      "step": 5902
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.097e-05,
+      "loss": 0.0922,
+      "step": 5903
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.096e-05,
+      "loss": 0.069,
+      "step": 5904
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.095e-05,
+      "loss": 0.5213,
+      "step": 5905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.094e-05,
+      "loss": 1.312,
+      "step": 5906
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.093e-05,
+      "loss": 1.4147,
+      "step": 5907
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.092e-05,
+      "loss": 0.9861,
+      "step": 5908
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0910000000000006e-05,
+      "loss": 1.5225,
+      "step": 5909
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.09e-05,
+      "loss": 1.32,
+      "step": 5910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.089e-05,
+      "loss": 0.9554,
+      "step": 5911
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.088e-05,
+      "loss": 1.1032,
+      "step": 5912
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.087e-05,
+      "loss": 1.6749,
+      "step": 5913
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0860000000000005e-05,
+      "loss": 0.7507,
+      "step": 5914
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.085e-05,
+      "loss": 0.1028,
+      "step": 5915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.084e-05,
+      "loss": 0.3191,
+      "step": 5916
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.083e-05,
+      "loss": 0.0844,
+      "step": 5917
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0820000000000006e-05,
+      "loss": 0.4157,
+      "step": 5918
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0810000000000004e-05,
+      "loss": 1.3759,
+      "step": 5919
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.08e-05,
+      "loss": 1.0556,
+      "step": 5920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.079e-05,
+      "loss": 1.1146,
+      "step": 5921
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.078e-05,
+      "loss": 1.0654,
+      "step": 5922
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0770000000000004e-05,
+      "loss": 1.163,
+      "step": 5923
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.076e-05,
+      "loss": 1.082,
+      "step": 5924
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.075e-05,
+      "loss": 1.2779,
+      "step": 5925
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.074e-05,
+      "loss": 0.9454,
+      "step": 5926
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0730000000000005e-05,
+      "loss": 0.2353,
+      "step": 5927
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.072e-05,
+      "loss": 0.0028,
+      "step": 5928
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.071e-05,
+      "loss": 0.5275,
+      "step": 5929
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.07e-05,
+      "loss": 0.5987,
+      "step": 5930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.069e-05,
+      "loss": 0.9041,
+      "step": 5931
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0680000000000004e-05,
+      "loss": 1.5812,
+      "step": 5932
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.067e-05,
+      "loss": 1.2861,
+      "step": 5933
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.066e-05,
+      "loss": 1.4724,
+      "step": 5934
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.065e-05,
+      "loss": 0.9368,
+      "step": 5935
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.064e-05,
+      "loss": 1.0358,
+      "step": 5936
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.063e-05,
+      "loss": 1.0555,
+      "step": 5937
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.062e-05,
+      "loss": 1.2885,
+      "step": 5938
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0610000000000006e-05,
+      "loss": 0.5409,
+      "step": 5939
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 0.1906,
+      "step": 5940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.059e-05,
+      "loss": 0.0568,
+      "step": 5941
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.058e-05,
+      "loss": 0.1165,
+      "step": 5942
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.057e-05,
+      "loss": 0.128,
+      "step": 5943
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0560000000000005e-05,
+      "loss": 1.0674,
+      "step": 5944
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.055e-05,
+      "loss": 1.4521,
+      "step": 5945
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.054e-05,
+      "loss": 1.1351,
+      "step": 5946
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.053e-05,
+      "loss": 1.2456,
+      "step": 5947
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0520000000000005e-05,
+      "loss": 0.7645,
+      "step": 5948
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0510000000000003e-05,
+      "loss": 1.7158,
+      "step": 5949
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.05e-05,
+      "loss": 1.2858,
+      "step": 5950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.049e-05,
+      "loss": 1.15,
+      "step": 5951
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.048e-05,
+      "loss": 0.2481,
+      "step": 5952
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0470000000000004e-05,
+      "loss": 0.1365,
+      "step": 5953
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.046e-05,
+      "loss": 0.1647,
+      "step": 5954
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.045000000000001e-05,
+      "loss": 0.3332,
+      "step": 5955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.044e-05,
+      "loss": 0.8063,
+      "step": 5956
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0430000000000004e-05,
+      "loss": 0.9348,
+      "step": 5957
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.042e-05,
+      "loss": 1.4339,
+      "step": 5958
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.041e-05,
+      "loss": 1.1698,
+      "step": 5959
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 0.9827,
+      "step": 5960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.039e-05,
+      "loss": 1.0336,
+      "step": 5961
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.038e-05,
+      "loss": 1.1376,
+      "step": 5962
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.037e-05,
+      "loss": 1.2794,
+      "step": 5963
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0360000000000007e-05,
+      "loss": 0.6294,
+      "step": 5964
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0350000000000005e-05,
+      "loss": 0.1642,
+      "step": 5965
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.034e-05,
+      "loss": 0.1874,
+      "step": 5966
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.033e-05,
+      "loss": 0.2764,
+      "step": 5967
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.032e-05,
+      "loss": 0.774,
+      "step": 5968
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0310000000000005e-05,
+      "loss": 1.0702,
+      "step": 5969
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0300000000000004e-05,
+      "loss": 0.8688,
+      "step": 5970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.029e-05,
+      "loss": 0.796,
+      "step": 5971
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.028e-05,
+      "loss": 1.4026,
+      "step": 5972
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.027e-05,
+      "loss": 1.4462,
+      "step": 5973
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0260000000000004e-05,
+      "loss": 0.9604,
+      "step": 5974
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.025e-05,
+      "loss": 1.0041,
+      "step": 5975
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.024e-05,
+      "loss": 1.5481,
+      "step": 5976
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.023e-05,
+      "loss": 0.2645,
+      "step": 5977
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0220000000000005e-05,
+      "loss": 0.0956,
+      "step": 5978
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.021e-05,
+      "loss": 0.2357,
+      "step": 5979
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.02e-05,
+      "loss": 0.6741,
+      "step": 5980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.019e-05,
+      "loss": 1.2147,
+      "step": 5981
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.018e-05,
+      "loss": 1.209,
+      "step": 5982
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.017e-05,
+      "loss": 1.3593,
+      "step": 5983
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.016e-05,
+      "loss": 1.3167,
+      "step": 5984
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.015000000000001e-05,
+      "loss": 0.674,
+      "step": 5985
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.014e-05,
+      "loss": 0.7676,
+      "step": 5986
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0130000000000004e-05,
+      "loss": 1.6603,
+      "step": 5987
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.012e-05,
+      "loss": 1.1664,
+      "step": 5988
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.011e-05,
+      "loss": 0.5595,
+      "step": 5989
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0100000000000006e-05,
+      "loss": 0.5468,
+      "step": 5990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.009e-05,
+      "loss": 0.819,
+      "step": 5991
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.008e-05,
+      "loss": 0.0126,
+      "step": 5992
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.007e-05,
+      "loss": 0.7977,
+      "step": 5993
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0060000000000006e-05,
+      "loss": 0.8078,
+      "step": 5994
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0050000000000004e-05,
+      "loss": 1.3397,
+      "step": 5995
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.004e-05,
+      "loss": 1.1908,
+      "step": 5996
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.003e-05,
+      "loss": 1.545,
+      "step": 5997
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.002e-05,
+      "loss": 0.9858,
+      "step": 5998
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.0010000000000005e-05,
+      "loss": 0.7808,
+      "step": 5999
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4e-05,
+      "loss": 1.4056,
+      "step": 6000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.999e-05,
+      "loss": 1.2696,
+      "step": 6001
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.998e-05,
+      "loss": 0.5891,
+      "step": 6002
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9970000000000005e-05,
+      "loss": 0.0601,
+      "step": 6003
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9960000000000004e-05,
+      "loss": 0.0045,
+      "step": 6004
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.995e-05,
+      "loss": 0.2565,
+      "step": 6005
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.994e-05,
+      "loss": 0.2381,
+      "step": 6006
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993e-05,
+      "loss": 1.0424,
+      "step": 6007
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9920000000000004e-05,
+      "loss": 0.9999,
+      "step": 6008
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991e-05,
+      "loss": 0.9418,
+      "step": 6009
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.99e-05,
+      "loss": 1.0179,
+      "step": 6010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989e-05,
+      "loss": 0.7088,
+      "step": 6011
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988e-05,
+      "loss": 1.0541,
+      "step": 6012
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987e-05,
+      "loss": 1.3974,
+      "step": 6013
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.986e-05,
+      "loss": 0.9603,
+      "step": 6014
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9850000000000006e-05,
+      "loss": 0.0928,
+      "step": 6015
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.984e-05,
+      "loss": 0.1365,
+      "step": 6016
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.983e-05,
+      "loss": 0.0849,
+      "step": 6017
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.982e-05,
+      "loss": 1.041,
+      "step": 6018
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.981e-05,
+      "loss": 0.9722,
+      "step": 6019
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 1.0902,
+      "step": 6020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.979e-05,
+      "loss": 0.9968,
+      "step": 6021
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.978e-05,
+      "loss": 1.2901,
+      "step": 6022
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.977e-05,
+      "loss": 1.2572,
+      "step": 6023
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9760000000000006e-05,
+      "loss": 1.51,
+      "step": 6024
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9750000000000004e-05,
+      "loss": 1.7849,
+      "step": 6025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.974e-05,
+      "loss": 1.0506,
+      "step": 6026
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.973e-05,
+      "loss": 0.5037,
+      "step": 6027
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.972e-05,
+      "loss": 0.1325,
+      "step": 6028
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9710000000000004e-05,
+      "loss": 0.02,
+      "step": 6029
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.97e-05,
+      "loss": 0.6723,
+      "step": 6030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.969e-05,
+      "loss": 0.4039,
+      "step": 6031
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.968e-05,
+      "loss": 1.4176,
+      "step": 6032
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9670000000000005e-05,
+      "loss": 1.3419,
+      "step": 6033
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.966e-05,
+      "loss": 0.9454,
+      "step": 6034
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.965e-05,
+      "loss": 1.1041,
+      "step": 6035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.964e-05,
+      "loss": 1.4976,
+      "step": 6036
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.963e-05,
+      "loss": 1.4994,
+      "step": 6037
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9620000000000004e-05,
+      "loss": 1.18,
+      "step": 6038
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.961e-05,
+      "loss": 0.8115,
+      "step": 6039
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 0.1534,
+      "step": 6040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.959e-05,
+      "loss": 0.1012,
+      "step": 6041
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.958e-05,
+      "loss": 0.0069,
+      "step": 6042
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.957e-05,
+      "loss": 0.3262,
+      "step": 6043
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.956e-05,
+      "loss": 0.8057,
+      "step": 6044
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9550000000000006e-05,
+      "loss": 1.4052,
+      "step": 6045
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.954e-05,
+      "loss": 1.4813,
+      "step": 6046
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.953e-05,
+      "loss": 1.2994,
+      "step": 6047
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.952e-05,
+      "loss": 1.001,
+      "step": 6048
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.951e-05,
+      "loss": 0.9968,
+      "step": 6049
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 1.4702,
+      "step": 6050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9489999999999996e-05,
+      "loss": 1.3915,
+      "step": 6051
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.948e-05,
+      "loss": 0.1876,
+      "step": 6052
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.947e-05,
+      "loss": 0.2309,
+      "step": 6053
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9460000000000005e-05,
+      "loss": 0.3028,
+      "step": 6054
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9450000000000003e-05,
+      "loss": 0.1469,
+      "step": 6055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.944e-05,
+      "loss": 0.7147,
+      "step": 6056
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.943e-05,
+      "loss": 1.0227,
+      "step": 6057
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.942e-05,
+      "loss": 1.094,
+      "step": 6058
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9410000000000004e-05,
+      "loss": 1.2631,
+      "step": 6059
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.94e-05,
+      "loss": 0.8617,
+      "step": 6060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.939e-05,
+      "loss": 0.9828,
+      "step": 6061
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.938e-05,
+      "loss": 1.1696,
+      "step": 6062
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9370000000000004e-05,
+      "loss": 1.0899,
+      "step": 6063
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.936e-05,
+      "loss": 1.0945,
+      "step": 6064
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.935e-05,
+      "loss": 0.351,
+      "step": 6065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9340000000000006e-05,
+      "loss": 0.5475,
+      "step": 6066
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.933e-05,
+      "loss": 0.0471,
+      "step": 6067
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.932e-05,
+      "loss": 0.5103,
+      "step": 6068
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.931e-05,
+      "loss": 0.8149,
+      "step": 6069
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9300000000000007e-05,
+      "loss": 0.9974,
+      "step": 6070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9290000000000005e-05,
+      "loss": 1.5561,
+      "step": 6071
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9280000000000003e-05,
+      "loss": 1.1635,
+      "step": 6072
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.927e-05,
+      "loss": 1.1617,
+      "step": 6073
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.926e-05,
+      "loss": 0.8855,
+      "step": 6074
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9250000000000005e-05,
+      "loss": 0.9195,
+      "step": 6075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9240000000000004e-05,
+      "loss": 0.8684,
+      "step": 6076
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.923e-05,
+      "loss": 0.3073,
+      "step": 6077
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.922e-05,
+      "loss": 0.0641,
+      "step": 6078
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.921e-05,
+      "loss": 0.3133,
+      "step": 6079
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 0.9772,
+      "step": 6080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.919e-05,
+      "loss": 1.3696,
+      "step": 6081
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.918e-05,
+      "loss": 0.9082,
+      "step": 6082
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.917e-05,
+      "loss": 1.0449,
+      "step": 6083
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9160000000000005e-05,
+      "loss": 1.2165,
+      "step": 6084
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.915e-05,
+      "loss": 1.3865,
+      "step": 6085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.914e-05,
+      "loss": 1.0279,
+      "step": 6086
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.913e-05,
+      "loss": 1.259,
+      "step": 6087
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.912e-05,
+      "loss": 1.716,
+      "step": 6088
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.911e-05,
+      "loss": 0.4253,
+      "step": 6089
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.91e-05,
+      "loss": 0.0012,
+      "step": 6090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.909000000000001e-05,
+      "loss": 0.003,
+      "step": 6091
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.908e-05,
+      "loss": 0.4967,
+      "step": 6092
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9070000000000004e-05,
+      "loss": 0.5517,
+      "step": 6093
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.906e-05,
+      "loss": 1.3046,
+      "step": 6094
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.905e-05,
+      "loss": 1.1125,
+      "step": 6095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9040000000000006e-05,
+      "loss": 1.0192,
+      "step": 6096
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.903e-05,
+      "loss": 0.995,
+      "step": 6097
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.902e-05,
+      "loss": 1.0943,
+      "step": 6098
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.901e-05,
+      "loss": 1.2691,
+      "step": 6099
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 1.2947,
+      "step": 6100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8990000000000004e-05,
+      "loss": 0.9839,
+      "step": 6101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.898e-05,
+      "loss": 0.56,
+      "step": 6102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.897e-05,
+      "loss": 0.0793,
+      "step": 6103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.896e-05,
+      "loss": 0.0433,
+      "step": 6104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8950000000000005e-05,
+      "loss": 0.4851,
+      "step": 6105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.894e-05,
+      "loss": 0.728,
+      "step": 6106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.893e-05,
+      "loss": 1.0246,
+      "step": 6107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.892e-05,
+      "loss": 1.1215,
+      "step": 6108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8910000000000005e-05,
+      "loss": 1.461,
+      "step": 6109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8900000000000004e-05,
+      "loss": 1.1552,
+      "step": 6110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.889e-05,
+      "loss": 1.041,
+      "step": 6111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.888e-05,
+      "loss": 1.174,
+      "step": 6112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.887e-05,
+      "loss": 1.9689,
+      "step": 6113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8860000000000004e-05,
+      "loss": 0.4561,
+      "step": 6114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.885e-05,
+      "loss": 0.0431,
+      "step": 6115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.884e-05,
+      "loss": 0.2006,
+      "step": 6116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.883e-05,
+      "loss": 0.3513,
+      "step": 6117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.882e-05,
+      "loss": 0.2162,
+      "step": 6118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.881e-05,
+      "loss": 1.101,
+      "step": 6119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.88e-05,
+      "loss": 1.1823,
+      "step": 6120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8790000000000006e-05,
+      "loss": 1.3393,
+      "step": 6121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.878e-05,
+      "loss": 1.3386,
+      "step": 6122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.877e-05,
+      "loss": 0.8985,
+      "step": 6123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.876e-05,
+      "loss": 1.1988,
+      "step": 6124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.875e-05,
+      "loss": 1.2433,
+      "step": 6125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8740000000000005e-05,
+      "loss": 0.5731,
+      "step": 6126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.873e-05,
+      "loss": 0.2358,
+      "step": 6127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.872e-05,
+      "loss": 0.0312,
+      "step": 6128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.871e-05,
+      "loss": 0.001,
+      "step": 6129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8700000000000006e-05,
+      "loss": 0.2962,
+      "step": 6130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8690000000000004e-05,
+      "loss": 0.8787,
+      "step": 6131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.868e-05,
+      "loss": 0.7818,
+      "step": 6132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.867e-05,
+      "loss": 0.5961,
+      "step": 6133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.866e-05,
+      "loss": 1.0896,
+      "step": 6134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8650000000000004e-05,
+      "loss": 1.2533,
+      "step": 6135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.864e-05,
+      "loss": 1.0878,
+      "step": 6136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.863e-05,
+      "loss": 1.0052,
+      "step": 6137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.862e-05,
+      "loss": 1.4188,
+      "step": 6138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8610000000000005e-05,
+      "loss": 0.7905,
+      "step": 6139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.86e-05,
+      "loss": 0.3499,
+      "step": 6140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.859e-05,
+      "loss": 0.2061,
+      "step": 6141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.858e-05,
+      "loss": 0.0026,
+      "step": 6142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.857e-05,
+      "loss": 0.7312,
+      "step": 6143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8560000000000004e-05,
+      "loss": 0.8354,
+      "step": 6144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.855e-05,
+      "loss": 1.0299,
+      "step": 6145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.854000000000001e-05,
+      "loss": 1.2824,
+      "step": 6146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.853e-05,
+      "loss": 1.4592,
+      "step": 6147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8520000000000004e-05,
+      "loss": 1.3543,
+      "step": 6148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.851e-05,
+      "loss": 0.7232,
+      "step": 6149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.85e-05,
+      "loss": 1.1139,
+      "step": 6150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8490000000000006e-05,
+      "loss": 1.4098,
+      "step": 6151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.848e-05,
+      "loss": 0.4333,
+      "step": 6152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.847e-05,
+      "loss": 0.211,
+      "step": 6153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.846e-05,
+      "loss": 0.0686,
+      "step": 6154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.845e-05,
+      "loss": 0.3534,
+      "step": 6155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8440000000000005e-05,
+      "loss": 0.8385,
+      "step": 6156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8429999999999996e-05,
+      "loss": 1.2251,
+      "step": 6157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.842e-05,
+      "loss": 1.162,
+      "step": 6158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.841e-05,
+      "loss": 1.3914,
+      "step": 6159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 0.8589,
+      "step": 6160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8390000000000003e-05,
+      "loss": 1.2137,
+      "step": 6161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.838e-05,
+      "loss": 1.7803,
+      "step": 6162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.837e-05,
+      "loss": 1.1262,
+      "step": 6163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.836e-05,
+      "loss": 0.4185,
+      "step": 6164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8350000000000004e-05,
+      "loss": 0.1457,
+      "step": 6165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.834e-05,
+      "loss": 0.1516,
+      "step": 6166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.833e-05,
+      "loss": 0.0484,
+      "step": 6167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.832e-05,
+      "loss": 0.1555,
+      "step": 6168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8310000000000004e-05,
+      "loss": 0.5962,
+      "step": 6169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.83e-05,
+      "loss": 1.2111,
+      "step": 6170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.829e-05,
+      "loss": 1.3601,
+      "step": 6171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.828e-05,
+      "loss": 0.9726,
+      "step": 6172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.827e-05,
+      "loss": 1.307,
+      "step": 6173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.826e-05,
+      "loss": 1.0288,
+      "step": 6174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.825e-05,
+      "loss": 1.4915,
+      "step": 6175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8240000000000007e-05,
+      "loss": 1.14,
+      "step": 6176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.823e-05,
+      "loss": 0.2291,
+      "step": 6177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.822e-05,
+      "loss": 0.0792,
+      "step": 6178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.821e-05,
+      "loss": 0.391,
+      "step": 6179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.82e-05,
+      "loss": 1.106,
+      "step": 6180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8190000000000005e-05,
+      "loss": 1.1809,
+      "step": 6181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.818e-05,
+      "loss": 1.1645,
+      "step": 6182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.817e-05,
+      "loss": 1.0655,
+      "step": 6183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.816e-05,
+      "loss": 0.905,
+      "step": 6184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8150000000000006e-05,
+      "loss": 0.933,
+      "step": 6185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8140000000000004e-05,
+      "loss": 0.789,
+      "step": 6186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8129999999999996e-05,
+      "loss": 1.001,
+      "step": 6187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.812e-05,
+      "loss": 0.7459,
+      "step": 6188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.811e-05,
+      "loss": 0.3556,
+      "step": 6189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8100000000000005e-05,
+      "loss": 0.0013,
+      "step": 6190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.809e-05,
+      "loss": 0.1232,
+      "step": 6191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.808e-05,
+      "loss": 0.0845,
+      "step": 6192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.807e-05,
+      "loss": 0.22,
+      "step": 6193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.806e-05,
+      "loss": 1.209,
+      "step": 6194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.805e-05,
+      "loss": 1.3712,
+      "step": 6195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.804e-05,
+      "loss": 1.12,
+      "step": 6196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.803000000000001e-05,
+      "loss": 1.2744,
+      "step": 6197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.802e-05,
+      "loss": 1.0083,
+      "step": 6198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8010000000000004e-05,
+      "loss": 1.6687,
+      "step": 6199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8e-05,
+      "loss": 1.0539,
+      "step": 6200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.799e-05,
+      "loss": 1.1538,
+      "step": 6201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7980000000000006e-05,
+      "loss": 0.6887,
+      "step": 6202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.797e-05,
+      "loss": 0.096,
+      "step": 6203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.796e-05,
+      "loss": 0.1001,
+      "step": 6204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.795e-05,
+      "loss": 0.2641,
+      "step": 6205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7940000000000006e-05,
+      "loss": 0.6986,
+      "step": 6206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7930000000000004e-05,
+      "loss": 1.0524,
+      "step": 6207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.792e-05,
+      "loss": 0.9214,
+      "step": 6208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.791e-05,
+      "loss": 1.3189,
+      "step": 6209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.79e-05,
+      "loss": 1.1783,
+      "step": 6210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7890000000000005e-05,
+      "loss": 1.5066,
+      "step": 6211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.788e-05,
+      "loss": 1.399,
+      "step": 6212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.787e-05,
+      "loss": 0.6235,
+      "step": 6213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.786e-05,
+      "loss": 1.1184,
+      "step": 6214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7850000000000005e-05,
+      "loss": 0.2108,
+      "step": 6215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7840000000000004e-05,
+      "loss": 0.0103,
+      "step": 6216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.783e-05,
+      "loss": 0.1868,
+      "step": 6217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.782e-05,
+      "loss": 0.4145,
+      "step": 6218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.781e-05,
+      "loss": 1.0614,
+      "step": 6219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 1.3044,
+      "step": 6220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.779e-05,
+      "loss": 1.1552,
+      "step": 6221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.778000000000001e-05,
+      "loss": 1.1841,
+      "step": 6222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.777e-05,
+      "loss": 0.9735,
+      "step": 6223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.776e-05,
+      "loss": 1.3441,
+      "step": 6224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.775e-05,
+      "loss": 1.6606,
+      "step": 6225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.774e-05,
+      "loss": 0.9957,
+      "step": 6226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7730000000000006e-05,
+      "loss": 0.3991,
+      "step": 6227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.772e-05,
+      "loss": 0.0972,
+      "step": 6228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.771e-05,
+      "loss": 0.6051,
+      "step": 6229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.77e-05,
+      "loss": 0.3776,
+      "step": 6230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.769e-05,
+      "loss": 1.2473,
+      "step": 6231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7680000000000005e-05,
+      "loss": 1.1983,
+      "step": 6232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.767e-05,
+      "loss": 1.0873,
+      "step": 6233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.766e-05,
+      "loss": 1.3342,
+      "step": 6234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.765e-05,
+      "loss": 1.1425,
+      "step": 6235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7640000000000006e-05,
+      "loss": 0.8323,
+      "step": 6236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7630000000000004e-05,
+      "loss": 0.9404,
+      "step": 6237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.762e-05,
+      "loss": 1.056,
+      "step": 6238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.761e-05,
+      "loss": 0.8717,
+      "step": 6239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.76e-05,
+      "loss": 0.1094,
+      "step": 6240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7590000000000004e-05,
+      "loss": 0.1088,
+      "step": 6241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.758e-05,
+      "loss": 0.0027,
+      "step": 6242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.757e-05,
+      "loss": 0.6994,
+      "step": 6243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.756e-05,
+      "loss": 1.2185,
+      "step": 6244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7550000000000005e-05,
+      "loss": 0.7088,
+      "step": 6245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.754e-05,
+      "loss": 0.9143,
+      "step": 6246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.753e-05,
+      "loss": 0.7815,
+      "step": 6247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.752e-05,
+      "loss": 1.2769,
+      "step": 6248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.751e-05,
+      "loss": 0.9923,
+      "step": 6249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.0523,
+      "step": 6250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.749e-05,
+      "loss": 1.2624,
+      "step": 6251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.748000000000001e-05,
+      "loss": 0.3549,
+      "step": 6252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.747e-05,
+      "loss": 0.2247,
+      "step": 6253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7460000000000004e-05,
+      "loss": 0.0336,
+      "step": 6254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.745e-05,
+      "loss": 0.1106,
+      "step": 6255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.744e-05,
+      "loss": 0.8706,
+      "step": 6256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7430000000000006e-05,
+      "loss": 1.0012,
+      "step": 6257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.742e-05,
+      "loss": 1.055,
+      "step": 6258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.741e-05,
+      "loss": 1.211,
+      "step": 6259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.74e-05,
+      "loss": 1.5273,
+      "step": 6260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.739e-05,
+      "loss": 0.936,
+      "step": 6261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7380000000000005e-05,
+      "loss": 1.2609,
+      "step": 6262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7369999999999996e-05,
+      "loss": 1.0707,
+      "step": 6263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.736e-05,
+      "loss": 0.3852,
+      "step": 6264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.735e-05,
+      "loss": 0.053,
+      "step": 6265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7340000000000005e-05,
+      "loss": 0.3587,
+      "step": 6266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7330000000000003e-05,
+      "loss": 0.068,
+      "step": 6267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.732e-05,
+      "loss": 0.9335,
+      "step": 6268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.731e-05,
+      "loss": 1.009,
+      "step": 6269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.73e-05,
+      "loss": 1.0565,
+      "step": 6270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7290000000000004e-05,
+      "loss": 0.872,
+      "step": 6271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.728e-05,
+      "loss": 1.0348,
+      "step": 6272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.727e-05,
+      "loss": 0.6081,
+      "step": 6273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.726e-05,
+      "loss": 1.3073,
+      "step": 6274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7250000000000004e-05,
+      "loss": 1.0612,
+      "step": 6275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.724e-05,
+      "loss": 1.22,
+      "step": 6276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.723e-05,
+      "loss": 0.5728,
+      "step": 6277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.722e-05,
+      "loss": 0.0742,
+      "step": 6278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.721e-05,
+      "loss": 0.0592,
+      "step": 6279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.72e-05,
+      "loss": 0.1064,
+      "step": 6280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.719e-05,
+      "loss": 0.6286,
+      "step": 6281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7180000000000007e-05,
+      "loss": 0.9699,
+      "step": 6282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.717e-05,
+      "loss": 0.9908,
+      "step": 6283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.716e-05,
+      "loss": 1.2715,
+      "step": 6284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.715e-05,
+      "loss": 0.8497,
+      "step": 6285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.714e-05,
+      "loss": 1.1236,
+      "step": 6286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7130000000000005e-05,
+      "loss": 0.8326,
+      "step": 6287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.712e-05,
+      "loss": 1.3502,
+      "step": 6288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.711e-05,
+      "loss": 0.574,
+      "step": 6289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.71e-05,
+      "loss": 0.176,
+      "step": 6290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7090000000000006e-05,
+      "loss": 0.1234,
+      "step": 6291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7080000000000004e-05,
+      "loss": 0.0475,
+      "step": 6292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.707e-05,
+      "loss": 0.743,
+      "step": 6293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.706e-05,
+      "loss": 1.3417,
+      "step": 6294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.705e-05,
+      "loss": 1.3712,
+      "step": 6295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7040000000000005e-05,
+      "loss": 1.1485,
+      "step": 6296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.703e-05,
+      "loss": 1.1798,
+      "step": 6297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.702e-05,
+      "loss": 1.1858,
+      "step": 6298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.701e-05,
+      "loss": 1.1744,
+      "step": 6299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7e-05,
+      "loss": 0.6071,
+      "step": 6300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.699e-05,
+      "loss": 1.3575,
+      "step": 6301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.698e-05,
+      "loss": 0.1154,
+      "step": 6302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.697e-05,
+      "loss": 0.3268,
+      "step": 6303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.696e-05,
+      "loss": 0.1996,
+      "step": 6304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6950000000000004e-05,
+      "loss": 0.3882,
+      "step": 6305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.694e-05,
+      "loss": 0.635,
+      "step": 6306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.693e-05,
+      "loss": 1.2577,
+      "step": 6307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.692e-05,
+      "loss": 1.3101,
+      "step": 6308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.691e-05,
+      "loss": 1.2396,
+      "step": 6309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.69e-05,
+      "loss": 1.3602,
+      "step": 6310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.689e-05,
+      "loss": 1.2722,
+      "step": 6311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6880000000000006e-05,
+      "loss": 1.07,
+      "step": 6312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6870000000000004e-05,
+      "loss": 1.2375,
+      "step": 6313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.686e-05,
+      "loss": 0.5192,
+      "step": 6314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.685e-05,
+      "loss": 0.0708,
+      "step": 6315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.684e-05,
+      "loss": 0.1197,
+      "step": 6316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6830000000000005e-05,
+      "loss": 0.3704,
+      "step": 6317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.682e-05,
+      "loss": 0.9543,
+      "step": 6318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.681e-05,
+      "loss": 1.2365,
+      "step": 6319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.68e-05,
+      "loss": 1.2765,
+      "step": 6320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6790000000000005e-05,
+      "loss": 0.905,
+      "step": 6321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6780000000000004e-05,
+      "loss": 1.554,
+      "step": 6322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.677e-05,
+      "loss": 1.0554,
+      "step": 6323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.676e-05,
+      "loss": 1.0504,
+      "step": 6324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.675e-05,
+      "loss": 1.0801,
+      "step": 6325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6740000000000004e-05,
+      "loss": 1.0643,
+      "step": 6326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.673e-05,
+      "loss": 0.5577,
+      "step": 6327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.672000000000001e-05,
+      "loss": 0.0701,
+      "step": 6328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.671e-05,
+      "loss": 0.1994,
+      "step": 6329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6700000000000004e-05,
+      "loss": 0.2786,
+      "step": 6330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.669e-05,
+      "loss": 0.1504,
+      "step": 6331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.668e-05,
+      "loss": 0.5495,
+      "step": 6332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6670000000000006e-05,
+      "loss": 1.5205,
+      "step": 6333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.666e-05,
+      "loss": 1.1674,
+      "step": 6334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.665e-05,
+      "loss": 1.1222,
+      "step": 6335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.664e-05,
+      "loss": 0.9604,
+      "step": 6336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.663e-05,
+      "loss": 1.3077,
+      "step": 6337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6620000000000005e-05,
+      "loss": 1.0446,
+      "step": 6338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.661e-05,
+      "loss": 0.907,
+      "step": 6339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.66e-05,
+      "loss": 0.1331,
+      "step": 6340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.659e-05,
+      "loss": 0.1937,
+      "step": 6341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6580000000000006e-05,
+      "loss": 0.0155,
+      "step": 6342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6570000000000004e-05,
+      "loss": 0.7555,
+      "step": 6343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.656e-05,
+      "loss": 1.7842,
+      "step": 6344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.655e-05,
+      "loss": 1.2792,
+      "step": 6345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.654e-05,
+      "loss": 0.7883,
+      "step": 6346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6530000000000004e-05,
+      "loss": 1.3295,
+      "step": 6347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.652e-05,
+      "loss": 1.1867,
+      "step": 6348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.651e-05,
+      "loss": 1.3867,
+      "step": 6349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.65e-05,
+      "loss": 1.8375,
+      "step": 6350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6490000000000005e-05,
+      "loss": 1.4829,
+      "step": 6351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.648e-05,
+      "loss": 0.4905,
+      "step": 6352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.647e-05,
+      "loss": 0.0683,
+      "step": 6353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.646e-05,
+      "loss": 0.1805,
+      "step": 6354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.645e-05,
+      "loss": 0.3969,
+      "step": 6355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6440000000000003e-05,
+      "loss": 0.977,
+      "step": 6356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.643e-05,
+      "loss": 0.9676,
+      "step": 6357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.642000000000001e-05,
+      "loss": 1.2745,
+      "step": 6358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.641e-05,
+      "loss": 0.9244,
+      "step": 6359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 1.1,
+      "step": 6360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.639e-05,
+      "loss": 0.9628,
+      "step": 6361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.638e-05,
+      "loss": 1.4993,
+      "step": 6362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6370000000000006e-05,
+      "loss": 1.3132,
+      "step": 6363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.636e-05,
+      "loss": 0.6652,
+      "step": 6364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.635e-05,
+      "loss": 0.1072,
+      "step": 6365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.634e-05,
+      "loss": 0.5308,
+      "step": 6366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6330000000000006e-05,
+      "loss": 0.1655,
+      "step": 6367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6320000000000005e-05,
+      "loss": 1.0679,
+      "step": 6368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6309999999999996e-05,
+      "loss": 1.0947,
+      "step": 6369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.63e-05,
+      "loss": 1.1994,
+      "step": 6370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.629e-05,
+      "loss": 0.9101,
+      "step": 6371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6280000000000005e-05,
+      "loss": 1.2183,
+      "step": 6372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6270000000000003e-05,
+      "loss": 1.2827,
+      "step": 6373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.626e-05,
+      "loss": 0.7452,
+      "step": 6374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.625e-05,
+      "loss": 1.3086,
+      "step": 6375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.624e-05,
+      "loss": 1.0616,
+      "step": 6376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6230000000000004e-05,
+      "loss": 0.2094,
+      "step": 6377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.622e-05,
+      "loss": 0.022,
+      "step": 6378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.621e-05,
+      "loss": 0.17,
+      "step": 6379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.62e-05,
+      "loss": 0.676,
+      "step": 6380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6190000000000004e-05,
+      "loss": 1.1794,
+      "step": 6381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.618e-05,
+      "loss": 1.2156,
+      "step": 6382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.617e-05,
+      "loss": 1.1632,
+      "step": 6383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.616e-05,
+      "loss": 1.2371,
+      "step": 6384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.615e-05,
+      "loss": 1.067,
+      "step": 6385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.614e-05,
+      "loss": 0.5549,
+      "step": 6386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.613e-05,
+      "loss": 0.7378,
+      "step": 6387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6120000000000007e-05,
+      "loss": 1.1958,
+      "step": 6388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.611e-05,
+      "loss": 0.5274,
+      "step": 6389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.61e-05,
+      "loss": 0.1976,
+      "step": 6390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.609e-05,
+      "loss": 0.1766,
+      "step": 6391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.608e-05,
+      "loss": 0.1427,
+      "step": 6392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6070000000000005e-05,
+      "loss": 1.511,
+      "step": 6393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.606e-05,
+      "loss": 1.0895,
+      "step": 6394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.605e-05,
+      "loss": 1.0187,
+      "step": 6395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.604e-05,
+      "loss": 0.6835,
+      "step": 6396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6030000000000006e-05,
+      "loss": 1.5032,
+      "step": 6397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6020000000000004e-05,
+      "loss": 1.0902,
+      "step": 6398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.601e-05,
+      "loss": 0.868,
+      "step": 6399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6e-05,
+      "loss": 0.9837,
+      "step": 6400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.599e-05,
+      "loss": 1.191,
+      "step": 6401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5980000000000004e-05,
+      "loss": 0.575,
+      "step": 6402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.597e-05,
+      "loss": 0.3185,
+      "step": 6403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.596e-05,
+      "loss": 0.3516,
+      "step": 6404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.595e-05,
+      "loss": 0.457,
+      "step": 6405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.594e-05,
+      "loss": 0.8769,
+      "step": 6406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.593e-05,
+      "loss": 1.1528,
+      "step": 6407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.592e-05,
+      "loss": 1.2928,
+      "step": 6408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.591e-05,
+      "loss": 1.0081,
+      "step": 6409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.59e-05,
+      "loss": 1.0259,
+      "step": 6410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5890000000000004e-05,
+      "loss": 0.8267,
+      "step": 6411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.588e-05,
+      "loss": 1.3071,
+      "step": 6412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.587e-05,
+      "loss": 1.2429,
+      "step": 6413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.586e-05,
+      "loss": 0.6975,
+      "step": 6414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.585e-05,
+      "loss": 0.3023,
+      "step": 6415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.584e-05,
+      "loss": 0.2489,
+      "step": 6416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.583e-05,
+      "loss": 0.0144,
+      "step": 6417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5820000000000006e-05,
+      "loss": 0.4302,
+      "step": 6418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.581e-05,
+      "loss": 0.8606,
+      "step": 6419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.58e-05,
+      "loss": 0.9246,
+      "step": 6420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.579e-05,
+      "loss": 1.2033,
+      "step": 6421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.578e-05,
+      "loss": 1.4471,
+      "step": 6422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5770000000000005e-05,
+      "loss": 1.294,
+      "step": 6423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5759999999999996e-05,
+      "loss": 1.0949,
+      "step": 6424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.575e-05,
+      "loss": 0.8776,
+      "step": 6425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.574e-05,
+      "loss": 1.1291,
+      "step": 6426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5730000000000005e-05,
+      "loss": 0.1379,
+      "step": 6427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5720000000000004e-05,
+      "loss": 0.0185,
+      "step": 6428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.571e-05,
+      "loss": 0.0455,
+      "step": 6429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.57e-05,
+      "loss": 1.1282,
+      "step": 6430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.569e-05,
+      "loss": 1.0316,
+      "step": 6431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5680000000000004e-05,
+      "loss": 1.1287,
+      "step": 6432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.567e-05,
+      "loss": 1.3587,
+      "step": 6433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.566e-05,
+      "loss": 1.0446,
+      "step": 6434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.565e-05,
+      "loss": 0.7846,
+      "step": 6435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5640000000000004e-05,
+      "loss": 1.4302,
+      "step": 6436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.563e-05,
+      "loss": 1.2394,
+      "step": 6437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.562e-05,
+      "loss": 1.3822,
+      "step": 6438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5610000000000006e-05,
+      "loss": 0.6435,
+      "step": 6439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.56e-05,
+      "loss": 0.1726,
+      "step": 6440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.559e-05,
+      "loss": 0.1527,
+      "step": 6441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.558e-05,
+      "loss": 0.0513,
+      "step": 6442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.557e-05,
+      "loss": 0.5396,
+      "step": 6443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5560000000000005e-05,
+      "loss": 0.8598,
+      "step": 6444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.555e-05,
+      "loss": 1.0503,
+      "step": 6445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.554e-05,
+      "loss": 1.0551,
+      "step": 6446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.553e-05,
+      "loss": 0.9437,
+      "step": 6447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5520000000000006e-05,
+      "loss": 1.195,
+      "step": 6448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5510000000000004e-05,
+      "loss": 1.1359,
+      "step": 6449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.55e-05,
+      "loss": 1.4307,
+      "step": 6450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.549e-05,
+      "loss": 1.277,
+      "step": 6451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.548e-05,
+      "loss": 0.4843,
+      "step": 6452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5470000000000004e-05,
+      "loss": 0.0694,
+      "step": 6453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.546e-05,
+      "loss": 0.1541,
+      "step": 6454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.545e-05,
+      "loss": 0.2736,
+      "step": 6455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.544e-05,
+      "loss": 0.4804,
+      "step": 6456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5430000000000005e-05,
+      "loss": 1.1528,
+      "step": 6457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.542e-05,
+      "loss": 1.4858,
+      "step": 6458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.541e-05,
+      "loss": 1.0726,
+      "step": 6459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.54e-05,
+      "loss": 0.9463,
+      "step": 6460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.539e-05,
+      "loss": 1.1468,
+      "step": 6461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5380000000000003e-05,
+      "loss": 1.3967,
+      "step": 6462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.537e-05,
+      "loss": 1.6107,
+      "step": 6463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.536000000000001e-05,
+      "loss": 0.8574,
+      "step": 6464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.535e-05,
+      "loss": 0.1503,
+      "step": 6465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5340000000000004e-05,
+      "loss": 0.148,
+      "step": 6466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.533e-05,
+      "loss": 0.1811,
+      "step": 6467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.532e-05,
+      "loss": 0.0833,
+      "step": 6468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5310000000000006e-05,
+      "loss": 1.0193,
+      "step": 6469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.53e-05,
+      "loss": 1.4071,
+      "step": 6470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.529e-05,
+      "loss": 1.0586,
+      "step": 6471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.528e-05,
+      "loss": 1.2396,
+      "step": 6472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5270000000000006e-05,
+      "loss": 1.0018,
+      "step": 6473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5260000000000005e-05,
+      "loss": 1.4568,
+      "step": 6474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.525e-05,
+      "loss": 1.324,
+      "step": 6475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.524e-05,
+      "loss": 0.7887,
+      "step": 6476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.523e-05,
+      "loss": 0.0513,
+      "step": 6477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5220000000000005e-05,
+      "loss": 0.1921,
+      "step": 6478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5210000000000003e-05,
+      "loss": 0.3127,
+      "step": 6479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.52e-05,
+      "loss": 0.3163,
+      "step": 6480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.519e-05,
+      "loss": 0.7757,
+      "step": 6481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.518e-05,
+      "loss": 1.1163,
+      "step": 6482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5170000000000004e-05,
+      "loss": 1.3409,
+      "step": 6483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.516e-05,
+      "loss": 1.3018,
+      "step": 6484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.515e-05,
+      "loss": 1.0243,
+      "step": 6485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.514e-05,
+      "loss": 1.0274,
+      "step": 6486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5130000000000004e-05,
+      "loss": 0.9484,
+      "step": 6487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.512e-05,
+      "loss": 1.106,
+      "step": 6488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.511e-05,
+      "loss": 1.2596,
+      "step": 6489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.51e-05,
+      "loss": 0.2944,
+      "step": 6490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.509e-05,
+      "loss": 0.1182,
+      "step": 6491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.508e-05,
+      "loss": 0.2247,
+      "step": 6492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.507e-05,
+      "loss": 0.3142,
+      "step": 6493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5060000000000007e-05,
+      "loss": 1.2905,
+      "step": 6494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.505e-05,
+      "loss": 1.1608,
+      "step": 6495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.504e-05,
+      "loss": 1.251,
+      "step": 6496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.503e-05,
+      "loss": 1.1698,
+      "step": 6497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.502e-05,
+      "loss": 1.1183,
+      "step": 6498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5010000000000005e-05,
+      "loss": 0.9427,
+      "step": 6499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4919,
+      "step": 6500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.499e-05,
+      "loss": 1.4466,
+      "step": 6501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.498e-05,
+      "loss": 0.6134,
+      "step": 6502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4970000000000006e-05,
+      "loss": 0.1849,
+      "step": 6503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4960000000000004e-05,
+      "loss": 0.0083,
+      "step": 6504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.495e-05,
+      "loss": 0.6853,
+      "step": 6505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.494e-05,
+      "loss": 1.0102,
+      "step": 6506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.493e-05,
+      "loss": 1.0787,
+      "step": 6507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4920000000000004e-05,
+      "loss": 0.949,
+      "step": 6508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.491e-05,
+      "loss": 0.961,
+      "step": 6509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.49e-05,
+      "loss": 1.1386,
+      "step": 6510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.489e-05,
+      "loss": 1.3075,
+      "step": 6511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4880000000000005e-05,
+      "loss": 1.0348,
+      "step": 6512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.487e-05,
+      "loss": 1.0742,
+      "step": 6513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.486e-05,
+      "loss": 0.6875,
+      "step": 6514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.485e-05,
+      "loss": 0.3249,
+      "step": 6515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.484e-05,
+      "loss": 0.2857,
+      "step": 6516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4830000000000004e-05,
+      "loss": 0.6018,
+      "step": 6517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.482e-05,
+      "loss": 0.707,
+      "step": 6518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.481e-05,
+      "loss": 0.8231,
+      "step": 6519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.48e-05,
+      "loss": 0.8251,
+      "step": 6520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.479e-05,
+      "loss": 1.4992,
+      "step": 6521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.478e-05,
+      "loss": 1.1495,
+      "step": 6522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.477e-05,
+      "loss": 1.0477,
+      "step": 6523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4760000000000006e-05,
+      "loss": 1.2041,
+      "step": 6524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.475e-05,
+      "loss": 1.5061,
+      "step": 6525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.474e-05,
+      "loss": 1.3642,
+      "step": 6526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.473e-05,
+      "loss": 0.2524,
+      "step": 6527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.472e-05,
+      "loss": 0.012,
+      "step": 6528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4710000000000005e-05,
+      "loss": 0.0051,
+      "step": 6529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4699999999999996e-05,
+      "loss": 0.2675,
+      "step": 6530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.469e-05,
+      "loss": 0.7083,
+      "step": 6531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.468e-05,
+      "loss": 1.1027,
+      "step": 6532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4670000000000005e-05,
+      "loss": 1.278,
+      "step": 6533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4660000000000004e-05,
+      "loss": 1.0983,
+      "step": 6534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.465e-05,
+      "loss": 1.2676,
+      "step": 6535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.464e-05,
+      "loss": 1.2836,
+      "step": 6536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.463e-05,
+      "loss": 1.1369,
+      "step": 6537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4620000000000004e-05,
+      "loss": 1.3921,
+      "step": 6538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.461e-05,
+      "loss": 0.6643,
+      "step": 6539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.46e-05,
+      "loss": 0.1792,
+      "step": 6540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.459e-05,
+      "loss": 0.2081,
+      "step": 6541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4580000000000004e-05,
+      "loss": 0.2756,
+      "step": 6542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.457e-05,
+      "loss": 0.6141,
+      "step": 6543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.456e-05,
+      "loss": 1.0794,
+      "step": 6544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.455e-05,
+      "loss": 1.0961,
+      "step": 6545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.454e-05,
+      "loss": 1.1892,
+      "step": 6546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.453e-05,
+      "loss": 1.6442,
+      "step": 6547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.452e-05,
+      "loss": 1.0998,
+      "step": 6548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.451000000000001e-05,
+      "loss": 0.7912,
+      "step": 6549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.45e-05,
+      "loss": 1.2526,
+      "step": 6550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.449e-05,
+      "loss": 0.6971,
+      "step": 6551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.448e-05,
+      "loss": 0.2318,
+      "step": 6552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.447e-05,
+      "loss": 0.0082,
+      "step": 6553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4460000000000005e-05,
+      "loss": 0.1507,
+      "step": 6554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.445e-05,
+      "loss": 0.3743,
+      "step": 6555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.444e-05,
+      "loss": 0.4413,
+      "step": 6556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.443e-05,
+      "loss": 1.0624,
+      "step": 6557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.442e-05,
+      "loss": 1.1205,
+      "step": 6558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4410000000000004e-05,
+      "loss": 1.2863,
+      "step": 6559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 1.4859,
+      "step": 6560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.439e-05,
+      "loss": 1.6601,
+      "step": 6561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.438e-05,
+      "loss": 1.0565,
+      "step": 6562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4370000000000005e-05,
+      "loss": 1.2664,
+      "step": 6563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.436e-05,
+      "loss": 0.8016,
+      "step": 6564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.435e-05,
+      "loss": 0.1295,
+      "step": 6565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.434e-05,
+      "loss": 0.1718,
+      "step": 6566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.433e-05,
+      "loss": 0.4687,
+      "step": 6567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4320000000000003e-05,
+      "loss": 1.2298,
+      "step": 6568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.431e-05,
+      "loss": 1.4414,
+      "step": 6569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.430000000000001e-05,
+      "loss": 1.3191,
+      "step": 6570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.429e-05,
+      "loss": 1.9175,
+      "step": 6571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4280000000000004e-05,
+      "loss": 0.9915,
+      "step": 6572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.427e-05,
+      "loss": 1.2405,
+      "step": 6573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.426e-05,
+      "loss": 1.2149,
+      "step": 6574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4250000000000006e-05,
+      "loss": 1.0552,
+      "step": 6575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.424e-05,
+      "loss": 1.0075,
+      "step": 6576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.423e-05,
+      "loss": 0.3572,
+      "step": 6577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.422e-05,
+      "loss": 0.2192,
+      "step": 6578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4210000000000006e-05,
+      "loss": 0.2012,
+      "step": 6579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 0.0928,
+      "step": 6580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.419e-05,
+      "loss": 0.9512,
+      "step": 6581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.418e-05,
+      "loss": 1.4542,
+      "step": 6582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.417e-05,
+      "loss": 1.0848,
+      "step": 6583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4160000000000005e-05,
+      "loss": 1.4333,
+      "step": 6584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.415e-05,
+      "loss": 1.6313,
+      "step": 6585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.414e-05,
+      "loss": 0.98,
+      "step": 6586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.413e-05,
+      "loss": 1.2501,
+      "step": 6587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.412e-05,
+      "loss": 1.5477,
+      "step": 6588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4110000000000004e-05,
+      "loss": 0.1124,
+      "step": 6589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.41e-05,
+      "loss": 0.2069,
+      "step": 6590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.409e-05,
+      "loss": 0.1927,
+      "step": 6591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.408e-05,
+      "loss": 0.1786,
+      "step": 6592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4070000000000004e-05,
+      "loss": 1.1722,
+      "step": 6593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.406e-05,
+      "loss": 1.0162,
+      "step": 6594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.405e-05,
+      "loss": 0.8601,
+      "step": 6595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.404e-05,
+      "loss": 1.2207,
+      "step": 6596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.403e-05,
+      "loss": 1.4054,
+      "step": 6597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.402e-05,
+      "loss": 1.2446,
+      "step": 6598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.401e-05,
+      "loss": 1.3013,
+      "step": 6599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.4674,
+      "step": 6600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.399e-05,
+      "loss": 1.0784,
+      "step": 6601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.398e-05,
+      "loss": 0.2902,
+      "step": 6602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.397e-05,
+      "loss": 0.0234,
+      "step": 6603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.396e-05,
+      "loss": 0.1782,
+      "step": 6604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3950000000000005e-05,
+      "loss": 0.0279,
+      "step": 6605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.394e-05,
+      "loss": 0.7705,
+      "step": 6606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.393e-05,
+      "loss": 1.1935,
+      "step": 6607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.392e-05,
+      "loss": 1.2859,
+      "step": 6608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3910000000000006e-05,
+      "loss": 1.2776,
+      "step": 6609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3900000000000004e-05,
+      "loss": 1.2818,
+      "step": 6610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.389e-05,
+      "loss": 1.1116,
+      "step": 6611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.388e-05,
+      "loss": 1.3507,
+      "step": 6612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.387e-05,
+      "loss": 1.3618,
+      "step": 6613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3860000000000004e-05,
+      "loss": 0.9993,
+      "step": 6614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.385e-05,
+      "loss": 0.1642,
+      "step": 6615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.384e-05,
+      "loss": 0.1169,
+      "step": 6616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.383e-05,
+      "loss": 0.5182,
+      "step": 6617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3820000000000005e-05,
+      "loss": 0.8633,
+      "step": 6618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.381e-05,
+      "loss": 0.8429,
+      "step": 6619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.38e-05,
+      "loss": 1.5579,
+      "step": 6620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.379e-05,
+      "loss": 1.0432,
+      "step": 6621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.378e-05,
+      "loss": 1.06,
+      "step": 6622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3770000000000004e-05,
+      "loss": 1.1086,
+      "step": 6623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.376e-05,
+      "loss": 1.3031,
+      "step": 6624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.375000000000001e-05,
+      "loss": 1.3286,
+      "step": 6625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.374e-05,
+      "loss": 1.0235,
+      "step": 6626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.373e-05,
+      "loss": 0.2308,
+      "step": 6627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.372e-05,
+      "loss": 0.1164,
+      "step": 6628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.371e-05,
+      "loss": 0.0628,
+      "step": 6629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3700000000000006e-05,
+      "loss": 0.6677,
+      "step": 6630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.369e-05,
+      "loss": 1.1741,
+      "step": 6631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.368e-05,
+      "loss": 1.342,
+      "step": 6632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.367e-05,
+      "loss": 0.7563,
+      "step": 6633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.366e-05,
+      "loss": 0.9381,
+      "step": 6634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3650000000000005e-05,
+      "loss": 1.2371,
+      "step": 6635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3639999999999996e-05,
+      "loss": 0.9602,
+      "step": 6636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.363e-05,
+      "loss": 1.2671,
+      "step": 6637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.362e-05,
+      "loss": 2.307,
+      "step": 6638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3610000000000005e-05,
+      "loss": 0.2709,
+      "step": 6639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 0.1409,
+      "step": 6640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.359e-05,
+      "loss": 0.0938,
+      "step": 6641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.358e-05,
+      "loss": 0.6609,
+      "step": 6642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.357e-05,
+      "loss": 0.9281,
+      "step": 6643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3560000000000004e-05,
+      "loss": 1.0536,
+      "step": 6644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.355e-05,
+      "loss": 0.8775,
+      "step": 6645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.354e-05,
+      "loss": 1.1392,
+      "step": 6646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.353e-05,
+      "loss": 1.4458,
+      "step": 6647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3520000000000004e-05,
+      "loss": 1.1276,
+      "step": 6648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.351e-05,
+      "loss": 1.1198,
+      "step": 6649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.35e-05,
+      "loss": 1.6394,
+      "step": 6650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.349e-05,
+      "loss": 1.3492,
+      "step": 6651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.348e-05,
+      "loss": 0.1442,
+      "step": 6652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.347e-05,
+      "loss": 0.0939,
+      "step": 6653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.346e-05,
+      "loss": 0.0732,
+      "step": 6654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.345000000000001e-05,
+      "loss": 0.2815,
+      "step": 6655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.344e-05,
+      "loss": 0.6773,
+      "step": 6656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3430000000000003e-05,
+      "loss": 1.4797,
+      "step": 6657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.342e-05,
+      "loss": 1.2818,
+      "step": 6658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.341e-05,
+      "loss": 1.017,
+      "step": 6659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 1.4058,
+      "step": 6660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.339e-05,
+      "loss": 1.0271,
+      "step": 6661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.338e-05,
+      "loss": 1.6085,
+      "step": 6662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.337e-05,
+      "loss": 1.6565,
+      "step": 6663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.336e-05,
+      "loss": 0.3995,
+      "step": 6664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3350000000000004e-05,
+      "loss": 0.0382,
+      "step": 6665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3339999999999996e-05,
+      "loss": 0.1003,
+      "step": 6666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.333e-05,
+      "loss": 0.1478,
+      "step": 6667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.332e-05,
+      "loss": 0.3192,
+      "step": 6668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3310000000000005e-05,
+      "loss": 0.9428,
+      "step": 6669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.33e-05,
+      "loss": 1.2884,
+      "step": 6670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.329e-05,
+      "loss": 1.3544,
+      "step": 6671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.328e-05,
+      "loss": 1.2468,
+      "step": 6672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.327e-05,
+      "loss": 1.1579,
+      "step": 6673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3260000000000003e-05,
+      "loss": 1.0183,
+      "step": 6674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.325e-05,
+      "loss": 0.4639,
+      "step": 6675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.324e-05,
+      "loss": 0.7892,
+      "step": 6676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.323e-05,
+      "loss": 0.1825,
+      "step": 6677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3220000000000004e-05,
+      "loss": 0.0596,
+      "step": 6678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.321e-05,
+      "loss": 0.0381,
+      "step": 6679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.32e-05,
+      "loss": 0.5899,
+      "step": 6680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.319e-05,
+      "loss": 0.6351,
+      "step": 6681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.318e-05,
+      "loss": 0.9785,
+      "step": 6682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.317e-05,
+      "loss": 1.3484,
+      "step": 6683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.316e-05,
+      "loss": 0.8998,
+      "step": 6684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3150000000000006e-05,
+      "loss": 0.9527,
+      "step": 6685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.314e-05,
+      "loss": 1.381,
+      "step": 6686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.313e-05,
+      "loss": 1.3687,
+      "step": 6687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.312e-05,
+      "loss": 1.0627,
+      "step": 6688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.311e-05,
+      "loss": 0.7071,
+      "step": 6689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3100000000000005e-05,
+      "loss": 0.3087,
+      "step": 6690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.309e-05,
+      "loss": 0.0241,
+      "step": 6691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.308e-05,
+      "loss": 0.0481,
+      "step": 6692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.307e-05,
+      "loss": 0.1917,
+      "step": 6693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3060000000000005e-05,
+      "loss": 1.0551,
+      "step": 6694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3050000000000004e-05,
+      "loss": 0.7928,
+      "step": 6695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.304e-05,
+      "loss": 1.1755,
+      "step": 6696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.303e-05,
+      "loss": 1.3316,
+      "step": 6697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.302e-05,
+      "loss": 1.108,
+      "step": 6698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3010000000000004e-05,
+      "loss": 1.225,
+      "step": 6699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3e-05,
+      "loss": 1.008,
+      "step": 6700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.299e-05,
+      "loss": 1.4173,
+      "step": 6701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.298e-05,
+      "loss": 0.3027,
+      "step": 6702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.297e-05,
+      "loss": 0.2448,
+      "step": 6703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.296e-05,
+      "loss": 0.2438,
+      "step": 6704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.295e-05,
+      "loss": 0.684,
+      "step": 6705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2940000000000006e-05,
+      "loss": 1.1073,
+      "step": 6706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.293e-05,
+      "loss": 1.2785,
+      "step": 6707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.292e-05,
+      "loss": 0.8985,
+      "step": 6708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.291e-05,
+      "loss": 1.2482,
+      "step": 6709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.29e-05,
+      "loss": 1.3861,
+      "step": 6710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2890000000000005e-05,
+      "loss": 1.218,
+      "step": 6711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.288e-05,
+      "loss": 1.1512,
+      "step": 6712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.287e-05,
+      "loss": 1.3416,
+      "step": 6713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.286e-05,
+      "loss": 1.0149,
+      "step": 6714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2850000000000006e-05,
+      "loss": 0.387,
+      "step": 6715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2840000000000004e-05,
+      "loss": 0.0826,
+      "step": 6716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.283e-05,
+      "loss": 0.3355,
+      "step": 6717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.282e-05,
+      "loss": 0.4608,
+      "step": 6718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.281e-05,
+      "loss": 0.7091,
+      "step": 6719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 0.8785,
+      "step": 6720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.279e-05,
+      "loss": 1.229,
+      "step": 6721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.278e-05,
+      "loss": 1.1472,
+      "step": 6722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.277e-05,
+      "loss": 0.7198,
+      "step": 6723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2760000000000005e-05,
+      "loss": 0.7243,
+      "step": 6724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.275e-05,
+      "loss": 1.3522,
+      "step": 6725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.274e-05,
+      "loss": 1.4975,
+      "step": 6726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.273e-05,
+      "loss": 0.3712,
+      "step": 6727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.272e-05,
+      "loss": 0.023,
+      "step": 6728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2710000000000004e-05,
+      "loss": 0.5813,
+      "step": 6729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.27e-05,
+      "loss": 1.2069,
+      "step": 6730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.269000000000001e-05,
+      "loss": 0.978,
+      "step": 6731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.268e-05,
+      "loss": 1.3722,
+      "step": 6732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.267e-05,
+      "loss": 1.121,
+      "step": 6733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.266e-05,
+      "loss": 1.0627,
+      "step": 6734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.265e-05,
+      "loss": 0.9246,
+      "step": 6735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2640000000000006e-05,
+      "loss": 0.9539,
+      "step": 6736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.263e-05,
+      "loss": 1.1914,
+      "step": 6737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.262e-05,
+      "loss": 1.3635,
+      "step": 6738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.261e-05,
+      "loss": 0.6569,
+      "step": 6739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.26e-05,
+      "loss": 0.135,
+      "step": 6740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2590000000000005e-05,
+      "loss": 0.1923,
+      "step": 6741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2579999999999996e-05,
+      "loss": 0.0429,
+      "step": 6742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.257e-05,
+      "loss": 0.1479,
+      "step": 6743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.256e-05,
+      "loss": 1.2821,
+      "step": 6744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2550000000000005e-05,
+      "loss": 1.2884,
+      "step": 6745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2540000000000004e-05,
+      "loss": 1.0556,
+      "step": 6746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.253e-05,
+      "loss": 0.9546,
+      "step": 6747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.252e-05,
+      "loss": 1.086,
+      "step": 6748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.251e-05,
+      "loss": 0.8973,
+      "step": 6749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 1.2201,
+      "step": 6750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.249e-05,
+      "loss": 1.0215,
+      "step": 6751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.248e-05,
+      "loss": 0.5176,
+      "step": 6752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.247e-05,
+      "loss": 0.0459,
+      "step": 6753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2460000000000004e-05,
+      "loss": 0.0129,
+      "step": 6754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.245e-05,
+      "loss": 0.3396,
+      "step": 6755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.244e-05,
+      "loss": 1.0535,
+      "step": 6756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.243e-05,
+      "loss": 1.0489,
+      "step": 6757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.242e-05,
+      "loss": 1.3116,
+      "step": 6758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.241e-05,
+      "loss": 0.9609,
+      "step": 6759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.24e-05,
+      "loss": 1.0001,
+      "step": 6760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.239000000000001e-05,
+      "loss": 1.1786,
+      "step": 6761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.238e-05,
+      "loss": 1.0695,
+      "step": 6762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2370000000000003e-05,
+      "loss": 1.346,
+      "step": 6763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.236e-05,
+      "loss": 1.0982,
+      "step": 6764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.235e-05,
+      "loss": 0.2468,
+      "step": 6765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2340000000000005e-05,
+      "loss": 0.0441,
+      "step": 6766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.233e-05,
+      "loss": 0.0314,
+      "step": 6767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.232e-05,
+      "loss": 0.925,
+      "step": 6768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.231e-05,
+      "loss": 1.0687,
+      "step": 6769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2300000000000006e-05,
+      "loss": 1.7006,
+      "step": 6770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2290000000000004e-05,
+      "loss": 1.469,
+      "step": 6771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2279999999999996e-05,
+      "loss": 1.2453,
+      "step": 6772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.227e-05,
+      "loss": 0.947,
+      "step": 6773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.226e-05,
+      "loss": 0.9492,
+      "step": 6774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2250000000000005e-05,
+      "loss": 1.3786,
+      "step": 6775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.224e-05,
+      "loss": 0.7954,
+      "step": 6776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.223e-05,
+      "loss": 0.1689,
+      "step": 6777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.222e-05,
+      "loss": 0.2611,
+      "step": 6778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.221e-05,
+      "loss": 0.5262,
+      "step": 6779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 0.3612,
+      "step": 6780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.219e-05,
+      "loss": 1.0987,
+      "step": 6781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.218e-05,
+      "loss": 1.2232,
+      "step": 6782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.217e-05,
+      "loss": 1.2372,
+      "step": 6783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2160000000000004e-05,
+      "loss": 1.1857,
+      "step": 6784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.215e-05,
+      "loss": 0.9351,
+      "step": 6785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.214e-05,
+      "loss": 1.3293,
+      "step": 6786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.213e-05,
+      "loss": 1.5048,
+      "step": 6787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.212e-05,
+      "loss": 0.8893,
+      "step": 6788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.211e-05,
+      "loss": 0.5809,
+      "step": 6789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.21e-05,
+      "loss": 0.2519,
+      "step": 6790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2090000000000006e-05,
+      "loss": 0.0241,
+      "step": 6791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.208e-05,
+      "loss": 0.0905,
+      "step": 6792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.207e-05,
+      "loss": 0.8735,
+      "step": 6793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.206e-05,
+      "loss": 1.4296,
+      "step": 6794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.205e-05,
+      "loss": 0.7699,
+      "step": 6795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2040000000000005e-05,
+      "loss": 1.1762,
+      "step": 6796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2029999999999997e-05,
+      "loss": 1.105,
+      "step": 6797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.202e-05,
+      "loss": 1.203,
+      "step": 6798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.201e-05,
+      "loss": 1.3042,
+      "step": 6799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.3903,
+      "step": 6800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1990000000000004e-05,
+      "loss": 0.67,
+      "step": 6801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.198e-05,
+      "loss": 0.3587,
+      "step": 6802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.197e-05,
+      "loss": 0.0748,
+      "step": 6803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.196e-05,
+      "loss": 0.2934,
+      "step": 6804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1950000000000004e-05,
+      "loss": 0.1647,
+      "step": 6805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.194e-05,
+      "loss": 0.8161,
+      "step": 6806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.193e-05,
+      "loss": 1.0139,
+      "step": 6807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.192e-05,
+      "loss": 1.2605,
+      "step": 6808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.191e-05,
+      "loss": 1.1719,
+      "step": 6809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.19e-05,
+      "loss": 0.77,
+      "step": 6810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.189e-05,
+      "loss": 1.0305,
+      "step": 6811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.188e-05,
+      "loss": 1.4119,
+      "step": 6812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.187e-05,
+      "loss": 1.0671,
+      "step": 6813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.186e-05,
+      "loss": 0.3447,
+      "step": 6814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.185e-05,
+      "loss": 0.1376,
+      "step": 6815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.184e-05,
+      "loss": 0.406,
+      "step": 6816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1830000000000005e-05,
+      "loss": 0.3973,
+      "step": 6817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.182e-05,
+      "loss": 0.7174,
+      "step": 6818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.181e-05,
+      "loss": 1.0205,
+      "step": 6819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.18e-05,
+      "loss": 1.5244,
+      "step": 6820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1790000000000006e-05,
+      "loss": 1.0013,
+      "step": 6821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1780000000000004e-05,
+      "loss": 1.3292,
+      "step": 6822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.177e-05,
+      "loss": 1.0104,
+      "step": 6823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.176e-05,
+      "loss": 0.9364,
+      "step": 6824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.175e-05,
+      "loss": 1.4095,
+      "step": 6825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1740000000000004e-05,
+      "loss": 1.1413,
+      "step": 6826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.173e-05,
+      "loss": 0.3619,
+      "step": 6827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.172e-05,
+      "loss": 0.4142,
+      "step": 6828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.171e-05,
+      "loss": 0.0068,
+      "step": 6829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1700000000000005e-05,
+      "loss": 0.1065,
+      "step": 6830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.169e-05,
+      "loss": 0.524,
+      "step": 6831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.168e-05,
+      "loss": 1.2329,
+      "step": 6832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.167e-05,
+      "loss": 0.9473,
+      "step": 6833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.166e-05,
+      "loss": 1.5047,
+      "step": 6834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1650000000000004e-05,
+      "loss": 1.1035,
+      "step": 6835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.164e-05,
+      "loss": 1.3539,
+      "step": 6836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.163000000000001e-05,
+      "loss": 0.9426,
+      "step": 6837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.162e-05,
+      "loss": 1.3692,
+      "step": 6838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1610000000000004e-05,
+      "loss": 0.6006,
+      "step": 6839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.16e-05,
+      "loss": 0.1913,
+      "step": 6840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.159e-05,
+      "loss": 0.1061,
+      "step": 6841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1580000000000006e-05,
+      "loss": 0.1745,
+      "step": 6842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.157e-05,
+      "loss": 0.361,
+      "step": 6843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.156e-05,
+      "loss": 1.1637,
+      "step": 6844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.155e-05,
+      "loss": 1.1865,
+      "step": 6845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.154e-05,
+      "loss": 1.1074,
+      "step": 6846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1530000000000005e-05,
+      "loss": 0.9469,
+      "step": 6847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1519999999999996e-05,
+      "loss": 1.0344,
+      "step": 6848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.151e-05,
+      "loss": 0.8148,
+      "step": 6849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.15e-05,
+      "loss": 1.0408,
+      "step": 6850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1490000000000005e-05,
+      "loss": 0.966,
+      "step": 6851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1480000000000004e-05,
+      "loss": 0.0854,
+      "step": 6852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.147e-05,
+      "loss": 0.0436,
+      "step": 6853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.146e-05,
+      "loss": 0.1949,
+      "step": 6854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.145e-05,
+      "loss": 0.0778,
+      "step": 6855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1440000000000004e-05,
+      "loss": 0.8393,
+      "step": 6856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.143e-05,
+      "loss": 1.1911,
+      "step": 6857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.142e-05,
+      "loss": 1.1383,
+      "step": 6858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.141e-05,
+      "loss": 0.9844,
+      "step": 6859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 0.991,
+      "step": 6860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.139e-05,
+      "loss": 1.0177,
+      "step": 6861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.138e-05,
+      "loss": 0.8261,
+      "step": 6862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.137e-05,
+      "loss": 1.3407,
+      "step": 6863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.136e-05,
+      "loss": 0.9385,
+      "step": 6864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.135e-05,
+      "loss": 0.2606,
+      "step": 6865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.134e-05,
+      "loss": 0.0523,
+      "step": 6866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.133000000000001e-05,
+      "loss": 0.0507,
+      "step": 6867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.132e-05,
+      "loss": 0.8417,
+      "step": 6868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1310000000000003e-05,
+      "loss": 1.3533,
+      "step": 6869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.13e-05,
+      "loss": 1.154,
+      "step": 6870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.129e-05,
+      "loss": 1.4669,
+      "step": 6871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1280000000000005e-05,
+      "loss": 1.0511,
+      "step": 6872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.127e-05,
+      "loss": 1.1564,
+      "step": 6873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.126e-05,
+      "loss": 1.4431,
+      "step": 6874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.125e-05,
+      "loss": 0.9257,
+      "step": 6875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1240000000000006e-05,
+      "loss": 1.1738,
+      "step": 6876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1230000000000004e-05,
+      "loss": 0.5492,
+      "step": 6877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.122e-05,
+      "loss": 0.0567,
+      "step": 6878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.121e-05,
+      "loss": 0.4105,
+      "step": 6879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.12e-05,
+      "loss": 0.6517,
+      "step": 6880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1190000000000005e-05,
+      "loss": 1.2521,
+      "step": 6881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.118e-05,
+      "loss": 1.2959,
+      "step": 6882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.117e-05,
+      "loss": 1.2437,
+      "step": 6883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.116e-05,
+      "loss": 0.9836,
+      "step": 6884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.115e-05,
+      "loss": 1.1771,
+      "step": 6885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1140000000000003e-05,
+      "loss": 1.4416,
+      "step": 6886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.113e-05,
+      "loss": 0.8981,
+      "step": 6887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.112e-05,
+      "loss": 1.3557,
+      "step": 6888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.111e-05,
+      "loss": 0.9825,
+      "step": 6889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1100000000000004e-05,
+      "loss": 0.1269,
+      "step": 6890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.109e-05,
+      "loss": 0.0093,
+      "step": 6891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.108e-05,
+      "loss": 0.4307,
+      "step": 6892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.107e-05,
+      "loss": 0.8394,
+      "step": 6893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.106e-05,
+      "loss": 1.1704,
+      "step": 6894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.105e-05,
+      "loss": 0.9691,
+      "step": 6895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.104e-05,
+      "loss": 0.9217,
+      "step": 6896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1030000000000006e-05,
+      "loss": 1.0343,
+      "step": 6897
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.102e-05,
+      "loss": 1.1021,
+      "step": 6898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.101e-05,
+      "loss": 0.9033,
+      "step": 6899
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1e-05,
+      "loss": 1.4352,
+      "step": 6900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.099e-05,
+      "loss": 1.2719,
+      "step": 6901
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0980000000000005e-05,
+      "loss": 0.4986,
+      "step": 6902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0969999999999997e-05,
+      "loss": 0.1758,
+      "step": 6903
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.096e-05,
+      "loss": 0.0642,
+      "step": 6904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.095e-05,
+      "loss": 0.1007,
+      "step": 6905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0940000000000005e-05,
+      "loss": 0.2652,
+      "step": 6906
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0930000000000004e-05,
+      "loss": 1.1465,
+      "step": 6907
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.092e-05,
+      "loss": 1.1723,
+      "step": 6908
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.091e-05,
+      "loss": 1.2547,
+      "step": 6909
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.09e-05,
+      "loss": 1.0677,
+      "step": 6910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0890000000000004e-05,
+      "loss": 1.2432,
+      "step": 6911
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.088e-05,
+      "loss": 1.3995,
+      "step": 6912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.087e-05,
+      "loss": 1.299,
+      "step": 6913
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.086e-05,
+      "loss": 0.5233,
+      "step": 6914
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0850000000000004e-05,
+      "loss": 0.5497,
+      "step": 6915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.084e-05,
+      "loss": 0.2346,
+      "step": 6916
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.083e-05,
+      "loss": 0.1939,
+      "step": 6917
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.082e-05,
+      "loss": 0.7418,
+      "step": 6918
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.081e-05,
+      "loss": 1.3301,
+      "step": 6919
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.08e-05,
+      "loss": 1.2177,
+      "step": 6920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.079e-05,
+      "loss": 1.2593,
+      "step": 6921
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.078e-05,
+      "loss": 1.4839,
+      "step": 6922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.077e-05,
+      "loss": 1.1006,
+      "step": 6923
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.076e-05,
+      "loss": 1.0034,
+      "step": 6924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.075e-05,
+      "loss": 1.2168,
+      "step": 6925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.074e-05,
+      "loss": 1.1971,
+      "step": 6926
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0730000000000006e-05,
+      "loss": 0.1512,
+      "step": 6927
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.072e-05,
+      "loss": 0.0019,
+      "step": 6928
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.071e-05,
+      "loss": 0.0203,
+      "step": 6929
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.07e-05,
+      "loss": 0.0722,
+      "step": 6930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.069e-05,
+      "loss": 1.1625,
+      "step": 6931
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0680000000000004e-05,
+      "loss": 1.2711,
+      "step": 6932
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0669999999999996e-05,
+      "loss": 0.9458,
+      "step": 6933
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.066e-05,
+      "loss": 1.0291,
+      "step": 6934
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.065e-05,
+      "loss": 1.4432,
+      "step": 6935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0640000000000005e-05,
+      "loss": 1.1683,
+      "step": 6936
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.063e-05,
+      "loss": 0.6184,
+      "step": 6937
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.062e-05,
+      "loss": 1.3389,
+      "step": 6938
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.061e-05,
+      "loss": 0.7628,
+      "step": 6939
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.06e-05,
+      "loss": 0.2068,
+      "step": 6940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0590000000000004e-05,
+      "loss": 0.181,
+      "step": 6941
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.058e-05,
+      "loss": 0.0432,
+      "step": 6942
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.057000000000001e-05,
+      "loss": 0.5329,
+      "step": 6943
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.056e-05,
+      "loss": 1.0223,
+      "step": 6944
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0550000000000004e-05,
+      "loss": 1.3955,
+      "step": 6945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.054e-05,
+      "loss": 0.8877,
+      "step": 6946
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.053e-05,
+      "loss": 1.3253,
+      "step": 6947
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0520000000000006e-05,
+      "loss": 1.0533,
+      "step": 6948
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.051e-05,
+      "loss": 0.6926,
+      "step": 6949
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.05e-05,
+      "loss": 1.5714,
+      "step": 6950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.049e-05,
+      "loss": 1.2288,
+      "step": 6951
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0480000000000003e-05,
+      "loss": 0.3296,
+      "step": 6952
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0470000000000005e-05,
+      "loss": 0.1256,
+      "step": 6953
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.046e-05,
+      "loss": 0.1304,
+      "step": 6954
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.045e-05,
+      "loss": 0.1664,
+      "step": 6955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0440000000000003e-05,
+      "loss": 0.8546,
+      "step": 6956
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0430000000000002e-05,
+      "loss": 1.2832,
+      "step": 6957
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0420000000000004e-05,
+      "loss": 1.0151,
+      "step": 6958
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.041e-05,
+      "loss": 1.2642,
+      "step": 6959
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.04e-05,
+      "loss": 1.1015,
+      "step": 6960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0390000000000002e-05,
+      "loss": 1.2261,
+      "step": 6961
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0380000000000004e-05,
+      "loss": 0.813,
+      "step": 6962
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0370000000000006e-05,
+      "loss": 1.1297,
+      "step": 6963
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.036e-05,
+      "loss": 0.7576,
+      "step": 6964
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.035e-05,
+      "loss": 0.4699,
+      "step": 6965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.034e-05,
+      "loss": 0.1038,
+      "step": 6966
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0330000000000003e-05,
+      "loss": 0.1052,
+      "step": 6967
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0320000000000004e-05,
+      "loss": 0.9025,
+      "step": 6968
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.031e-05,
+      "loss": 1.2169,
+      "step": 6969
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.03e-05,
+      "loss": 1.1264,
+      "step": 6970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0290000000000003e-05,
+      "loss": 1.044,
+      "step": 6971
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.028e-05,
+      "loss": 0.8757,
+      "step": 6972
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0270000000000003e-05,
+      "loss": 1.1926,
+      "step": 6973
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0259999999999998e-05,
+      "loss": 0.8275,
+      "step": 6974
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.025e-05,
+      "loss": 1.3295,
+      "step": 6975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0240000000000002e-05,
+      "loss": 1.4305,
+      "step": 6976
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0230000000000004e-05,
+      "loss": 0.654,
+      "step": 6977
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0220000000000005e-05,
+      "loss": 0.1501,
+      "step": 6978
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.021e-05,
+      "loss": 0.097,
+      "step": 6979
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.02e-05,
+      "loss": 0.0841,
+      "step": 6980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.019e-05,
+      "loss": 0.2289,
+      "step": 6981
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0180000000000002e-05,
+      "loss": 0.9812,
+      "step": 6982
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0170000000000004e-05,
+      "loss": 1.6692,
+      "step": 6983
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.016e-05,
+      "loss": 1.6509,
+      "step": 6984
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.015e-05,
+      "loss": 1.0945,
+      "step": 6985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0140000000000003e-05,
+      "loss": 1.4408,
+      "step": 6986
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.013e-05,
+      "loss": 1.1384,
+      "step": 6987
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0120000000000003e-05,
+      "loss": 0.8029,
+      "step": 6988
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0109999999999998e-05,
+      "loss": 1.1222,
+      "step": 6989
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.01e-05,
+      "loss": 0.0439,
+      "step": 6990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.009e-05,
+      "loss": 0.069,
+      "step": 6991
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0080000000000003e-05,
+      "loss": 0.2212,
+      "step": 6992
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0070000000000005e-05,
+      "loss": 0.7549,
+      "step": 6993
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.006e-05,
+      "loss": 1.1104,
+      "step": 6994
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0050000000000002e-05,
+      "loss": 1.3288,
+      "step": 6995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.004e-05,
+      "loss": 1.2525,
+      "step": 6996
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0030000000000002e-05,
+      "loss": 1.4042,
+      "step": 6997
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.0020000000000004e-05,
+      "loss": 1.1726,
+      "step": 6998
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.001e-05,
+      "loss": 1.1666,
+      "step": 6999
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3e-05,
+      "loss": 1.5618,
+      "step": 7000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990000000000003e-05,
+      "loss": 1.2818,
+      "step": 7001
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.998e-05,
+      "loss": 0.3047,
+      "step": 7002
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9970000000000003e-05,
+      "loss": 0.2751,
+      "step": 7003
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9959999999999998e-05,
+      "loss": 0.1835,
+      "step": 7004
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.995e-05,
+      "loss": 0.0256,
+      "step": 7005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.994e-05,
+      "loss": 1.0076,
+      "step": 7006
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9930000000000003e-05,
+      "loss": 1.1474,
+      "step": 7007
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9920000000000005e-05,
+      "loss": 1.3529,
+      "step": 7008
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.991e-05,
+      "loss": 1.5133,
+      "step": 7009
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9900000000000002e-05,
+      "loss": 1.1832,
+      "step": 7010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.989e-05,
+      "loss": 0.772,
+      "step": 7011
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9880000000000002e-05,
+      "loss": 1.0569,
+      "step": 7012
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9870000000000004e-05,
+      "loss": 0.9185,
+      "step": 7013
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.986e-05,
+      "loss": 0.3607,
+      "step": 7014
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.985e-05,
+      "loss": 0.0282,
+      "step": 7015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9840000000000002e-05,
+      "loss": 0.0407,
+      "step": 7016
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9830000000000004e-05,
+      "loss": 0.2695,
+      "step": 7017
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9820000000000002e-05,
+      "loss": 0.5471,
+      "step": 7018
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9809999999999997e-05,
+      "loss": 0.8805,
+      "step": 7019
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.98e-05,
+      "loss": 1.5178,
+      "step": 7020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.979e-05,
+      "loss": 0.94,
+      "step": 7021
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9780000000000003e-05,
+      "loss": 0.8382,
+      "step": 7022
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9770000000000005e-05,
+      "loss": 1.3205,
+      "step": 7023
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976e-05,
+      "loss": 1.4607,
+      "step": 7024
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975e-05,
+      "loss": 1.2549,
+      "step": 7025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974e-05,
+      "loss": 0.873,
+      "step": 7026
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.973e-05,
+      "loss": 0.1676,
+      "step": 7027
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9720000000000003e-05,
+      "loss": 0.0248,
+      "step": 7028
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.971e-05,
+      "loss": 0.2913,
+      "step": 7029
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.97e-05,
+      "loss": 0.2939,
+      "step": 7030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9690000000000002e-05,
+      "loss": 0.9408,
+      "step": 7031
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9680000000000004e-05,
+      "loss": 1.5015,
+      "step": 7032
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9670000000000002e-05,
+      "loss": 1.2734,
+      "step": 7033
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9659999999999997e-05,
+      "loss": 1.0328,
+      "step": 7034
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.965e-05,
+      "loss": 1.4326,
+      "step": 7035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.964e-05,
+      "loss": 0.9043,
+      "step": 7036
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9630000000000003e-05,
+      "loss": 0.7865,
+      "step": 7037
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9620000000000004e-05,
+      "loss": 1.0334,
+      "step": 7038
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.961e-05,
+      "loss": 0.5501,
+      "step": 7039
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.96e-05,
+      "loss": 0.1398,
+      "step": 7040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.959e-05,
+      "loss": 0.1735,
+      "step": 7041
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.958e-05,
+      "loss": 0.1521,
+      "step": 7042
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9570000000000003e-05,
+      "loss": 0.4923,
+      "step": 7043
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9559999999999998e-05,
+      "loss": 0.7044,
+      "step": 7044
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.955e-05,
+      "loss": 0.8687,
+      "step": 7045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9540000000000002e-05,
+      "loss": 1.2606,
+      "step": 7046
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9530000000000004e-05,
+      "loss": 1.0975,
+      "step": 7047
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9520000000000002e-05,
+      "loss": 1.0956,
+      "step": 7048
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.951e-05,
+      "loss": 1.1072,
+      "step": 7049
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.95e-05,
+      "loss": 1.4773,
+      "step": 7050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.949e-05,
+      "loss": 1.2232,
+      "step": 7051
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9480000000000002e-05,
+      "loss": 0.6102,
+      "step": 7052
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9470000000000004e-05,
+      "loss": 0.0127,
+      "step": 7053
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.946e-05,
+      "loss": 0.2428,
+      "step": 7054
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.945e-05,
+      "loss": 0.3764,
+      "step": 7055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.944e-05,
+      "loss": 0.6671,
+      "step": 7056
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.943e-05,
+      "loss": 0.9166,
+      "step": 7057
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9420000000000003e-05,
+      "loss": 1.3828,
+      "step": 7058
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9409999999999998e-05,
+      "loss": 1.1808,
+      "step": 7059
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.94e-05,
+      "loss": 1.2681,
+      "step": 7060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.939e-05,
+      "loss": 1.3502,
+      "step": 7061
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9380000000000003e-05,
+      "loss": 0.9646,
+      "step": 7062
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9370000000000002e-05,
+      "loss": 1.1753,
+      "step": 7063
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9360000000000003e-05,
+      "loss": 0.4124,
+      "step": 7064
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.935e-05,
+      "loss": 0.0737,
+      "step": 7065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.934e-05,
+      "loss": 0.2172,
+      "step": 7066
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9330000000000002e-05,
+      "loss": 0.1453,
+      "step": 7067
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9320000000000004e-05,
+      "loss": 0.5463,
+      "step": 7068
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9310000000000006e-05,
+      "loss": 0.8225,
+      "step": 7069
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.93e-05,
+      "loss": 0.9729,
+      "step": 7070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.929e-05,
+      "loss": 1.0484,
+      "step": 7071
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.928e-05,
+      "loss": 1.1231,
+      "step": 7072
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9270000000000003e-05,
+      "loss": 1.1996,
+      "step": 7073
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9260000000000004e-05,
+      "loss": 0.9695,
+      "step": 7074
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.925e-05,
+      "loss": 0.773,
+      "step": 7075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.924e-05,
+      "loss": 0.9663,
+      "step": 7076
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9230000000000003e-05,
+      "loss": 0.3698,
+      "step": 7077
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.922e-05,
+      "loss": 0.2522,
+      "step": 7078
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9210000000000003e-05,
+      "loss": 0.1671,
+      "step": 7079
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 0.1101,
+      "step": 7080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.919e-05,
+      "loss": 0.911,
+      "step": 7081
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9180000000000002e-05,
+      "loss": 1.1618,
+      "step": 7082
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9170000000000004e-05,
+      "loss": 1.1395,
+      "step": 7083
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9160000000000005e-05,
+      "loss": 1.1589,
+      "step": 7084
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.915e-05,
+      "loss": 0.9773,
+      "step": 7085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9140000000000002e-05,
+      "loss": 1.2383,
+      "step": 7086
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.913e-05,
+      "loss": 1.3512,
+      "step": 7087
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9120000000000002e-05,
+      "loss": 1.385,
+      "step": 7088
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9110000000000004e-05,
+      "loss": 0.6141,
+      "step": 7089
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.91e-05,
+      "loss": 0.2638,
+      "step": 7090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.909e-05,
+      "loss": 0.1373,
+      "step": 7091
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9080000000000003e-05,
+      "loss": 0.0688,
+      "step": 7092
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.907e-05,
+      "loss": 0.2833,
+      "step": 7093
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9060000000000003e-05,
+      "loss": 1.0893,
+      "step": 7094
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9049999999999998e-05,
+      "loss": 1.2967,
+      "step": 7095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.904e-05,
+      "loss": 1.5364,
+      "step": 7096
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.903e-05,
+      "loss": 1.1954,
+      "step": 7097
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9020000000000003e-05,
+      "loss": 0.8599,
+      "step": 7098
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9010000000000005e-05,
+      "loss": 0.8786,
+      "step": 7099
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9e-05,
+      "loss": 0.8008,
+      "step": 7100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8990000000000002e-05,
+      "loss": 1.1267,
+      "step": 7101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.898e-05,
+      "loss": 0.2584,
+      "step": 7102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8970000000000002e-05,
+      "loss": 0.0168,
+      "step": 7103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8960000000000004e-05,
+      "loss": 0.2901,
+      "step": 7104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.895e-05,
+      "loss": 0.4461,
+      "step": 7105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.894e-05,
+      "loss": 1.0486,
+      "step": 7106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8930000000000003e-05,
+      "loss": 1.1019,
+      "step": 7107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8920000000000004e-05,
+      "loss": 1.3993,
+      "step": 7108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8910000000000003e-05,
+      "loss": 1.1998,
+      "step": 7109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8899999999999998e-05,
+      "loss": 1.3409,
+      "step": 7110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.889e-05,
+      "loss": 0.8869,
+      "step": 7111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.888e-05,
+      "loss": 0.8583,
+      "step": 7112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8870000000000003e-05,
+      "loss": 1.0561,
+      "step": 7113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8860000000000005e-05,
+      "loss": 0.4167,
+      "step": 7114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.885e-05,
+      "loss": 0.2192,
+      "step": 7115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8840000000000002e-05,
+      "loss": 0.025,
+      "step": 7116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.883e-05,
+      "loss": 0.2958,
+      "step": 7117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8820000000000002e-05,
+      "loss": 1.0058,
+      "step": 7118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8810000000000004e-05,
+      "loss": 1.121,
+      "step": 7119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.88e-05,
+      "loss": 0.9242,
+      "step": 7120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.879e-05,
+      "loss": 1.2716,
+      "step": 7121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8780000000000002e-05,
+      "loss": 0.9149,
+      "step": 7122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8770000000000004e-05,
+      "loss": 1.1966,
+      "step": 7123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8760000000000002e-05,
+      "loss": 1.268,
+      "step": 7124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8749999999999997e-05,
+      "loss": 1.6679,
+      "step": 7125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.874e-05,
+      "loss": 1.1823,
+      "step": 7126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.873e-05,
+      "loss": 0.3268,
+      "step": 7127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8720000000000003e-05,
+      "loss": 0.0874,
+      "step": 7128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8710000000000005e-05,
+      "loss": 0.2757,
+      "step": 7129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.87e-05,
+      "loss": 0.358,
+      "step": 7130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.869e-05,
+      "loss": 0.6155,
+      "step": 7131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.868e-05,
+      "loss": 1.0907,
+      "step": 7132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.867e-05,
+      "loss": 1.3734,
+      "step": 7133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8660000000000003e-05,
+      "loss": 1.3245,
+      "step": 7134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.865e-05,
+      "loss": 1.1319,
+      "step": 7135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.864e-05,
+      "loss": 1.1161,
+      "step": 7136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8630000000000002e-05,
+      "loss": 1.0862,
+      "step": 7137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8620000000000004e-05,
+      "loss": 1.0822,
+      "step": 7138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8610000000000002e-05,
+      "loss": 0.6371,
+      "step": 7139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.86e-05,
+      "loss": 0.0868,
+      "step": 7140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.859e-05,
+      "loss": 0.4099,
+      "step": 7141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.858e-05,
+      "loss": 0.0013,
+      "step": 7142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8570000000000003e-05,
+      "loss": 0.4026,
+      "step": 7143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8560000000000004e-05,
+      "loss": 0.7888,
+      "step": 7144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.855e-05,
+      "loss": 1.0549,
+      "step": 7145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.854e-05,
+      "loss": 1.4557,
+      "step": 7146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.853e-05,
+      "loss": 1.3307,
+      "step": 7147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.852e-05,
+      "loss": 1.1453,
+      "step": 7148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8510000000000003e-05,
+      "loss": 1.1735,
+      "step": 7149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 1.6231,
+      "step": 7150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.849e-05,
+      "loss": 1.3132,
+      "step": 7151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8480000000000002e-05,
+      "loss": 0.0424,
+      "step": 7152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8470000000000004e-05,
+      "loss": 0.039,
+      "step": 7153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8460000000000002e-05,
+      "loss": 0.0393,
+      "step": 7154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.845e-05,
+      "loss": 0.7442,
+      "step": 7155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.844e-05,
+      "loss": 1.1005,
+      "step": 7156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.843e-05,
+      "loss": 0.9096,
+      "step": 7157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8420000000000002e-05,
+      "loss": 1.1307,
+      "step": 7158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8410000000000004e-05,
+      "loss": 0.9543,
+      "step": 7159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.84e-05,
+      "loss": 0.8281,
+      "step": 7160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.839e-05,
+      "loss": 1.1174,
+      "step": 7161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8380000000000003e-05,
+      "loss": 0.9943,
+      "step": 7162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.837e-05,
+      "loss": 0.9778,
+      "step": 7163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8360000000000003e-05,
+      "loss": 1.053,
+      "step": 7164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8349999999999998e-05,
+      "loss": 0.1118,
+      "step": 7165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.834e-05,
+      "loss": 0.0087,
+      "step": 7166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.833e-05,
+      "loss": 0.5369,
+      "step": 7167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8320000000000003e-05,
+      "loss": 0.6213,
+      "step": 7168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8310000000000002e-05,
+      "loss": 1.2606,
+      "step": 7169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.83e-05,
+      "loss": 1.6496,
+      "step": 7170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.829e-05,
+      "loss": 1.0789,
+      "step": 7171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.828e-05,
+      "loss": 0.5871,
+      "step": 7172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8270000000000002e-05,
+      "loss": 1.5256,
+      "step": 7173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8260000000000004e-05,
+      "loss": 0.9149,
+      "step": 7174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.825e-05,
+      "loss": 1.2061,
+      "step": 7175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.824e-05,
+      "loss": 0.9039,
+      "step": 7176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8230000000000002e-05,
+      "loss": 0.4468,
+      "step": 7177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.822e-05,
+      "loss": 0.0223,
+      "step": 7178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8210000000000003e-05,
+      "loss": 0.0522,
+      "step": 7179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 0.1444,
+      "step": 7180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.819e-05,
+      "loss": 0.7536,
+      "step": 7181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.818e-05,
+      "loss": 1.0404,
+      "step": 7182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8170000000000003e-05,
+      "loss": 1.0584,
+      "step": 7183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.816e-05,
+      "loss": 1.1995,
+      "step": 7184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.815e-05,
+      "loss": 1.2594,
+      "step": 7185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8139999999999998e-05,
+      "loss": 1.4004,
+      "step": 7186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.813e-05,
+      "loss": 1.6736,
+      "step": 7187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8120000000000002e-05,
+      "loss": 1.3109,
+      "step": 7188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8110000000000004e-05,
+      "loss": 0.5273,
+      "step": 7189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8100000000000005e-05,
+      "loss": 0.1913,
+      "step": 7190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.809e-05,
+      "loss": 0.0375,
+      "step": 7191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8080000000000002e-05,
+      "loss": 0.1139,
+      "step": 7192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.807e-05,
+      "loss": 0.7078,
+      "step": 7193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8060000000000002e-05,
+      "loss": 0.6584,
+      "step": 7194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8050000000000004e-05,
+      "loss": 1.0153,
+      "step": 7195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.804e-05,
+      "loss": 1.2393,
+      "step": 7196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.803e-05,
+      "loss": 1.1328,
+      "step": 7197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8020000000000003e-05,
+      "loss": 0.5784,
+      "step": 7198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8010000000000005e-05,
+      "loss": 1.3362,
+      "step": 7199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.524,
+      "step": 7200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7989999999999998e-05,
+      "loss": 1.5056,
+      "step": 7201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.798e-05,
+      "loss": 0.1132,
+      "step": 7202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.797e-05,
+      "loss": 0.0599,
+      "step": 7203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7960000000000003e-05,
+      "loss": 0.1465,
+      "step": 7204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7950000000000005e-05,
+      "loss": 0.1609,
+      "step": 7205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.794e-05,
+      "loss": 0.7287,
+      "step": 7206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7930000000000002e-05,
+      "loss": 1.1759,
+      "step": 7207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.792e-05,
+      "loss": 1.3992,
+      "step": 7208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7910000000000002e-05,
+      "loss": 1.0692,
+      "step": 7209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7900000000000004e-05,
+      "loss": 1.019,
+      "step": 7210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.789e-05,
+      "loss": 1.048,
+      "step": 7211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.788e-05,
+      "loss": 1.1017,
+      "step": 7212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7870000000000003e-05,
+      "loss": 1.2277,
+      "step": 7213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7860000000000004e-05,
+      "loss": 0.6907,
+      "step": 7214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7850000000000003e-05,
+      "loss": 0.1258,
+      "step": 7215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7839999999999998e-05,
+      "loss": 0.1827,
+      "step": 7216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.783e-05,
+      "loss": 0.3013,
+      "step": 7217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.782e-05,
+      "loss": 1.0023,
+      "step": 7218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7810000000000003e-05,
+      "loss": 1.5301,
+      "step": 7219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 1.2519,
+      "step": 7220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.779e-05,
+      "loss": 1.0286,
+      "step": 7221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.778e-05,
+      "loss": 1.1133,
+      "step": 7222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.777e-05,
+      "loss": 1.2042,
+      "step": 7223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7760000000000002e-05,
+      "loss": 0.9915,
+      "step": 7224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7750000000000004e-05,
+      "loss": 1.0171,
+      "step": 7225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.774e-05,
+      "loss": 1.0124,
+      "step": 7226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.773e-05,
+      "loss": 0.0477,
+      "step": 7227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7720000000000002e-05,
+      "loss": 0.2675,
+      "step": 7228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7710000000000004e-05,
+      "loss": 0.293,
+      "step": 7229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7700000000000002e-05,
+      "loss": 0.1456,
+      "step": 7230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.769e-05,
+      "loss": 0.8732,
+      "step": 7231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.768e-05,
+      "loss": 1.1379,
+      "step": 7232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.767e-05,
+      "loss": 1.215,
+      "step": 7233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7660000000000003e-05,
+      "loss": 1.2665,
+      "step": 7234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7650000000000005e-05,
+      "loss": 1.1982,
+      "step": 7235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.764e-05,
+      "loss": 0.9378,
+      "step": 7236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.763e-05,
+      "loss": 1.1653,
+      "step": 7237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.762e-05,
+      "loss": 1.3819,
+      "step": 7238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.761e-05,
+      "loss": 1.157,
+      "step": 7239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 0.2,
+      "step": 7240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.759e-05,
+      "loss": 0.247,
+      "step": 7241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.758e-05,
+      "loss": 0.3394,
+      "step": 7242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7570000000000002e-05,
+      "loss": 0.3108,
+      "step": 7243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7560000000000004e-05,
+      "loss": 0.8565,
+      "step": 7244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7550000000000002e-05,
+      "loss": 0.8947,
+      "step": 7245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.754e-05,
+      "loss": 1.4762,
+      "step": 7246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.753e-05,
+      "loss": 1.1543,
+      "step": 7247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.752e-05,
+      "loss": 1.3229,
+      "step": 7248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7510000000000003e-05,
+      "loss": 1.2376,
+      "step": 7249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.8314,
+      "step": 7250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.749e-05,
+      "loss": 1.5132,
+      "step": 7251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.748e-05,
+      "loss": 0.1994,
+      "step": 7252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7470000000000003e-05,
+      "loss": 0.1433,
+      "step": 7253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.746e-05,
+      "loss": 0.273,
+      "step": 7254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7450000000000003e-05,
+      "loss": 0.4706,
+      "step": 7255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7439999999999998e-05,
+      "loss": 0.7176,
+      "step": 7256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.743e-05,
+      "loss": 0.9984,
+      "step": 7257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7420000000000002e-05,
+      "loss": 1.4204,
+      "step": 7258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7410000000000004e-05,
+      "loss": 0.8026,
+      "step": 7259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 1.1069,
+      "step": 7260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.739e-05,
+      "loss": 1.3937,
+      "step": 7261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.738e-05,
+      "loss": 1.1958,
+      "step": 7262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.737e-05,
+      "loss": 1.3502,
+      "step": 7263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7360000000000002e-05,
+      "loss": 0.6981,
+      "step": 7264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7350000000000004e-05,
+      "loss": 0.3647,
+      "step": 7265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.734e-05,
+      "loss": 0.1801,
+      "step": 7266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.733e-05,
+      "loss": 0.1208,
+      "step": 7267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7320000000000003e-05,
+      "loss": 1.2854,
+      "step": 7268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.731e-05,
+      "loss": 0.9216,
+      "step": 7269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 1.1803,
+      "step": 7270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7289999999999998e-05,
+      "loss": 1.4852,
+      "step": 7271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.728e-05,
+      "loss": 1.2645,
+      "step": 7272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.727e-05,
+      "loss": 1.3675,
+      "step": 7273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7260000000000003e-05,
+      "loss": 1.2712,
+      "step": 7274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.725e-05,
+      "loss": 1.539,
+      "step": 7275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.724e-05,
+      "loss": 0.9691,
+      "step": 7276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.723e-05,
+      "loss": 0.364,
+      "step": 7277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.722e-05,
+      "loss": 0.1259,
+      "step": 7278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7210000000000002e-05,
+      "loss": 0.0794,
+      "step": 7279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 0.2251,
+      "step": 7280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.719e-05,
+      "loss": 0.8031,
+      "step": 7281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.718e-05,
+      "loss": 1.256,
+      "step": 7282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7170000000000002e-05,
+      "loss": 1.1859,
+      "step": 7283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.716e-05,
+      "loss": 1.3197,
+      "step": 7284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7150000000000003e-05,
+      "loss": 1.3179,
+      "step": 7285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7139999999999998e-05,
+      "loss": 0.6855,
+      "step": 7286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.713e-05,
+      "loss": 0.8422,
+      "step": 7287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.712e-05,
+      "loss": 0.8569,
+      "step": 7288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7110000000000003e-05,
+      "loss": 0.17,
+      "step": 7289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7100000000000005e-05,
+      "loss": 0.2434,
+      "step": 7290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.709e-05,
+      "loss": 0.4263,
+      "step": 7291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7079999999999998e-05,
+      "loss": 0.0077,
+      "step": 7292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.707e-05,
+      "loss": 1.1251,
+      "step": 7293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7060000000000002e-05,
+      "loss": 0.8929,
+      "step": 7294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7050000000000004e-05,
+      "loss": 0.9989,
+      "step": 7295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.704e-05,
+      "loss": 0.7687,
+      "step": 7296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.703e-05,
+      "loss": 0.835,
+      "step": 7297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7020000000000002e-05,
+      "loss": 1.1959,
+      "step": 7298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.701e-05,
+      "loss": 1.0828,
+      "step": 7299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.4756,
+      "step": 7300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6989999999999997e-05,
+      "loss": 1.2191,
+      "step": 7301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.698e-05,
+      "loss": 0.295,
+      "step": 7302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.697e-05,
+      "loss": 0.3863,
+      "step": 7303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6960000000000003e-05,
+      "loss": 0.1577,
+      "step": 7304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6950000000000005e-05,
+      "loss": 0.2499,
+      "step": 7305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.694e-05,
+      "loss": 1.2228,
+      "step": 7306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.693e-05,
+      "loss": 0.6334,
+      "step": 7307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.692e-05,
+      "loss": 1.3363,
+      "step": 7308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.691e-05,
+      "loss": 0.957,
+      "step": 7309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6900000000000003e-05,
+      "loss": 1.2475,
+      "step": 7310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.689e-05,
+      "loss": 1.1203,
+      "step": 7311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.688e-05,
+      "loss": 0.7104,
+      "step": 7312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6870000000000002e-05,
+      "loss": 1.2166,
+      "step": 7313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.686e-05,
+      "loss": 0.615,
+      "step": 7314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6850000000000002e-05,
+      "loss": 0.0624,
+      "step": 7315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6840000000000004e-05,
+      "loss": 0.1281,
+      "step": 7316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.683e-05,
+      "loss": 0.0378,
+      "step": 7317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.682e-05,
+      "loss": 0.5763,
+      "step": 7318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6810000000000003e-05,
+      "loss": 1.3414,
+      "step": 7319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 1.4209,
+      "step": 7320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6790000000000003e-05,
+      "loss": 1.0287,
+      "step": 7321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.678e-05,
+      "loss": 0.6575,
+      "step": 7322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.677e-05,
+      "loss": 0.9529,
+      "step": 7323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.676e-05,
+      "loss": 1.0862,
+      "step": 7324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6750000000000003e-05,
+      "loss": 1.1275,
+      "step": 7325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6740000000000005e-05,
+      "loss": 1.441,
+      "step": 7326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.673e-05,
+      "loss": 0.5622,
+      "step": 7327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.672e-05,
+      "loss": 0.0156,
+      "step": 7328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.671e-05,
+      "loss": 0.0505,
+      "step": 7329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 0.1211,
+      "step": 7330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6690000000000004e-05,
+      "loss": 0.868,
+      "step": 7331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.668e-05,
+      "loss": 0.8292,
+      "step": 7332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.667e-05,
+      "loss": 1.21,
+      "step": 7333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6660000000000002e-05,
+      "loss": 1.2189,
+      "step": 7334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6650000000000004e-05,
+      "loss": 0.949,
+      "step": 7335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6640000000000002e-05,
+      "loss": 1.0111,
+      "step": 7336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.663e-05,
+      "loss": 1.2335,
+      "step": 7337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.662e-05,
+      "loss": 1.4676,
+      "step": 7338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.661e-05,
+      "loss": 0.8262,
+      "step": 7339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 0.3118,
+      "step": 7340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6590000000000005e-05,
+      "loss": 0.0374,
+      "step": 7341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.658e-05,
+      "loss": 0.4973,
+      "step": 7342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.657e-05,
+      "loss": 0.4268,
+      "step": 7343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6560000000000003e-05,
+      "loss": 0.9576,
+      "step": 7344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.655e-05,
+      "loss": 1.3165,
+      "step": 7345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6540000000000003e-05,
+      "loss": 1.4179,
+      "step": 7346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.653e-05,
+      "loss": 1.0685,
+      "step": 7347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.652e-05,
+      "loss": 0.9428,
+      "step": 7348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6510000000000002e-05,
+      "loss": 1.1892,
+      "step": 7349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 0.9407,
+      "step": 7350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6490000000000002e-05,
+      "loss": 1.1274,
+      "step": 7351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.648e-05,
+      "loss": 0.1459,
+      "step": 7352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.647e-05,
+      "loss": 0.305,
+      "step": 7353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.646e-05,
+      "loss": 0.7521,
+      "step": 7354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6450000000000003e-05,
+      "loss": 0.5627,
+      "step": 7355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6440000000000004e-05,
+      "loss": 0.9946,
+      "step": 7356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.643e-05,
+      "loss": 0.8687,
+      "step": 7357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.642e-05,
+      "loss": 0.9037,
+      "step": 7358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6410000000000003e-05,
+      "loss": 0.9451,
+      "step": 7359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.64e-05,
+      "loss": 1.2289,
+      "step": 7360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6390000000000003e-05,
+      "loss": 1.4672,
+      "step": 7361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6379999999999998e-05,
+      "loss": 1.2397,
+      "step": 7362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.637e-05,
+      "loss": 1.0149,
+      "step": 7363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6360000000000002e-05,
+      "loss": 0.2662,
+      "step": 7364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6350000000000004e-05,
+      "loss": 0.1186,
+      "step": 7365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6340000000000002e-05,
+      "loss": 0.3938,
+      "step": 7366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.633e-05,
+      "loss": 0.064,
+      "step": 7367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.632e-05,
+      "loss": 0.8206,
+      "step": 7368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.631e-05,
+      "loss": 0.8009,
+      "step": 7369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6300000000000002e-05,
+      "loss": 1.114,
+      "step": 7370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6290000000000004e-05,
+      "loss": 0.9707,
+      "step": 7371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.628e-05,
+      "loss": 0.7793,
+      "step": 7372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.627e-05,
+      "loss": 1.2394,
+      "step": 7373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6260000000000003e-05,
+      "loss": 1.3486,
+      "step": 7374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.625e-05,
+      "loss": 0.8877,
+      "step": 7375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6240000000000003e-05,
+      "loss": 1.1424,
+      "step": 7376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6229999999999998e-05,
+      "loss": 0.3408,
+      "step": 7377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.622e-05,
+      "loss": 0.0778,
+      "step": 7378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.621e-05,
+      "loss": 0.3137,
+      "step": 7379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 0.6126,
+      "step": 7380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6190000000000005e-05,
+      "loss": 0.8577,
+      "step": 7381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.618e-05,
+      "loss": 0.8904,
+      "step": 7382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.617e-05,
+      "loss": 1.524,
+      "step": 7383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.616e-05,
+      "loss": 1.1349,
+      "step": 7384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6150000000000002e-05,
+      "loss": 1.0076,
+      "step": 7385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6140000000000004e-05,
+      "loss": 1.0222,
+      "step": 7386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.613e-05,
+      "loss": 1.1431,
+      "step": 7387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.612e-05,
+      "loss": 1.4659,
+      "step": 7388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6110000000000002e-05,
+      "loss": 1.1548,
+      "step": 7389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.61e-05,
+      "loss": 0.588,
+      "step": 7390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6090000000000003e-05,
+      "loss": 0.2387,
+      "step": 7391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6079999999999998e-05,
+      "loss": 0.0233,
+      "step": 7392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.607e-05,
+      "loss": 0.1961,
+      "step": 7393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.606e-05,
+      "loss": 0.6587,
+      "step": 7394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6050000000000003e-05,
+      "loss": 1.3383,
+      "step": 7395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6040000000000005e-05,
+      "loss": 0.8893,
+      "step": 7396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.603e-05,
+      "loss": 0.9956,
+      "step": 7397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.602e-05,
+      "loss": 1.0732,
+      "step": 7398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.601e-05,
+      "loss": 0.9664,
+      "step": 7399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.9634,
+      "step": 7400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5990000000000004e-05,
+      "loss": 0.8789,
+      "step": 7401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.598e-05,
+      "loss": 0.0452,
+      "step": 7402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.597e-05,
+      "loss": 0.3323,
+      "step": 7403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5960000000000002e-05,
+      "loss": 0.3961,
+      "step": 7404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.595e-05,
+      "loss": 0.8194,
+      "step": 7405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5940000000000002e-05,
+      "loss": 0.9411,
+      "step": 7406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5929999999999997e-05,
+      "loss": 1.2856,
+      "step": 7407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.592e-05,
+      "loss": 0.991,
+      "step": 7408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.591e-05,
+      "loss": 0.5241,
+      "step": 7409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5900000000000003e-05,
+      "loss": 1.1055,
+      "step": 7410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5890000000000005e-05,
+      "loss": 1.5491,
+      "step": 7411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.588e-05,
+      "loss": 0.8499,
+      "step": 7412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.587e-05,
+      "loss": 1.0415,
+      "step": 7413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.586e-05,
+      "loss": 1.1245,
+      "step": 7414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.585e-05,
+      "loss": 0.2768,
+      "step": 7415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5840000000000003e-05,
+      "loss": 0.105,
+      "step": 7416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.583e-05,
+      "loss": 0.0465,
+      "step": 7417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.582e-05,
+      "loss": 0.4098,
+      "step": 7418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5810000000000002e-05,
+      "loss": 0.7716,
+      "step": 7419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.58e-05,
+      "loss": 1.3988,
+      "step": 7420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5790000000000002e-05,
+      "loss": 1.2233,
+      "step": 7421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5779999999999997e-05,
+      "loss": 0.9127,
+      "step": 7422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.577e-05,
+      "loss": 0.8226,
+      "step": 7423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.576e-05,
+      "loss": 1.2025,
+      "step": 7424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5750000000000002e-05,
+      "loss": 1.327,
+      "step": 7425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5740000000000004e-05,
+      "loss": 1.0361,
+      "step": 7426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.573e-05,
+      "loss": 0.5108,
+      "step": 7427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.572e-05,
+      "loss": 0.0851,
+      "step": 7428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.571e-05,
+      "loss": 0.3036,
+      "step": 7429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.57e-05,
+      "loss": 0.3644,
+      "step": 7430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5690000000000003e-05,
+      "loss": 0.8521,
+      "step": 7431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5679999999999998e-05,
+      "loss": 0.8103,
+      "step": 7432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.567e-05,
+      "loss": 1.417,
+      "step": 7433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.566e-05,
+      "loss": 1.1449,
+      "step": 7434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5650000000000003e-05,
+      "loss": 1.3091,
+      "step": 7435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5640000000000002e-05,
+      "loss": 1.159,
+      "step": 7436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5629999999999997e-05,
+      "loss": 1.2519,
+      "step": 7437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.562e-05,
+      "loss": 1.3434,
+      "step": 7438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.561e-05,
+      "loss": 1.0046,
+      "step": 7439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 0.4893,
+      "step": 7440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5590000000000004e-05,
+      "loss": 0.2042,
+      "step": 7441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5580000000000002e-05,
+      "loss": 0.0016,
+      "step": 7442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.557e-05,
+      "loss": 0.2809,
+      "step": 7443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.556e-05,
+      "loss": 0.9489,
+      "step": 7444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.555e-05,
+      "loss": 1.1621,
+      "step": 7445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5540000000000003e-05,
+      "loss": 1.4863,
+      "step": 7446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5530000000000005e-05,
+      "loss": 1.0225,
+      "step": 7447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.552e-05,
+      "loss": 1.0562,
+      "step": 7448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.551e-05,
+      "loss": 0.7883,
+      "step": 7449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 1.1265,
+      "step": 7450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.549e-05,
+      "loss": 0.972,
+      "step": 7451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5480000000000003e-05,
+      "loss": 0.1407,
+      "step": 7452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.547e-05,
+      "loss": 0.2271,
+      "step": 7453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.546e-05,
+      "loss": 0.0494,
+      "step": 7454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5450000000000002e-05,
+      "loss": 0.7802,
+      "step": 7455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5440000000000004e-05,
+      "loss": 1.0018,
+      "step": 7456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5430000000000002e-05,
+      "loss": 1.0647,
+      "step": 7457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.542e-05,
+      "loss": 1.147,
+      "step": 7458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.541e-05,
+      "loss": 1.047,
+      "step": 7459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.54e-05,
+      "loss": 1.0053,
+      "step": 7460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5390000000000003e-05,
+      "loss": 1.0225,
+      "step": 7461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5380000000000004e-05,
+      "loss": 1.0019,
+      "step": 7462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.537e-05,
+      "loss": 1.1218,
+      "step": 7463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.536e-05,
+      "loss": 0.8052,
+      "step": 7464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5350000000000003e-05,
+      "loss": 0.2696,
+      "step": 7465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.534e-05,
+      "loss": 0.0893,
+      "step": 7466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5330000000000003e-05,
+      "loss": 0.0406,
+      "step": 7467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5319999999999998e-05,
+      "loss": 0.8427,
+      "step": 7468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.531e-05,
+      "loss": 1.0332,
+      "step": 7469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5300000000000002e-05,
+      "loss": 1.0303,
+      "step": 7470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5290000000000004e-05,
+      "loss": 1.2864,
+      "step": 7471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5280000000000005e-05,
+      "loss": 1.2284,
+      "step": 7472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.527e-05,
+      "loss": 1.2021,
+      "step": 7473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.526e-05,
+      "loss": 0.9287,
+      "step": 7474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.525e-05,
+      "loss": 1.3238,
+      "step": 7475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5240000000000002e-05,
+      "loss": 1.2455,
+      "step": 7476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5230000000000004e-05,
+      "loss": 0.0203,
+      "step": 7477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.522e-05,
+      "loss": 0.5354,
+      "step": 7478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.521e-05,
+      "loss": 0.0042,
+      "step": 7479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 0.2546,
+      "step": 7480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.519e-05,
+      "loss": 1.0332,
+      "step": 7481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5180000000000003e-05,
+      "loss": 1.3771,
+      "step": 7482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5169999999999998e-05,
+      "loss": 1.004,
+      "step": 7483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.516e-05,
+      "loss": 1.3441,
+      "step": 7484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.515e-05,
+      "loss": 1.5682,
+      "step": 7485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5140000000000003e-05,
+      "loss": 1.0538,
+      "step": 7486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5130000000000005e-05,
+      "loss": 1.1779,
+      "step": 7487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.512e-05,
+      "loss": 1.5092,
+      "step": 7488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5110000000000002e-05,
+      "loss": 0.621,
+      "step": 7489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.51e-05,
+      "loss": 0.1872,
+      "step": 7490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5090000000000002e-05,
+      "loss": 0.4854,
+      "step": 7491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5080000000000004e-05,
+      "loss": 0.3085,
+      "step": 7492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.507e-05,
+      "loss": 0.6721,
+      "step": 7493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.506e-05,
+      "loss": 0.5978,
+      "step": 7494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5050000000000002e-05,
+      "loss": 1.31,
+      "step": 7495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.504e-05,
+      "loss": 1.0845,
+      "step": 7496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5030000000000003e-05,
+      "loss": 1.0956,
+      "step": 7497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5019999999999998e-05,
+      "loss": 0.9699,
+      "step": 7498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.501e-05,
+      "loss": 1.3278,
+      "step": 7499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5e-05,
+      "loss": 1.2082,
+      "step": 7500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4990000000000003e-05,
+      "loss": 0.6983,
+      "step": 7501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.498e-05,
+      "loss": 0.2855,
+      "step": 7502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4970000000000003e-05,
+      "loss": 0.0172,
+      "step": 7503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.496e-05,
+      "loss": 0.0198,
+      "step": 7504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.495e-05,
+      "loss": 0.3103,
+      "step": 7505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4940000000000002e-05,
+      "loss": 1.5524,
+      "step": 7506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.493e-05,
+      "loss": 1.0885,
+      "step": 7507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4920000000000002e-05,
+      "loss": 1.0617,
+      "step": 7508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.491e-05,
+      "loss": 1.2464,
+      "step": 7509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4900000000000002e-05,
+      "loss": 1.5857,
+      "step": 7510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.489e-05,
+      "loss": 1.2418,
+      "step": 7511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.488e-05,
+      "loss": 1.3319,
+      "step": 7512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.487e-05,
+      "loss": 1.1851,
+      "step": 7513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.486e-05,
+      "loss": 0.578,
+      "step": 7514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.485e-05,
+      "loss": 0.3391,
+      "step": 7515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4840000000000003e-05,
+      "loss": 0.0861,
+      "step": 7516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.483e-05,
+      "loss": 0.08,
+      "step": 7517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4820000000000003e-05,
+      "loss": 0.7123,
+      "step": 7518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.481e-05,
+      "loss": 1.235,
+      "step": 7519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.48e-05,
+      "loss": 1.0482,
+      "step": 7520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.479e-05,
+      "loss": 1.0359,
+      "step": 7521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.478e-05,
+      "loss": 1.2446,
+      "step": 7522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4770000000000002e-05,
+      "loss": 0.9563,
+      "step": 7523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.476e-05,
+      "loss": 1.26,
+      "step": 7524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4750000000000002e-05,
+      "loss": 1.0562,
+      "step": 7525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4740000000000004e-05,
+      "loss": 0.9948,
+      "step": 7526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.473e-05,
+      "loss": 0.299,
+      "step": 7527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.472e-05,
+      "loss": 0.1757,
+      "step": 7528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.471e-05,
+      "loss": 0.0143,
+      "step": 7529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.47e-05,
+      "loss": 0.2436,
+      "step": 7530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4690000000000002e-05,
+      "loss": 0.9538,
+      "step": 7531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.468e-05,
+      "loss": 0.7716,
+      "step": 7532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4670000000000003e-05,
+      "loss": 1.0482,
+      "step": 7533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.466e-05,
+      "loss": 1.2879,
+      "step": 7534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.465e-05,
+      "loss": 1.0746,
+      "step": 7535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.464e-05,
+      "loss": 0.9683,
+      "step": 7536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.463e-05,
+      "loss": 1.0082,
+      "step": 7537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.462e-05,
+      "loss": 1.031,
+      "step": 7538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4610000000000003e-05,
+      "loss": 0.4689,
+      "step": 7539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.46e-05,
+      "loss": 0.0559,
+      "step": 7540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4590000000000003e-05,
+      "loss": 0.1045,
+      "step": 7541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4580000000000002e-05,
+      "loss": 0.2618,
+      "step": 7542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.457e-05,
+      "loss": 1.2874,
+      "step": 7543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4560000000000002e-05,
+      "loss": 1.257,
+      "step": 7544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.455e-05,
+      "loss": 0.9286,
+      "step": 7545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4540000000000002e-05,
+      "loss": 1.5087,
+      "step": 7546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.453e-05,
+      "loss": 0.9025,
+      "step": 7547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4520000000000002e-05,
+      "loss": 1.056,
+      "step": 7548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.451e-05,
+      "loss": 1.1351,
+      "step": 7549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.45e-05,
+      "loss": 1.5545,
+      "step": 7550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.449e-05,
+      "loss": 1.3792,
+      "step": 7551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.448e-05,
+      "loss": 0.4548,
+      "step": 7552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.447e-05,
+      "loss": 0.1711,
+      "step": 7553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4460000000000003e-05,
+      "loss": 0.4687,
+      "step": 7554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.445e-05,
+      "loss": 0.2748,
+      "step": 7555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4440000000000003e-05,
+      "loss": 0.7304,
+      "step": 7556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.443e-05,
+      "loss": 1.6974,
+      "step": 7557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.442e-05,
+      "loss": 0.9696,
+      "step": 7558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4410000000000002e-05,
+      "loss": 1.4566,
+      "step": 7559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.44e-05,
+      "loss": 0.9003,
+      "step": 7560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4390000000000002e-05,
+      "loss": 1.1105,
+      "step": 7561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.438e-05,
+      "loss": 0.8607,
+      "step": 7562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4370000000000002e-05,
+      "loss": 1.0462,
+      "step": 7563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4360000000000004e-05,
+      "loss": 0.7311,
+      "step": 7564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.435e-05,
+      "loss": 0.3814,
+      "step": 7565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.434e-05,
+      "loss": 0.1941,
+      "step": 7566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.433e-05,
+      "loss": 0.0161,
+      "step": 7567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.432e-05,
+      "loss": 0.4287,
+      "step": 7568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4310000000000003e-05,
+      "loss": 0.784,
+      "step": 7569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.43e-05,
+      "loss": 1.1501,
+      "step": 7570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4290000000000003e-05,
+      "loss": 0.7056,
+      "step": 7571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.428e-05,
+      "loss": 1.1388,
+      "step": 7572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.427e-05,
+      "loss": 1.1345,
+      "step": 7573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.426e-05,
+      "loss": 0.9097,
+      "step": 7574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.425e-05,
+      "loss": 1.2885,
+      "step": 7575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4240000000000002e-05,
+      "loss": 1.35,
+      "step": 7576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.423e-05,
+      "loss": 0.4393,
+      "step": 7577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4220000000000002e-05,
+      "loss": 0.0426,
+      "step": 7578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4210000000000004e-05,
+      "loss": 0.0066,
+      "step": 7579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 0.2169,
+      "step": 7580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.419e-05,
+      "loss": 0.8927,
+      "step": 7581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.418e-05,
+      "loss": 1.1053,
+      "step": 7582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.417e-05,
+      "loss": 1.0031,
+      "step": 7583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4160000000000002e-05,
+      "loss": 0.9909,
+      "step": 7584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.415e-05,
+      "loss": 1.6616,
+      "step": 7585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4140000000000003e-05,
+      "loss": 1.5363,
+      "step": 7586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.413e-05,
+      "loss": 1.214,
+      "step": 7587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.412e-05,
+      "loss": 0.9595,
+      "step": 7588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.411e-05,
+      "loss": 0.3562,
+      "step": 7589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.41e-05,
+      "loss": 0.0633,
+      "step": 7590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.409e-05,
+      "loss": 0.1651,
+      "step": 7591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.408e-05,
+      "loss": 0.1088,
+      "step": 7592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.407e-05,
+      "loss": 0.3715,
+      "step": 7593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4060000000000003e-05,
+      "loss": 1.1817,
+      "step": 7594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4050000000000002e-05,
+      "loss": 1.2084,
+      "step": 7595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.404e-05,
+      "loss": 1.1868,
+      "step": 7596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4030000000000002e-05,
+      "loss": 1.0879,
+      "step": 7597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.402e-05,
+      "loss": 1.3294,
+      "step": 7598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4010000000000002e-05,
+      "loss": 1.1185,
+      "step": 7599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4e-05,
+      "loss": 1.6464,
+      "step": 7600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3990000000000002e-05,
+      "loss": 1.0382,
+      "step": 7601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.398e-05,
+      "loss": 0.2826,
+      "step": 7602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.397e-05,
+      "loss": 0.253,
+      "step": 7603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.396e-05,
+      "loss": 0.0528,
+      "step": 7604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.395e-05,
+      "loss": 0.4938,
+      "step": 7605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.394e-05,
+      "loss": 0.7793,
+      "step": 7606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3930000000000003e-05,
+      "loss": 0.899,
+      "step": 7607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.392e-05,
+      "loss": 0.9059,
+      "step": 7608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3910000000000003e-05,
+      "loss": 0.9683,
+      "step": 7609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.39e-05,
+      "loss": 1.3959,
+      "step": 7610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.389e-05,
+      "loss": 1.0272,
+      "step": 7611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3880000000000002e-05,
+      "loss": 1.1919,
+      "step": 7612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.387e-05,
+      "loss": 1.4945,
+      "step": 7613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3860000000000002e-05,
+      "loss": 0.3531,
+      "step": 7614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.385e-05,
+      "loss": 0.0322,
+      "step": 7615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3840000000000002e-05,
+      "loss": 0.1014,
+      "step": 7616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3830000000000004e-05,
+      "loss": 0.0995,
+      "step": 7617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3820000000000002e-05,
+      "loss": 0.486,
+      "step": 7618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.381e-05,
+      "loss": 1.0067,
+      "step": 7619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.38e-05,
+      "loss": 1.1995,
+      "step": 7620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.379e-05,
+      "loss": 1.5411,
+      "step": 7621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3780000000000003e-05,
+      "loss": 0.8083,
+      "step": 7622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.377e-05,
+      "loss": 1.5488,
+      "step": 7623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3760000000000003e-05,
+      "loss": 0.6513,
+      "step": 7624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.375e-05,
+      "loss": 0.8508,
+      "step": 7625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.374e-05,
+      "loss": 1.4075,
+      "step": 7626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.373e-05,
+      "loss": 0.2581,
+      "step": 7627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.372e-05,
+      "loss": 0.3708,
+      "step": 7628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.371e-05,
+      "loss": 0.1221,
+      "step": 7629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.37e-05,
+      "loss": 0.5733,
+      "step": 7630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3690000000000002e-05,
+      "loss": 0.7504,
+      "step": 7631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3680000000000004e-05,
+      "loss": 1.3791,
+      "step": 7632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3670000000000002e-05,
+      "loss": 1.3356,
+      "step": 7633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.366e-05,
+      "loss": 0.9948,
+      "step": 7634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.365e-05,
+      "loss": 0.9686,
+      "step": 7635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.364e-05,
+      "loss": 1.2193,
+      "step": 7636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3630000000000002e-05,
+      "loss": 0.9676,
+      "step": 7637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.362e-05,
+      "loss": 1.1131,
+      "step": 7638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3610000000000003e-05,
+      "loss": 1.2057,
+      "step": 7639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.36e-05,
+      "loss": 0.4204,
+      "step": 7640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.359e-05,
+      "loss": 0.0956,
+      "step": 7641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.358e-05,
+      "loss": 0.1559,
+      "step": 7642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.357e-05,
+      "loss": 0.88,
+      "step": 7643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.356e-05,
+      "loss": 1.1144,
+      "step": 7644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.355e-05,
+      "loss": 1.1534,
+      "step": 7645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.354e-05,
+      "loss": 1.3974,
+      "step": 7646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3530000000000003e-05,
+      "loss": 1.1662,
+      "step": 7647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3520000000000002e-05,
+      "loss": 1.2189,
+      "step": 7648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.351e-05,
+      "loss": 1.0821,
+      "step": 7649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.35e-05,
+      "loss": 2.0112,
+      "step": 7650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.349e-05,
+      "loss": 0.6774,
+      "step": 7651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3480000000000002e-05,
+      "loss": 0.2041,
+      "step": 7652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.347e-05,
+      "loss": 0.0583,
+      "step": 7653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3460000000000002e-05,
+      "loss": 0.3229,
+      "step": 7654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.345e-05,
+      "loss": 0.731,
+      "step": 7655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.344e-05,
+      "loss": 1.1811,
+      "step": 7656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.343e-05,
+      "loss": 1.3704,
+      "step": 7657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.342e-05,
+      "loss": 1.2822,
+      "step": 7658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.341e-05,
+      "loss": 1.2782,
+      "step": 7659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 0.6511,
+      "step": 7660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.339e-05,
+      "loss": 0.712,
+      "step": 7661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3380000000000003e-05,
+      "loss": 1.1842,
+      "step": 7662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.337e-05,
+      "loss": 1.4497,
+      "step": 7663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.336e-05,
+      "loss": 1.0325,
+      "step": 7664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3350000000000002e-05,
+      "loss": 0.1188,
+      "step": 7665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.334e-05,
+      "loss": 0.0183,
+      "step": 7666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3330000000000002e-05,
+      "loss": 0.2689,
+      "step": 7667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.332e-05,
+      "loss": 0.7192,
+      "step": 7668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3310000000000002e-05,
+      "loss": 1.1292,
+      "step": 7669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3300000000000004e-05,
+      "loss": 1.2732,
+      "step": 7670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3290000000000002e-05,
+      "loss": 1.3762,
+      "step": 7671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.328e-05,
+      "loss": 0.8826,
+      "step": 7672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.327e-05,
+      "loss": 1.2505,
+      "step": 7673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.326e-05,
+      "loss": 1.2825,
+      "step": 7674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3250000000000003e-05,
+      "loss": 1.3546,
+      "step": 7675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.324e-05,
+      "loss": 1.284,
+      "step": 7676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3230000000000003e-05,
+      "loss": 0.3908,
+      "step": 7677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.322e-05,
+      "loss": 0.1205,
+      "step": 7678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.321e-05,
+      "loss": 0.0818,
+      "step": 7679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.32e-05,
+      "loss": 0.3033,
+      "step": 7680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.319e-05,
+      "loss": 0.1584,
+      "step": 7681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.318e-05,
+      "loss": 0.9809,
+      "step": 7682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.317e-05,
+      "loss": 1.6604,
+      "step": 7683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3160000000000002e-05,
+      "loss": 0.9775,
+      "step": 7684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3150000000000004e-05,
+      "loss": 1.0758,
+      "step": 7685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3140000000000002e-05,
+      "loss": 1.4928,
+      "step": 7686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.313e-05,
+      "loss": 1.4508,
+      "step": 7687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.312e-05,
+      "loss": 1.6192,
+      "step": 7688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.311e-05,
+      "loss": 0.5518,
+      "step": 7689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 0.0331,
+      "step": 7690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.309e-05,
+      "loss": 0.4869,
+      "step": 7691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3080000000000003e-05,
+      "loss": 0.1335,
+      "step": 7692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.307e-05,
+      "loss": 0.3317,
+      "step": 7693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.306e-05,
+      "loss": 1.0476,
+      "step": 7694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.305e-05,
+      "loss": 1.1566,
+      "step": 7695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.304e-05,
+      "loss": 1.0949,
+      "step": 7696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.303e-05,
+      "loss": 1.2439,
+      "step": 7697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.302e-05,
+      "loss": 0.8748,
+      "step": 7698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.301e-05,
+      "loss": 0.9469,
+      "step": 7699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.5421,
+      "step": 7700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2990000000000002e-05,
+      "loss": 1.3138,
+      "step": 7701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.298e-05,
+      "loss": 0.1989,
+      "step": 7702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.297e-05,
+      "loss": 0.3405,
+      "step": 7703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.296e-05,
+      "loss": 0.1906,
+      "step": 7704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2950000000000002e-05,
+      "loss": 0.3105,
+      "step": 7705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.294e-05,
+      "loss": 1.1425,
+      "step": 7706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2930000000000002e-05,
+      "loss": 1.7144,
+      "step": 7707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.292e-05,
+      "loss": 1.0818,
+      "step": 7708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2910000000000003e-05,
+      "loss": 0.7426,
+      "step": 7709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.29e-05,
+      "loss": 0.9465,
+      "step": 7710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.289e-05,
+      "loss": 1.075,
+      "step": 7711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.288e-05,
+      "loss": 0.6929,
+      "step": 7712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.287e-05,
+      "loss": 1.3863,
+      "step": 7713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.286e-05,
+      "loss": 1.2844,
+      "step": 7714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2850000000000003e-05,
+      "loss": 0.4048,
+      "step": 7715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.284e-05,
+      "loss": 0.067,
+      "step": 7716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.283e-05,
+      "loss": 0.1413,
+      "step": 7717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.282e-05,
+      "loss": 0.0799,
+      "step": 7718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.281e-05,
+      "loss": 0.3872,
+      "step": 7719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 0.8442,
+      "step": 7720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.279e-05,
+      "loss": 1.1289,
+      "step": 7721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2780000000000002e-05,
+      "loss": 1.4018,
+      "step": 7722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2770000000000004e-05,
+      "loss": 1.1321,
+      "step": 7723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2760000000000002e-05,
+      "loss": 0.7866,
+      "step": 7724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.275e-05,
+      "loss": 1.3374,
+      "step": 7725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.274e-05,
+      "loss": 1.3374,
+      "step": 7726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.273e-05,
+      "loss": 0.4856,
+      "step": 7727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2720000000000003e-05,
+      "loss": 0.2046,
+      "step": 7728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.271e-05,
+      "loss": 0.041,
+      "step": 7729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2700000000000003e-05,
+      "loss": 0.0594,
+      "step": 7730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.269e-05,
+      "loss": 0.6982,
+      "step": 7731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.268e-05,
+      "loss": 1.3237,
+      "step": 7732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.267e-05,
+      "loss": 1.4018,
+      "step": 7733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.266e-05,
+      "loss": 1.3405,
+      "step": 7734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.265e-05,
+      "loss": 1.3107,
+      "step": 7735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.264e-05,
+      "loss": 1.2991,
+      "step": 7736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2630000000000002e-05,
+      "loss": 1.0166,
+      "step": 7737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2620000000000004e-05,
+      "loss": 1.1543,
+      "step": 7738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2610000000000002e-05,
+      "loss": 0.3821,
+      "step": 7739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.26e-05,
+      "loss": 0.1431,
+      "step": 7740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.259e-05,
+      "loss": 0.0334,
+      "step": 7741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.258e-05,
+      "loss": 0.1415,
+      "step": 7742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2570000000000002e-05,
+      "loss": 0.7102,
+      "step": 7743
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.256e-05,
+      "loss": 0.7942,
+      "step": 7744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2550000000000003e-05,
+      "loss": 1.6785,
+      "step": 7745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.254e-05,
+      "loss": 1.2555,
+      "step": 7746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.253e-05,
+      "loss": 0.8742,
+      "step": 7747
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.252e-05,
+      "loss": 1.7663,
+      "step": 7748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.251e-05,
+      "loss": 1.2358,
+      "step": 7749
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.25e-05,
+      "loss": 1.5062,
+      "step": 7750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.249e-05,
+      "loss": 1.123,
+      "step": 7751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.248e-05,
+      "loss": 0.1552,
+      "step": 7752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2470000000000003e-05,
+      "loss": 0.1584,
+      "step": 7753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2460000000000002e-05,
+      "loss": 0.1409,
+      "step": 7754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.245e-05,
+      "loss": 0.0652,
+      "step": 7755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.244e-05,
+      "loss": 0.8416,
+      "step": 7756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.243e-05,
+      "loss": 1.0176,
+      "step": 7757
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2420000000000002e-05,
+      "loss": 1.1851,
+      "step": 7758
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.241e-05,
+      "loss": 1.1195,
+      "step": 7759
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 1.1753,
+      "step": 7760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.239e-05,
+      "loss": 0.9241,
+      "step": 7761
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2380000000000003e-05,
+      "loss": 0.9936,
+      "step": 7762
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.237e-05,
+      "loss": 1.5016,
+      "step": 7763
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.236e-05,
+      "loss": 0.5575,
+      "step": 7764
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.235e-05,
+      "loss": 0.4634,
+      "step": 7765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.234e-05,
+      "loss": 0.0007,
+      "step": 7766
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.233e-05,
+      "loss": 0.0037,
+      "step": 7767
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2320000000000003e-05,
+      "loss": 0.4304,
+      "step": 7768
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.231e-05,
+      "loss": 1.063,
+      "step": 7769
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.23e-05,
+      "loss": 1.6595,
+      "step": 7770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.229e-05,
+      "loss": 1.3452,
+      "step": 7771
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.228e-05,
+      "loss": 1.1228,
+      "step": 7772
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2270000000000002e-05,
+      "loss": 0.7959,
+      "step": 7773
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.226e-05,
+      "loss": 1.134,
+      "step": 7774
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2250000000000002e-05,
+      "loss": 0.987,
+      "step": 7775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.224e-05,
+      "loss": 1.4255,
+      "step": 7776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2230000000000002e-05,
+      "loss": 0.2176,
+      "step": 7777
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.222e-05,
+      "loss": 0.1123,
+      "step": 7778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.221e-05,
+      "loss": 0.1768,
+      "step": 7779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.22e-05,
+      "loss": 0.2928,
+      "step": 7780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.219e-05,
+      "loss": 0.7354,
+      "step": 7781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.218e-05,
+      "loss": 1.3124,
+      "step": 7782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2170000000000003e-05,
+      "loss": 1.4664,
+      "step": 7783
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.216e-05,
+      "loss": 1.286,
+      "step": 7784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.215e-05,
+      "loss": 0.9727,
+      "step": 7785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.214e-05,
+      "loss": 1.3684,
+      "step": 7786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.213e-05,
+      "loss": 1.5913,
+      "step": 7787
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.212e-05,
+      "loss": 1.4555,
+      "step": 7788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.211e-05,
+      "loss": 0.6302,
+      "step": 7789
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2100000000000002e-05,
+      "loss": 0.0129,
+      "step": 7790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2090000000000004e-05,
+      "loss": 0.0314,
+      "step": 7791
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2080000000000002e-05,
+      "loss": 0.2365,
+      "step": 7792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.207e-05,
+      "loss": 0.844,
+      "step": 7793
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.206e-05,
+      "loss": 1.125,
+      "step": 7794
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.205e-05,
+      "loss": 1.1924,
+      "step": 7795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2040000000000002e-05,
+      "loss": 1.2231,
+      "step": 7796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.203e-05,
+      "loss": 1.3758,
+      "step": 7797
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2020000000000003e-05,
+      "loss": 1.2892,
+      "step": 7798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.201e-05,
+      "loss": 1.2098,
+      "step": 7799
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.9364,
+      "step": 7800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.199e-05,
+      "loss": 1.0368,
+      "step": 7801
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.198e-05,
+      "loss": 0.4575,
+      "step": 7802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.197e-05,
+      "loss": 0.0762,
+      "step": 7803
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.196e-05,
+      "loss": 0.1984,
+      "step": 7804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.195e-05,
+      "loss": 0.3365,
+      "step": 7805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1940000000000003e-05,
+      "loss": 1.434,
+      "step": 7806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1930000000000002e-05,
+      "loss": 1.1557,
+      "step": 7807
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.192e-05,
+      "loss": 1.1763,
+      "step": 7808
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.191e-05,
+      "loss": 1.0179,
+      "step": 7809
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.19e-05,
+      "loss": 1.3928,
+      "step": 7810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1890000000000002e-05,
+      "loss": 1.209,
+      "step": 7811
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188e-05,
+      "loss": 1.3377,
+      "step": 7812
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1870000000000002e-05,
+      "loss": 1.5082,
+      "step": 7813
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186e-05,
+      "loss": 0.3964,
+      "step": 7814
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1850000000000003e-05,
+      "loss": 0.4186,
+      "step": 7815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184e-05,
+      "loss": 0.0155,
+      "step": 7816
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.183e-05,
+      "loss": 0.5128,
+      "step": 7817
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.182e-05,
+      "loss": 0.8642,
+      "step": 7818
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.181e-05,
+      "loss": 1.0618,
+      "step": 7819
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18e-05,
+      "loss": 0.8475,
+      "step": 7820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1790000000000003e-05,
+      "loss": 1.2601,
+      "step": 7821
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.178e-05,
+      "loss": 0.7301,
+      "step": 7822
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.177e-05,
+      "loss": 0.9512,
+      "step": 7823
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.176e-05,
+      "loss": 1.3023,
+      "step": 7824
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.175e-05,
+      "loss": 1.3302,
+      "step": 7825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1740000000000002e-05,
+      "loss": 1.0918,
+      "step": 7826
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.173e-05,
+      "loss": 0.0817,
+      "step": 7827
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1720000000000002e-05,
+      "loss": 0.2478,
+      "step": 7828
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.171e-05,
+      "loss": 0.3323,
+      "step": 7829
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1700000000000002e-05,
+      "loss": 0.763,
+      "step": 7830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.169e-05,
+      "loss": 0.7259,
+      "step": 7831
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.168e-05,
+      "loss": 1.0969,
+      "step": 7832
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.167e-05,
+      "loss": 1.2296,
+      "step": 7833
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.166e-05,
+      "loss": 0.8712,
+      "step": 7834
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.165e-05,
+      "loss": 1.3386,
+      "step": 7835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1640000000000003e-05,
+      "loss": 0.772,
+      "step": 7836
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.163e-05,
+      "loss": 1.0529,
+      "step": 7837
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.162e-05,
+      "loss": 1.1269,
+      "step": 7838
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1609999999999998e-05,
+      "loss": 0.95,
+      "step": 7839
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.16e-05,
+      "loss": 0.052,
+      "step": 7840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.159e-05,
+      "loss": 0.598,
+      "step": 7841
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.158e-05,
+      "loss": 0.0422,
+      "step": 7842
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1570000000000002e-05,
+      "loss": 0.5798,
+      "step": 7843
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1560000000000004e-05,
+      "loss": 1.2635,
+      "step": 7844
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1550000000000002e-05,
+      "loss": 0.9067,
+      "step": 7845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.154e-05,
+      "loss": 1.1093,
+      "step": 7846
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.153e-05,
+      "loss": 0.9889,
+      "step": 7847
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.152e-05,
+      "loss": 0.9853,
+      "step": 7848
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1510000000000002e-05,
+      "loss": 0.997,
+      "step": 7849
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.15e-05,
+      "loss": 1.1341,
+      "step": 7850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1490000000000003e-05,
+      "loss": 0.9365,
+      "step": 7851
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.148e-05,
+      "loss": 0.2147,
+      "step": 7852
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1470000000000003e-05,
+      "loss": 0.0763,
+      "step": 7853
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.146e-05,
+      "loss": 0.1601,
+      "step": 7854
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.145e-05,
+      "loss": 0.5077,
+      "step": 7855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.144e-05,
+      "loss": 1.461,
+      "step": 7856
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.143e-05,
+      "loss": 1.1985,
+      "step": 7857
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.142e-05,
+      "loss": 1.2531,
+      "step": 7858
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1410000000000003e-05,
+      "loss": 1.0342,
+      "step": 7859
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 0.7456,
+      "step": 7860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.139e-05,
+      "loss": 0.9838,
+      "step": 7861
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.138e-05,
+      "loss": 1.3527,
+      "step": 7862
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.137e-05,
+      "loss": 1.2058,
+      "step": 7863
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1360000000000002e-05,
+      "loss": 1.0724,
+      "step": 7864
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.135e-05,
+      "loss": 0.3732,
+      "step": 7865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1340000000000002e-05,
+      "loss": 0.161,
+      "step": 7866
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.133e-05,
+      "loss": 0.154,
+      "step": 7867
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1320000000000003e-05,
+      "loss": 0.5675,
+      "step": 7868
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.131e-05,
+      "loss": 0.8929,
+      "step": 7869
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.13e-05,
+      "loss": 0.6952,
+      "step": 7870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.129e-05,
+      "loss": 1.152,
+      "step": 7871
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.128e-05,
+      "loss": 1.2245,
+      "step": 7872
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.127e-05,
+      "loss": 1.3721,
+      "step": 7873
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1260000000000003e-05,
+      "loss": 1.0534,
+      "step": 7874
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.125e-05,
+      "loss": 1.7878,
+      "step": 7875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.124e-05,
+      "loss": 1.2336,
+      "step": 7876
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.123e-05,
+      "loss": 0.4058,
+      "step": 7877
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.122e-05,
+      "loss": 0.1083,
+      "step": 7878
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1210000000000002e-05,
+      "loss": 0.1423,
+      "step": 7879
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.12e-05,
+      "loss": 0.249,
+      "step": 7880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1190000000000002e-05,
+      "loss": 1.416,
+      "step": 7881
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.118e-05,
+      "loss": 1.0687,
+      "step": 7882
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1170000000000002e-05,
+      "loss": 1.0323,
+      "step": 7883
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.116e-05,
+      "loss": 1.3398,
+      "step": 7884
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.115e-05,
+      "loss": 1.1743,
+      "step": 7885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.114e-05,
+      "loss": 1.2396,
+      "step": 7886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.113e-05,
+      "loss": 0.9176,
+      "step": 7887
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.112e-05,
+      "loss": 1.085,
+      "step": 7888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1110000000000003e-05,
+      "loss": 1.3231,
+      "step": 7889
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.11e-05,
+      "loss": 0.1972,
+      "step": 7890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1090000000000003e-05,
+      "loss": 0.1203,
+      "step": 7891
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1079999999999998e-05,
+      "loss": 0.2864,
+      "step": 7892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.107e-05,
+      "loss": 1.1539,
+      "step": 7893
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.106e-05,
+      "loss": 1.5327,
+      "step": 7894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.105e-05,
+      "loss": 1.1598,
+      "step": 7895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1040000000000002e-05,
+      "loss": 1.1368,
+      "step": 7896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.103e-05,
+      "loss": 1.0709,
+      "step": 7897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1020000000000002e-05,
+      "loss": 1.6264,
+      "step": 7898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.101e-05,
+      "loss": 0.9449,
+      "step": 7899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1e-05,
+      "loss": 0.9582,
+      "step": 7900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.099e-05,
+      "loss": 1.0327,
+      "step": 7901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.098e-05,
+      "loss": 0.0905,
+      "step": 7902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.097e-05,
+      "loss": 0.0994,
+      "step": 7903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0960000000000003e-05,
+      "loss": 0.0369,
+      "step": 7904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.095e-05,
+      "loss": 0.4701,
+      "step": 7905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0940000000000003e-05,
+      "loss": 1.1288,
+      "step": 7906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.093e-05,
+      "loss": 0.9285,
+      "step": 7907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.092e-05,
+      "loss": 0.9314,
+      "step": 7908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.091e-05,
+      "loss": 1.1007,
+      "step": 7909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.09e-05,
+      "loss": 1.1889,
+      "step": 7910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.089e-05,
+      "loss": 1.078,
+      "step": 7911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0880000000000003e-05,
+      "loss": 0.8888,
+      "step": 7912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0870000000000002e-05,
+      "loss": 1.8117,
+      "step": 7913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.086e-05,
+      "loss": 0.4922,
+      "step": 7914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.085e-05,
+      "loss": 0.3637,
+      "step": 7915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.084e-05,
+      "loss": 0.0706,
+      "step": 7916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0830000000000002e-05,
+      "loss": 0.0128,
+      "step": 7917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.082e-05,
+      "loss": 0.1223,
+      "step": 7918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0810000000000002e-05,
+      "loss": 0.7099,
+      "step": 7919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.08e-05,
+      "loss": 0.9504,
+      "step": 7920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0790000000000003e-05,
+      "loss": 1.2222,
+      "step": 7921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.078e-05,
+      "loss": 1.2135,
+      "step": 7922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.077e-05,
+      "loss": 1.2907,
+      "step": 7923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.076e-05,
+      "loss": 1.0982,
+      "step": 7924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.075e-05,
+      "loss": 1.7289,
+      "step": 7925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.074e-05,
+      "loss": 1.4176,
+      "step": 7926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0730000000000003e-05,
+      "loss": 0.2246,
+      "step": 7927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.072e-05,
+      "loss": 0.0317,
+      "step": 7928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0710000000000003e-05,
+      "loss": 0.0156,
+      "step": 7929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.07e-05,
+      "loss": 0.3993,
+      "step": 7930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.069e-05,
+      "loss": 0.8949,
+      "step": 7931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0680000000000002e-05,
+      "loss": 1.4419,
+      "step": 7932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.067e-05,
+      "loss": 1.3265,
+      "step": 7933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0660000000000002e-05,
+      "loss": 1.2423,
+      "step": 7934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.065e-05,
+      "loss": 0.7591,
+      "step": 7935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0640000000000002e-05,
+      "loss": 1.8491,
+      "step": 7936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.063e-05,
+      "loss": 1.4896,
+      "step": 7937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.062e-05,
+      "loss": 1.1884,
+      "step": 7938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.061e-05,
+      "loss": 0.6709,
+      "step": 7939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.06e-05,
+      "loss": 0.0761,
+      "step": 7940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.059e-05,
+      "loss": 0.2304,
+      "step": 7941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 0.1634,
+      "step": 7942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.057e-05,
+      "loss": 0.1624,
+      "step": 7943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0560000000000003e-05,
+      "loss": 1.2411,
+      "step": 7944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.055e-05,
+      "loss": 0.8815,
+      "step": 7945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.054e-05,
+      "loss": 1.1751,
+      "step": 7946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.053e-05,
+      "loss": 0.992,
+      "step": 7947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.052e-05,
+      "loss": 0.898,
+      "step": 7948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0510000000000002e-05,
+      "loss": 0.86,
+      "step": 7949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.05e-05,
+      "loss": 0.8388,
+      "step": 7950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0490000000000002e-05,
+      "loss": 1.0861,
+      "step": 7951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.048e-05,
+      "loss": 0.3206,
+      "step": 7952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.047e-05,
+      "loss": 0.0667,
+      "step": 7953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.046e-05,
+      "loss": 0.0791,
+      "step": 7954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.045e-05,
+      "loss": 0.1448,
+      "step": 7955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.044e-05,
+      "loss": 0.7664,
+      "step": 7956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0430000000000003e-05,
+      "loss": 1.2411,
+      "step": 7957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.042e-05,
+      "loss": 1.0645,
+      "step": 7958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0410000000000003e-05,
+      "loss": 1.3045,
+      "step": 7959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.04e-05,
+      "loss": 1.2602,
+      "step": 7960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.039e-05,
+      "loss": 0.7824,
+      "step": 7961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.038e-05,
+      "loss": 1.5567,
+      "step": 7962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.037e-05,
+      "loss": 1.473,
+      "step": 7963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.036e-05,
+      "loss": 1.0373,
+      "step": 7964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.035e-05,
+      "loss": 0.111,
+      "step": 7965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0340000000000002e-05,
+      "loss": 0.1571,
+      "step": 7966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.033e-05,
+      "loss": 0.0499,
+      "step": 7967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.032e-05,
+      "loss": 0.1728,
+      "step": 7968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.031e-05,
+      "loss": 0.7748,
+      "step": 7969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0300000000000002e-05,
+      "loss": 1.3865,
+      "step": 7970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.029e-05,
+      "loss": 0.9329,
+      "step": 7971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0280000000000002e-05,
+      "loss": 1.0972,
+      "step": 7972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.027e-05,
+      "loss": 1.4543,
+      "step": 7973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0260000000000003e-05,
+      "loss": 1.0765,
+      "step": 7974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.025e-05,
+      "loss": 1.7112,
+      "step": 7975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.024e-05,
+      "loss": 1.2713,
+      "step": 7976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.023e-05,
+      "loss": 0.0806,
+      "step": 7977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.022e-05,
+      "loss": 0.1682,
+      "step": 7978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.021e-05,
+      "loss": 0.0315,
+      "step": 7979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 0.0158,
+      "step": 7980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.019e-05,
+      "loss": 0.8825,
+      "step": 7981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0180000000000003e-05,
+      "loss": 0.926,
+      "step": 7982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.017e-05,
+      "loss": 0.9685,
+      "step": 7983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.016e-05,
+      "loss": 1.4318,
+      "step": 7984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0150000000000002e-05,
+      "loss": 0.8804,
+      "step": 7985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.014e-05,
+      "loss": 1.1087,
+      "step": 7986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0130000000000002e-05,
+      "loss": 0.9325,
+      "step": 7987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.012e-05,
+      "loss": 1.6279,
+      "step": 7988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0110000000000002e-05,
+      "loss": 0.8933,
+      "step": 7989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.01e-05,
+      "loss": 0.2152,
+      "step": 7990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.009e-05,
+      "loss": 0.0024,
+      "step": 7991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.008e-05,
+      "loss": 0.322,
+      "step": 7992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.007e-05,
+      "loss": 0.8771,
+      "step": 7993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.006e-05,
+      "loss": 1.3814,
+      "step": 7994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0050000000000003e-05,
+      "loss": 1.1387,
+      "step": 7995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.004e-05,
+      "loss": 0.875,
+      "step": 7996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0030000000000003e-05,
+      "loss": 1.0976,
+      "step": 7997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.002e-05,
+      "loss": 1.538,
+      "step": 7998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.001e-05,
+      "loss": 1.5361,
+      "step": 7999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 1.6701,
+      "step": 8000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999e-05,
+      "loss": 1.2819,
+      "step": 8001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.3236,
+      "step": 8002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997e-05,
+      "loss": 0.1548,
+      "step": 8003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9960000000000002e-05,
+      "loss": 0.2101,
+      "step": 8004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995e-05,
+      "loss": 0.5373,
+      "step": 8005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994e-05,
+      "loss": 0.7632,
+      "step": 8006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993e-05,
+      "loss": 1.4303,
+      "step": 8007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.992e-05,
+      "loss": 1.1852,
+      "step": 8008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.991e-05,
+      "loss": 1.3857,
+      "step": 8009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9900000000000003e-05,
+      "loss": 1.1794,
+      "step": 8010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989e-05,
+      "loss": 1.4342,
+      "step": 8011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880000000000003e-05,
+      "loss": 0.7602,
+      "step": 8012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987e-05,
+      "loss": 1.0636,
+      "step": 8013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986e-05,
+      "loss": 1.1876,
+      "step": 8014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985e-05,
+      "loss": 0.2701,
+      "step": 8015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984e-05,
+      "loss": 0.1371,
+      "step": 8016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.983e-05,
+      "loss": 0.0113,
+      "step": 8017
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.982e-05,
+      "loss": 0.079,
+      "step": 8018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9810000000000002e-05,
+      "loss": 0.8048,
+      "step": 8019
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 1.3688,
+      "step": 8020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.979e-05,
+      "loss": 1.2696,
+      "step": 8021
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.978e-05,
+      "loss": 1.2105,
+      "step": 8022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.977e-05,
+      "loss": 1.2128,
+      "step": 8023
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976e-05,
+      "loss": 1.0678,
+      "step": 8024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9750000000000002e-05,
+      "loss": 0.7929,
+      "step": 8025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974e-05,
+      "loss": 0.9009,
+      "step": 8026
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9730000000000003e-05,
+      "loss": 0.3227,
+      "step": 8027
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972e-05,
+      "loss": 0.0355,
+      "step": 8028
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971e-05,
+      "loss": 0.2895,
+      "step": 8029
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97e-05,
+      "loss": 0.0743,
+      "step": 8030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969e-05,
+      "loss": 1.1282,
+      "step": 8031
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968e-05,
+      "loss": 0.8247,
+      "step": 8032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9670000000000003e-05,
+      "loss": 1.2207,
+      "step": 8033
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966e-05,
+      "loss": 0.9773,
+      "step": 8034
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9650000000000003e-05,
+      "loss": 0.9723,
+      "step": 8035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9640000000000002e-05,
+      "loss": 1.2744,
+      "step": 8036
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.963e-05,
+      "loss": 0.7679,
+      "step": 8037
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9620000000000002e-05,
+      "loss": 1.3984,
+      "step": 8038
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.961e-05,
+      "loss": 1.2403,
+      "step": 8039
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 0.269,
+      "step": 8040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.959e-05,
+      "loss": 0.0025,
+      "step": 8041
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.1675,
+      "step": 8042
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.957e-05,
+      "loss": 1.0226,
+      "step": 8043
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.956e-05,
+      "loss": 1.2052,
+      "step": 8044
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.955e-05,
+      "loss": 1.5755,
+      "step": 8045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.954e-05,
+      "loss": 1.1589,
+      "step": 8046
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.953e-05,
+      "loss": 1.2662,
+      "step": 8047
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9520000000000003e-05,
+      "loss": 1.3213,
+      "step": 8048
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.951e-05,
+      "loss": 1.3447,
+      "step": 8049
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 1.5497,
+      "step": 8050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.949e-05,
+      "loss": 1.3385,
+      "step": 8051
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.948e-05,
+      "loss": 0.1172,
+      "step": 8052
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.947e-05,
+      "loss": 0.0637,
+      "step": 8053
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.946e-05,
+      "loss": 0.2201,
+      "step": 8054
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9450000000000002e-05,
+      "loss": 0.8893,
+      "step": 8055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.944e-05,
+      "loss": 0.8236,
+      "step": 8056
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9430000000000002e-05,
+      "loss": 1.2013,
+      "step": 8057
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.942e-05,
+      "loss": 0.8791,
+      "step": 8058
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.941e-05,
+      "loss": 1.1212,
+      "step": 8059
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.94e-05,
+      "loss": 1.3463,
+      "step": 8060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.939e-05,
+      "loss": 0.9283,
+      "step": 8061
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.938e-05,
+      "loss": 1.4463,
+      "step": 8062
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9370000000000003e-05,
+      "loss": 1.1103,
+      "step": 8063
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.936e-05,
+      "loss": 0.4414,
+      "step": 8064
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9350000000000003e-05,
+      "loss": 0.2095,
+      "step": 8065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.934e-05,
+      "loss": 0.0118,
+      "step": 8066
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.933e-05,
+      "loss": 0.0675,
+      "step": 8067
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.932e-05,
+      "loss": 0.8658,
+      "step": 8068
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.931e-05,
+      "loss": 1.0566,
+      "step": 8069
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.93e-05,
+      "loss": 1.3609,
+      "step": 8070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.929e-05,
+      "loss": 1.4461,
+      "step": 8071
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9280000000000002e-05,
+      "loss": 1.0974,
+      "step": 8072
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9270000000000004e-05,
+      "loss": 0.8074,
+      "step": 8073
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9260000000000002e-05,
+      "loss": 0.8141,
+      "step": 8074
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.925e-05,
+      "loss": 1.3359,
+      "step": 8075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.924e-05,
+      "loss": 1.2932,
+      "step": 8076
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.923e-05,
+      "loss": 0.4889,
+      "step": 8077
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9220000000000002e-05,
+      "loss": 0.2262,
+      "step": 8078
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.921e-05,
+      "loss": 0.0199,
+      "step": 8079
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.1895,
+      "step": 8080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.919e-05,
+      "loss": 0.6989,
+      "step": 8081
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.918e-05,
+      "loss": 0.9703,
+      "step": 8082
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.917e-05,
+      "loss": 1.4386,
+      "step": 8083
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.916e-05,
+      "loss": 1.1514,
+      "step": 8084
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.915e-05,
+      "loss": 1.0321,
+      "step": 8085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.914e-05,
+      "loss": 0.903,
+      "step": 8086
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.913e-05,
+      "loss": 1.6223,
+      "step": 8087
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9120000000000003e-05,
+      "loss": 1.3302,
+      "step": 8088
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.911e-05,
+      "loss": 0.753,
+      "step": 8089
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.91e-05,
+      "loss": 0.258,
+      "step": 8090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.909e-05,
+      "loss": 0.2943,
+      "step": 8091
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.908e-05,
+      "loss": 0.0993,
+      "step": 8092
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9070000000000002e-05,
+      "loss": 0.391,
+      "step": 8093
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.906e-05,
+      "loss": 0.7104,
+      "step": 8094
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9050000000000002e-05,
+      "loss": 0.869,
+      "step": 8095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.904e-05,
+      "loss": 1.2447,
+      "step": 8096
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.903e-05,
+      "loss": 1.1003,
+      "step": 8097
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.902e-05,
+      "loss": 1.0829,
+      "step": 8098
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.901e-05,
+      "loss": 1.0882,
+      "step": 8099
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9e-05,
+      "loss": 1.4766,
+      "step": 8100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8990000000000003e-05,
+      "loss": 1.19,
+      "step": 8101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5512,
+      "step": 8102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8970000000000003e-05,
+      "loss": 0.143,
+      "step": 8103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.896e-05,
+      "loss": 0.1717,
+      "step": 8104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.895e-05,
+      "loss": 0.1951,
+      "step": 8105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.894e-05,
+      "loss": 0.8641,
+      "step": 8106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.893e-05,
+      "loss": 1.3228,
+      "step": 8107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8920000000000002e-05,
+      "loss": 0.773,
+      "step": 8108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.891e-05,
+      "loss": 1.3673,
+      "step": 8109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 1.2614,
+      "step": 8110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8890000000000004e-05,
+      "loss": 1.1341,
+      "step": 8111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.888e-05,
+      "loss": 1.0443,
+      "step": 8112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.887e-05,
+      "loss": 1.1356,
+      "step": 8113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.886e-05,
+      "loss": 0.4631,
+      "step": 8114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.885e-05,
+      "loss": 0.2708,
+      "step": 8115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8840000000000003e-05,
+      "loss": 0.1055,
+      "step": 8116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.883e-05,
+      "loss": 0.1737,
+      "step": 8117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8820000000000003e-05,
+      "loss": 0.6103,
+      "step": 8118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.881e-05,
+      "loss": 0.6425,
+      "step": 8119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.88e-05,
+      "loss": 1.2162,
+      "step": 8120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.879e-05,
+      "loss": 0.8394,
+      "step": 8121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.878e-05,
+      "loss": 0.9064,
+      "step": 8122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.877e-05,
+      "loss": 0.9599,
+      "step": 8123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.876e-05,
+      "loss": 1.3457,
+      "step": 8124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 1.3936,
+      "step": 8125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8740000000000004e-05,
+      "loss": 1.227,
+      "step": 8126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8730000000000002e-05,
+      "loss": 0.3525,
+      "step": 8127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.872e-05,
+      "loss": 0.1291,
+      "step": 8128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.871e-05,
+      "loss": 0.2102,
+      "step": 8129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.87e-05,
+      "loss": 0.6819,
+      "step": 8130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8690000000000002e-05,
+      "loss": 1.1609,
+      "step": 8131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.868e-05,
+      "loss": 0.7743,
+      "step": 8132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8670000000000003e-05,
+      "loss": 1.0759,
+      "step": 8133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.866e-05,
+      "loss": 1.4682,
+      "step": 8134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.865e-05,
+      "loss": 0.9865,
+      "step": 8135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.864e-05,
+      "loss": 0.9315,
+      "step": 8136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.863e-05,
+      "loss": 0.5166,
+      "step": 8137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.862e-05,
+      "loss": 1.5809,
+      "step": 8138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.861e-05,
+      "loss": 0.5857,
+      "step": 8139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.86e-05,
+      "loss": 0.1032,
+      "step": 8140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8590000000000003e-05,
+      "loss": 0.3747,
+      "step": 8141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.858e-05,
+      "loss": 0.0142,
+      "step": 8142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.857e-05,
+      "loss": 1.0348,
+      "step": 8143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.856e-05,
+      "loss": 0.7819,
+      "step": 8144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.855e-05,
+      "loss": 1.297,
+      "step": 8145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8540000000000002e-05,
+      "loss": 1.1608,
+      "step": 8146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.853e-05,
+      "loss": 1.1976,
+      "step": 8147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8520000000000002e-05,
+      "loss": 1.0003,
+      "step": 8148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.851e-05,
+      "loss": 1.2315,
+      "step": 8149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.85e-05,
+      "loss": 1.1725,
+      "step": 8150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.849e-05,
+      "loss": 1.0793,
+      "step": 8151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.848e-05,
+      "loss": 0.5335,
+      "step": 8152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.847e-05,
+      "loss": 0.0359,
+      "step": 8153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.846e-05,
+      "loss": 0.3228,
+      "step": 8154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.845e-05,
+      "loss": 0.6272,
+      "step": 8155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8440000000000003e-05,
+      "loss": 0.9905,
+      "step": 8156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.843e-05,
+      "loss": 1.2967,
+      "step": 8157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.842e-05,
+      "loss": 1.0064,
+      "step": 8158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.841e-05,
+      "loss": 1.5522,
+      "step": 8159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.84e-05,
+      "loss": 1.1455,
+      "step": 8160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8390000000000002e-05,
+      "loss": 0.8451,
+      "step": 8161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.838e-05,
+      "loss": 1.0329,
+      "step": 8162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8370000000000002e-05,
+      "loss": 0.9618,
+      "step": 8163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8360000000000004e-05,
+      "loss": 0.6129,
+      "step": 8164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8350000000000002e-05,
+      "loss": 0.0185,
+      "step": 8165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.834e-05,
+      "loss": 0.1088,
+      "step": 8166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.833e-05,
+      "loss": 0.1895,
+      "step": 8167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.832e-05,
+      "loss": 0.2227,
+      "step": 8168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8310000000000003e-05,
+      "loss": 0.4752,
+      "step": 8169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.83e-05,
+      "loss": 1.1051,
+      "step": 8170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8290000000000003e-05,
+      "loss": 1.2929,
+      "step": 8171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.828e-05,
+      "loss": 1.3806,
+      "step": 8172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.827e-05,
+      "loss": 1.2748,
+      "step": 8173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.826e-05,
+      "loss": 1.4653,
+      "step": 8174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.825e-05,
+      "loss": 0.8807,
+      "step": 8175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.824e-05,
+      "loss": 1.4032,
+      "step": 8176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.823e-05,
+      "loss": 0.1827,
+      "step": 8177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8220000000000002e-05,
+      "loss": 0.1882,
+      "step": 8178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8210000000000004e-05,
+      "loss": 0.1456,
+      "step": 8179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 0.3739,
+      "step": 8180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.819e-05,
+      "loss": 0.5745,
+      "step": 8181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.818e-05,
+      "loss": 1.3601,
+      "step": 8182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.817e-05,
+      "loss": 1.1459,
+      "step": 8183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8160000000000002e-05,
+      "loss": 1.2495,
+      "step": 8184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.815e-05,
+      "loss": 0.9101,
+      "step": 8185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8140000000000003e-05,
+      "loss": 0.8928,
+      "step": 8186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.813e-05,
+      "loss": 1.0403,
+      "step": 8187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.812e-05,
+      "loss": 1.272,
+      "step": 8188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.811e-05,
+      "loss": 0.8194,
+      "step": 8189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.81e-05,
+      "loss": 0.3319,
+      "step": 8190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.809e-05,
+      "loss": 0.0156,
+      "step": 8191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.808e-05,
+      "loss": 0.1828,
+      "step": 8192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.807e-05,
+      "loss": 1.2722,
+      "step": 8193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8060000000000003e-05,
+      "loss": 0.9556,
+      "step": 8194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.805e-05,
+      "loss": 1.0092,
+      "step": 8195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.804e-05,
+      "loss": 1.2877,
+      "step": 8196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.803e-05,
+      "loss": 1.2905,
+      "step": 8197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.802e-05,
+      "loss": 1.1357,
+      "step": 8198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8010000000000002e-05,
+      "loss": 0.7452,
+      "step": 8199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8e-05,
+      "loss": 1.0,
+      "step": 8200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7990000000000002e-05,
+      "loss": 1.1024,
+      "step": 8201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.798e-05,
+      "loss": 0.4192,
+      "step": 8202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.797e-05,
+      "loss": 0.0669,
+      "step": 8203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.796e-05,
+      "loss": 0.0747,
+      "step": 8204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.795e-05,
+      "loss": 0.6913,
+      "step": 8205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.794e-05,
+      "loss": 1.2032,
+      "step": 8206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.793e-05,
+      "loss": 1.3313,
+      "step": 8207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.792e-05,
+      "loss": 1.1019,
+      "step": 8208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7910000000000003e-05,
+      "loss": 0.7857,
+      "step": 8209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.79e-05,
+      "loss": 1.411,
+      "step": 8210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.789e-05,
+      "loss": 1.3926,
+      "step": 8211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7879999999999998e-05,
+      "loss": 0.9968,
+      "step": 8212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.787e-05,
+      "loss": 1.8236,
+      "step": 8213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7860000000000002e-05,
+      "loss": 0.5736,
+      "step": 8214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.785e-05,
+      "loss": 0.0248,
+      "step": 8215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7840000000000002e-05,
+      "loss": 0.0934,
+      "step": 8216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.783e-05,
+      "loss": 0.0693,
+      "step": 8217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7820000000000002e-05,
+      "loss": 0.1997,
+      "step": 8218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.781e-05,
+      "loss": 1.3814,
+      "step": 8219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.78e-05,
+      "loss": 1.273,
+      "step": 8220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.779e-05,
+      "loss": 1.3699,
+      "step": 8221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.8254,
+      "step": 8222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.777e-05,
+      "loss": 0.7143,
+      "step": 8223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7760000000000003e-05,
+      "loss": 1.2368,
+      "step": 8224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.775e-05,
+      "loss": 0.7395,
+      "step": 8225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.774e-05,
+      "loss": 1.4847,
+      "step": 8226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.773e-05,
+      "loss": 0.1889,
+      "step": 8227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.772e-05,
+      "loss": 0.2498,
+      "step": 8228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.771e-05,
+      "loss": 0.3636,
+      "step": 8229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.77e-05,
+      "loss": 0.1568,
+      "step": 8230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7690000000000002e-05,
+      "loss": 0.885,
+      "step": 8231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7680000000000004e-05,
+      "loss": 0.9436,
+      "step": 8232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7670000000000002e-05,
+      "loss": 1.0607,
+      "step": 8233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.766e-05,
+      "loss": 0.8741,
+      "step": 8234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.765e-05,
+      "loss": 0.9831,
+      "step": 8235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.764e-05,
+      "loss": 0.8072,
+      "step": 8236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7630000000000002e-05,
+      "loss": 0.991,
+      "step": 8237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.762e-05,
+      "loss": 1.1069,
+      "step": 8238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7610000000000002e-05,
+      "loss": 0.7944,
+      "step": 8239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.76e-05,
+      "loss": 0.2322,
+      "step": 8240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.759e-05,
+      "loss": 0.1515,
+      "step": 8241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.758e-05,
+      "loss": 0.2609,
+      "step": 8242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.757e-05,
+      "loss": 0.1452,
+      "step": 8243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.756e-05,
+      "loss": 1.1007,
+      "step": 8244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.755e-05,
+      "loss": 1.2486,
+      "step": 8245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.754e-05,
+      "loss": 1.173,
+      "step": 8246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7530000000000003e-05,
+      "loss": 0.96,
+      "step": 8247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.752e-05,
+      "loss": 1.4033,
+      "step": 8248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.751e-05,
+      "loss": 0.9006,
+      "step": 8249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.75e-05,
+      "loss": 1.4352,
+      "step": 8250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.749e-05,
+      "loss": 1.4348,
+      "step": 8251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7480000000000002e-05,
+      "loss": 0.1623,
+      "step": 8252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.747e-05,
+      "loss": 0.0655,
+      "step": 8253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7460000000000002e-05,
+      "loss": 0.362,
+      "step": 8254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.745e-05,
+      "loss": 1.1533,
+      "step": 8255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7440000000000002e-05,
+      "loss": 0.9135,
+      "step": 8256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.743e-05,
+      "loss": 1.1991,
+      "step": 8257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.742e-05,
+      "loss": 0.9052,
+      "step": 8258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.741e-05,
+      "loss": 0.94,
+      "step": 8259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.74e-05,
+      "loss": 1.2935,
+      "step": 8260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.739e-05,
+      "loss": 1.1119,
+      "step": 8261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.9696,
+      "step": 8262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.737e-05,
+      "loss": 1.9964,
+      "step": 8263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.736e-05,
+      "loss": 0.7979,
+      "step": 8264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7349999999999998e-05,
+      "loss": 0.2856,
+      "step": 8265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.734e-05,
+      "loss": 0.0706,
+      "step": 8266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7330000000000002e-05,
+      "loss": 0.2544,
+      "step": 8267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.732e-05,
+      "loss": 0.9011,
+      "step": 8268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7310000000000002e-05,
+      "loss": 1.0562,
+      "step": 8269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.73e-05,
+      "loss": 0.949,
+      "step": 8270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7290000000000002e-05,
+      "loss": 1.4612,
+      "step": 8271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.728e-05,
+      "loss": 1.2484,
+      "step": 8272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.727e-05,
+      "loss": 1.63,
+      "step": 8273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.726e-05,
+      "loss": 1.0887,
+      "step": 8274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.725e-05,
+      "loss": 0.732,
+      "step": 8275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.724e-05,
+      "loss": 1.3973,
+      "step": 8276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7230000000000003e-05,
+      "loss": 0.2763,
+      "step": 8277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.722e-05,
+      "loss": 0.0237,
+      "step": 8278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.721e-05,
+      "loss": 0.0622,
+      "step": 8279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 0.255,
+      "step": 8280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.719e-05,
+      "loss": 0.5132,
+      "step": 8281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.718e-05,
+      "loss": 1.2124,
+      "step": 8282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.717e-05,
+      "loss": 0.689,
+      "step": 8283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7160000000000002e-05,
+      "loss": 0.992,
+      "step": 8284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7150000000000004e-05,
+      "loss": 1.1898,
+      "step": 8285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7140000000000002e-05,
+      "loss": 1.2461,
+      "step": 8286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.713e-05,
+      "loss": 1.2926,
+      "step": 8287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.712e-05,
+      "loss": 1.5951,
+      "step": 8288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.711e-05,
+      "loss": 0.9665,
+      "step": 8289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7100000000000002e-05,
+      "loss": 0.5504,
+      "step": 8290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.709e-05,
+      "loss": 0.0018,
+      "step": 8291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7080000000000002e-05,
+      "loss": 0.2291,
+      "step": 8292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.707e-05,
+      "loss": 0.9972,
+      "step": 8293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.706e-05,
+      "loss": 1.1932,
+      "step": 8294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.705e-05,
+      "loss": 0.9545,
+      "step": 8295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.704e-05,
+      "loss": 1.5476,
+      "step": 8296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.703e-05,
+      "loss": 0.882,
+      "step": 8297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.702e-05,
+      "loss": 1.1925,
+      "step": 8298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.701e-05,
+      "loss": 1.048,
+      "step": 8299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.717,
+      "step": 8300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.699e-05,
+      "loss": 1.2267,
+      "step": 8301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.698e-05,
+      "loss": 0.4618,
+      "step": 8302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.697e-05,
+      "loss": 0.2114,
+      "step": 8303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.696e-05,
+      "loss": 0.491,
+      "step": 8304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6950000000000002e-05,
+      "loss": 0.6063,
+      "step": 8305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.694e-05,
+      "loss": 1.2543,
+      "step": 8306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6930000000000002e-05,
+      "loss": 1.2415,
+      "step": 8307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.692e-05,
+      "loss": 1.2075,
+      "step": 8308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6910000000000002e-05,
+      "loss": 0.968,
+      "step": 8309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.69e-05,
+      "loss": 0.7132,
+      "step": 8310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.689e-05,
+      "loss": 1.0503,
+      "step": 8311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.688e-05,
+      "loss": 1.0381,
+      "step": 8312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.687e-05,
+      "loss": 1.4345,
+      "step": 8313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.686e-05,
+      "loss": 0.7185,
+      "step": 8314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6850000000000003e-05,
+      "loss": 0.1828,
+      "step": 8315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.684e-05,
+      "loss": 0.004,
+      "step": 8316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.683e-05,
+      "loss": 0.1684,
+      "step": 8317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6819999999999998e-05,
+      "loss": 1.1619,
+      "step": 8318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.681e-05,
+      "loss": 0.8761,
+      "step": 8319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.5028,
+      "step": 8320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.679e-05,
+      "loss": 1.0441,
+      "step": 8321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.9107,
+      "step": 8322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.677e-05,
+      "loss": 1.3796,
+      "step": 8323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6760000000000002e-05,
+      "loss": 0.9711,
+      "step": 8324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.675e-05,
+      "loss": 0.6073,
+      "step": 8325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.674e-05,
+      "loss": 0.826,
+      "step": 8326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.673e-05,
+      "loss": 0.0468,
+      "step": 8327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.672e-05,
+      "loss": 0.0185,
+      "step": 8328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.671e-05,
+      "loss": 0.2325,
+      "step": 8329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6700000000000003e-05,
+      "loss": 0.7248,
+      "step": 8330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.669e-05,
+      "loss": 1.0039,
+      "step": 8331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.668e-05,
+      "loss": 1.2441,
+      "step": 8332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6669999999999998e-05,
+      "loss": 1.203,
+      "step": 8333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.666e-05,
+      "loss": 1.2838,
+      "step": 8334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.665e-05,
+      "loss": 1.0468,
+      "step": 8335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.664e-05,
+      "loss": 1.3774,
+      "step": 8336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6630000000000002e-05,
+      "loss": 1.0434,
+      "step": 8337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.662e-05,
+      "loss": 1.6643,
+      "step": 8338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6610000000000002e-05,
+      "loss": 0.7353,
+      "step": 8339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.66e-05,
+      "loss": 0.3715,
+      "step": 8340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.659e-05,
+      "loss": 0.1379,
+      "step": 8341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6171,
+      "step": 8342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.657e-05,
+      "loss": 0.6968,
+      "step": 8343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.656e-05,
+      "loss": 1.2801,
+      "step": 8344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6550000000000002e-05,
+      "loss": 1.1179,
+      "step": 8345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.654e-05,
+      "loss": 1.5037,
+      "step": 8346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6530000000000003e-05,
+      "loss": 0.6611,
+      "step": 8347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.652e-05,
+      "loss": 0.7043,
+      "step": 8348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.651e-05,
+      "loss": 0.6232,
+      "step": 8349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.65e-05,
+      "loss": 1.1141,
+      "step": 8350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.649e-05,
+      "loss": 1.1705,
+      "step": 8351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.648e-05,
+      "loss": 0.3256,
+      "step": 8352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6470000000000003e-05,
+      "loss": 0.2378,
+      "step": 8353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.646e-05,
+      "loss": 0.1443,
+      "step": 8354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.645e-05,
+      "loss": 0.2061,
+      "step": 8355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.644e-05,
+      "loss": 0.263,
+      "step": 8356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.643e-05,
+      "loss": 1.2403,
+      "step": 8357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6420000000000002e-05,
+      "loss": 1.0648,
+      "step": 8358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.641e-05,
+      "loss": 1.2109,
+      "step": 8359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 0.9323,
+      "step": 8360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.639e-05,
+      "loss": 1.0198,
+      "step": 8361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 1.106,
+      "step": 8362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.637e-05,
+      "loss": 1.2857,
+      "step": 8363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.636e-05,
+      "loss": 1.1605,
+      "step": 8364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.635e-05,
+      "loss": 0.1996,
+      "step": 8365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.634e-05,
+      "loss": 0.1527,
+      "step": 8366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.633e-05,
+      "loss": 0.0033,
+      "step": 8367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6320000000000003e-05,
+      "loss": 0.0256,
+      "step": 8368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.631e-05,
+      "loss": 1.1838,
+      "step": 8369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.63e-05,
+      "loss": 0.9828,
+      "step": 8370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6289999999999998e-05,
+      "loss": 1.2231,
+      "step": 8371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.628e-05,
+      "loss": 0.9531,
+      "step": 8372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6270000000000002e-05,
+      "loss": 0.9435,
+      "step": 8373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.626e-05,
+      "loss": 1.3118,
+      "step": 8374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 0.9337,
+      "step": 8375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.624e-05,
+      "loss": 1.5498,
+      "step": 8376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6230000000000002e-05,
+      "loss": 0.6892,
+      "step": 8377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.622e-05,
+      "loss": 0.3324,
+      "step": 8378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.621e-05,
+      "loss": 0.2546,
+      "step": 8379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.62e-05,
+      "loss": 0.2425,
+      "step": 8380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.619e-05,
+      "loss": 0.384,
+      "step": 8381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.618e-05,
+      "loss": 0.8854,
+      "step": 8382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6170000000000003e-05,
+      "loss": 1.1431,
+      "step": 8383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.616e-05,
+      "loss": 0.9327,
+      "step": 8384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6150000000000003e-05,
+      "loss": 0.8613,
+      "step": 8385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6139999999999998e-05,
+      "loss": 1.3858,
+      "step": 8386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.613e-05,
+      "loss": 0.9633,
+      "step": 8387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.612e-05,
+      "loss": 1.0201,
+      "step": 8388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.611e-05,
+      "loss": 0.3176,
+      "step": 8389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6100000000000002e-05,
+      "loss": 0.2349,
+      "step": 8390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.609e-05,
+      "loss": 0.0345,
+      "step": 8391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 0.076,
+      "step": 8392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.607e-05,
+      "loss": 0.4866,
+      "step": 8393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.606e-05,
+      "loss": 0.7663,
+      "step": 8394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.605e-05,
+      "loss": 1.3494,
+      "step": 8395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.604e-05,
+      "loss": 1.2003,
+      "step": 8396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.603e-05,
+      "loss": 1.2144,
+      "step": 8397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6020000000000002e-05,
+      "loss": 1.1147,
+      "step": 8398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.601e-05,
+      "loss": 0.9066,
+      "step": 8399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.947,
+      "step": 8400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.599e-05,
+      "loss": 0.8033,
+      "step": 8401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.598e-05,
+      "loss": 0.0748,
+      "step": 8402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.597e-05,
+      "loss": 0.0174,
+      "step": 8403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.596e-05,
+      "loss": 0.3409,
+      "step": 8404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.595e-05,
+      "loss": 1.078,
+      "step": 8405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.594e-05,
+      "loss": 1.2981,
+      "step": 8406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.593e-05,
+      "loss": 1.2874,
+      "step": 8407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.592e-05,
+      "loss": 1.1004,
+      "step": 8408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.591e-05,
+      "loss": 1.4034,
+      "step": 8409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.59e-05,
+      "loss": 0.8169,
+      "step": 8410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5890000000000002e-05,
+      "loss": 0.9998,
+      "step": 8411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.588e-05,
+      "loss": 1.1115,
+      "step": 8412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5870000000000002e-05,
+      "loss": 0.7404,
+      "step": 8413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.586e-05,
+      "loss": 1.0391,
+      "step": 8414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5850000000000002e-05,
+      "loss": 0.3182,
+      "step": 8415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.584e-05,
+      "loss": 0.0247,
+      "step": 8416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.583e-05,
+      "loss": 0.3761,
+      "step": 8417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.582e-05,
+      "loss": 0.3537,
+      "step": 8418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.581e-05,
+      "loss": 0.6685,
+      "step": 8419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.58e-05,
+      "loss": 1.4372,
+      "step": 8420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5790000000000003e-05,
+      "loss": 0.9035,
+      "step": 8421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.578e-05,
+      "loss": 1.1151,
+      "step": 8422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.577e-05,
+      "loss": 1.1451,
+      "step": 8423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5759999999999998e-05,
+      "loss": 1.3254,
+      "step": 8424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.575e-05,
+      "loss": 0.7098,
+      "step": 8425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5740000000000002e-05,
+      "loss": 1.0619,
+      "step": 8426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.573e-05,
+      "loss": 0.5496,
+      "step": 8427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5720000000000002e-05,
+      "loss": 0.1463,
+      "step": 8428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.571e-05,
+      "loss": 0.2713,
+      "step": 8429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5700000000000002e-05,
+      "loss": 0.0251,
+      "step": 8430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.569e-05,
+      "loss": 0.7595,
+      "step": 8431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.568e-05,
+      "loss": 1.3735,
+      "step": 8432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.567e-05,
+      "loss": 1.1531,
+      "step": 8433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.566e-05,
+      "loss": 0.9184,
+      "step": 8434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.565e-05,
+      "loss": 0.9848,
+      "step": 8435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5640000000000003e-05,
+      "loss": 1.0485,
+      "step": 8436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.563e-05,
+      "loss": 1.2093,
+      "step": 8437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5620000000000003e-05,
+      "loss": 1.4855,
+      "step": 8438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.561e-05,
+      "loss": 0.262,
+      "step": 8439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.56e-05,
+      "loss": 0.0033,
+      "step": 8440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.559e-05,
+      "loss": 0.0909,
+      "step": 8441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.558e-05,
+      "loss": 0.0359,
+      "step": 8442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5570000000000002e-05,
+      "loss": 0.2678,
+      "step": 8443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.556e-05,
+      "loss": 0.9,
+      "step": 8444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5550000000000002e-05,
+      "loss": 0.8939,
+      "step": 8445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.554e-05,
+      "loss": 1.211,
+      "step": 8446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.553e-05,
+      "loss": 1.0289,
+      "step": 8447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.552e-05,
+      "loss": 1.1521,
+      "step": 8448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.551e-05,
+      "loss": 1.0909,
+      "step": 8449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.55e-05,
+      "loss": 1.1584,
+      "step": 8450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5490000000000002e-05,
+      "loss": 1.083,
+      "step": 8451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.548e-05,
+      "loss": 0.4601,
+      "step": 8452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5470000000000003e-05,
+      "loss": 0.0474,
+      "step": 8453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.546e-05,
+      "loss": 0.3481,
+      "step": 8454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.545e-05,
+      "loss": 0.3016,
+      "step": 8455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.544e-05,
+      "loss": 1.2401,
+      "step": 8456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.543e-05,
+      "loss": 1.2014,
+      "step": 8457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.542e-05,
+      "loss": 1.1637,
+      "step": 8458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.541e-05,
+      "loss": 1.3391,
+      "step": 8459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.54e-05,
+      "loss": 1.0883,
+      "step": 8460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.539e-05,
+      "loss": 1.3719,
+      "step": 8461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.538e-05,
+      "loss": 1.2215,
+      "step": 8462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.537e-05,
+      "loss": 1.2111,
+      "step": 8463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.536e-05,
+      "loss": 0.6547,
+      "step": 8464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.535e-05,
+      "loss": 0.1965,
+      "step": 8465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5340000000000002e-05,
+      "loss": 0.029,
+      "step": 8466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.533e-05,
+      "loss": 0.0925,
+      "step": 8467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5320000000000002e-05,
+      "loss": 1.1576,
+      "step": 8468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.531e-05,
+      "loss": 0.8721,
+      "step": 8469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.53e-05,
+      "loss": 1.1854,
+      "step": 8470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.529e-05,
+      "loss": 0.7024,
+      "step": 8471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.528e-05,
+      "loss": 1.0286,
+      "step": 8472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.527e-05,
+      "loss": 1.3681,
+      "step": 8473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5260000000000003e-05,
+      "loss": 1.1818,
+      "step": 8474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.525e-05,
+      "loss": 1.3595,
+      "step": 8475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5240000000000001e-05,
+      "loss": 1.3426,
+      "step": 8476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.523e-05,
+      "loss": 0.2147,
+      "step": 8477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5220000000000002e-05,
+      "loss": 0.1016,
+      "step": 8478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5210000000000002e-05,
+      "loss": 0.0403,
+      "step": 8479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.52e-05,
+      "loss": 0.1385,
+      "step": 8480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5190000000000002e-05,
+      "loss": 0.8587,
+      "step": 8481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.518e-05,
+      "loss": 0.9554,
+      "step": 8482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.517e-05,
+      "loss": 0.9172,
+      "step": 8483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5160000000000002e-05,
+      "loss": 1.1149,
+      "step": 8484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.515e-05,
+      "loss": 1.3177,
+      "step": 8485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.514e-05,
+      "loss": 0.9928,
+      "step": 8486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5129999999999999e-05,
+      "loss": 1.094,
+      "step": 8487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 1.0713,
+      "step": 8488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5110000000000003e-05,
+      "loss": 0.5948,
+      "step": 8489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.51e-05,
+      "loss": 0.1134,
+      "step": 8490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5090000000000001e-05,
+      "loss": 0.0539,
+      "step": 8491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.508e-05,
+      "loss": 0.0094,
+      "step": 8492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5070000000000001e-05,
+      "loss": 0.2417,
+      "step": 8493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5060000000000001e-05,
+      "loss": 0.9479,
+      "step": 8494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.505e-05,
+      "loss": 1.3856,
+      "step": 8495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5040000000000002e-05,
+      "loss": 1.2397,
+      "step": 8496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.503e-05,
+      "loss": 0.9054,
+      "step": 8497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.502e-05,
+      "loss": 1.1579,
+      "step": 8498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5010000000000002e-05,
+      "loss": 1.165,
+      "step": 8499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5e-05,
+      "loss": 0.7333,
+      "step": 8500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.499e-05,
+      "loss": 1.6297,
+      "step": 8501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4979999999999999e-05,
+      "loss": 0.4035,
+      "step": 8502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.497e-05,
+      "loss": 0.0483,
+      "step": 8503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4960000000000002e-05,
+      "loss": 0.1652,
+      "step": 8504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4950000000000001e-05,
+      "loss": 0.627,
+      "step": 8505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 1.5205,
+      "step": 8506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.493e-05,
+      "loss": 0.8735,
+      "step": 8507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4920000000000001e-05,
+      "loss": 1.1595,
+      "step": 8508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4910000000000001e-05,
+      "loss": 1.0905,
+      "step": 8509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.49e-05,
+      "loss": 1.1584,
+      "step": 8510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4890000000000001e-05,
+      "loss": 0.9552,
+      "step": 8511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.488e-05,
+      "loss": 1.4242,
+      "step": 8512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.487e-05,
+      "loss": 0.6897,
+      "step": 8513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4860000000000002e-05,
+      "loss": 0.155,
+      "step": 8514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.485e-05,
+      "loss": 0.354,
+      "step": 8515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4840000000000002e-05,
+      "loss": 0.2042,
+      "step": 8516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4829999999999999e-05,
+      "loss": 0.3966,
+      "step": 8517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.482e-05,
+      "loss": 0.7209,
+      "step": 8518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4810000000000002e-05,
+      "loss": 0.9553,
+      "step": 8519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.48e-05,
+      "loss": 1.2332,
+      "step": 8520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.479e-05,
+      "loss": 0.8967,
+      "step": 8521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4779999999999999e-05,
+      "loss": 0.7492,
+      "step": 8522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4770000000000001e-05,
+      "loss": 0.7635,
+      "step": 8523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 0.9673,
+      "step": 8524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.475e-05,
+      "loss": 1.0359,
+      "step": 8525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4740000000000001e-05,
+      "loss": 1.4527,
+      "step": 8526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.473e-05,
+      "loss": 0.1744,
+      "step": 8527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.472e-05,
+      "loss": 0.0561,
+      "step": 8528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4710000000000001e-05,
+      "loss": 0.1359,
+      "step": 8529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.47e-05,
+      "loss": 0.1958,
+      "step": 8530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4690000000000002e-05,
+      "loss": 0.3849,
+      "step": 8531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4680000000000002e-05,
+      "loss": 1.1843,
+      "step": 8532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.467e-05,
+      "loss": 1.404,
+      "step": 8533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4660000000000002e-05,
+      "loss": 1.3163,
+      "step": 8534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.465e-05,
+      "loss": 1.0285,
+      "step": 8535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.464e-05,
+      "loss": 0.8174,
+      "step": 8536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4630000000000002e-05,
+      "loss": 1.4273,
+      "step": 8537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.462e-05,
+      "loss": 1.2173,
+      "step": 8538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.461e-05,
+      "loss": 0.5318,
+      "step": 8539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 0.089,
+      "step": 8540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4590000000000001e-05,
+      "loss": 0.0047,
+      "step": 8541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4580000000000003e-05,
+      "loss": 0.2951,
+      "step": 8542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4570000000000001e-05,
+      "loss": 0.3534,
+      "step": 8543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4560000000000001e-05,
+      "loss": 1.185,
+      "step": 8544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.455e-05,
+      "loss": 1.0773,
+      "step": 8545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4540000000000001e-05,
+      "loss": 1.105,
+      "step": 8546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4530000000000001e-05,
+      "loss": 1.1403,
+      "step": 8547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.452e-05,
+      "loss": 1.475,
+      "step": 8548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4510000000000002e-05,
+      "loss": 1.4945,
+      "step": 8549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.45e-05,
+      "loss": 1.4495,
+      "step": 8550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.449e-05,
+      "loss": 0.8792,
+      "step": 8551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4480000000000002e-05,
+      "loss": 0.3382,
+      "step": 8552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.447e-05,
+      "loss": 0.0908,
+      "step": 8553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4460000000000002e-05,
+      "loss": 0.1569,
+      "step": 8554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4449999999999999e-05,
+      "loss": 0.5524,
+      "step": 8555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.444e-05,
+      "loss": 0.5484,
+      "step": 8556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4430000000000002e-05,
+      "loss": 0.8149,
+      "step": 8557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4420000000000001e-05,
+      "loss": 1.2407,
+      "step": 8558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4410000000000001e-05,
+      "loss": 0.949,
+      "step": 8559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.44e-05,
+      "loss": 0.9799,
+      "step": 8560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4390000000000001e-05,
+      "loss": 1.3299,
+      "step": 8561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 1.484,
+      "step": 8562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.437e-05,
+      "loss": 1.2095,
+      "step": 8563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4360000000000001e-05,
+      "loss": 0.5785,
+      "step": 8564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.435e-05,
+      "loss": 0.2166,
+      "step": 8565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.434e-05,
+      "loss": 0.0324,
+      "step": 8566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4330000000000002e-05,
+      "loss": 0.0422,
+      "step": 8567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.432e-05,
+      "loss": 0.8208,
+      "step": 8568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4310000000000002e-05,
+      "loss": 0.8659,
+      "step": 8569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.43e-05,
+      "loss": 1.1139,
+      "step": 8570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.429e-05,
+      "loss": 1.4391,
+      "step": 8571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4280000000000002e-05,
+      "loss": 0.6262,
+      "step": 8572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.427e-05,
+      "loss": 1.0661,
+      "step": 8573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.426e-05,
+      "loss": 1.491,
+      "step": 8574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4249999999999999e-05,
+      "loss": 0.8233,
+      "step": 8575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4240000000000001e-05,
+      "loss": 1.0581,
+      "step": 8576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4230000000000001e-05,
+      "loss": 0.2126,
+      "step": 8577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.422e-05,
+      "loss": 0.3258,
+      "step": 8578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4210000000000001e-05,
+      "loss": 0.0215,
+      "step": 8579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.42e-05,
+      "loss": 0.2384,
+      "step": 8580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4190000000000001e-05,
+      "loss": 1.084,
+      "step": 8581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4180000000000001e-05,
+      "loss": 1.0898,
+      "step": 8582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.417e-05,
+      "loss": 1.0188,
+      "step": 8583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4160000000000002e-05,
+      "loss": 0.8603,
+      "step": 8584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.415e-05,
+      "loss": 1.3179,
+      "step": 8585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.414e-05,
+      "loss": 1.0964,
+      "step": 8586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4130000000000002e-05,
+      "loss": 1.4888,
+      "step": 8587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.412e-05,
+      "loss": 1.3982,
+      "step": 8588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.411e-05,
+      "loss": 0.6513,
+      "step": 8589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 0.3474,
+      "step": 8590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.409e-05,
+      "loss": 0.2257,
+      "step": 8591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.408e-05,
+      "loss": 0.023,
+      "step": 8592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4069999999999999e-05,
+      "loss": 0.3242,
+      "step": 8593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4060000000000001e-05,
+      "loss": 1.1281,
+      "step": 8594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4050000000000003e-05,
+      "loss": 1.6343,
+      "step": 8595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 1.191,
+      "step": 8596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4030000000000001e-05,
+      "loss": 1.269,
+      "step": 8597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.402e-05,
+      "loss": 1.0472,
+      "step": 8598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4010000000000001e-05,
+      "loss": 1.5165,
+      "step": 8599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.8523,
+      "step": 8600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.399e-05,
+      "loss": 1.3753,
+      "step": 8601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 0.2725,
+      "step": 8602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.397e-05,
+      "loss": 0.001,
+      "step": 8603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.396e-05,
+      "loss": 0.238,
+      "step": 8604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3950000000000002e-05,
+      "loss": 0.4992,
+      "step": 8605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.394e-05,
+      "loss": 0.5205,
+      "step": 8606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3930000000000002e-05,
+      "loss": 1.1905,
+      "step": 8607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3919999999999999e-05,
+      "loss": 1.0068,
+      "step": 8608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.391e-05,
+      "loss": 0.7764,
+      "step": 8609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3900000000000002e-05,
+      "loss": 1.3032,
+      "step": 8610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.389e-05,
+      "loss": 1.1063,
+      "step": 8611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3880000000000001e-05,
+      "loss": 0.9131,
+      "step": 8612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.387e-05,
+      "loss": 1.5581,
+      "step": 8613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 1.0659,
+      "step": 8614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3850000000000001e-05,
+      "loss": 0.101,
+      "step": 8615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.384e-05,
+      "loss": 0.3174,
+      "step": 8616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3830000000000001e-05,
+      "loss": 0.0956,
+      "step": 8617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.382e-05,
+      "loss": 0.0171,
+      "step": 8618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.381e-05,
+      "loss": 0.427,
+      "step": 8619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.4393,
+      "step": 8620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.379e-05,
+      "loss": 0.8051,
+      "step": 8621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3780000000000002e-05,
+      "loss": 1.1551,
+      "step": 8622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.377e-05,
+      "loss": 1.3998,
+      "step": 8623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.376e-05,
+      "loss": 1.0479,
+      "step": 8624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3750000000000002e-05,
+      "loss": 1.6267,
+      "step": 8625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.374e-05,
+      "loss": 1.1222,
+      "step": 8626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.373e-05,
+      "loss": 0.3713,
+      "step": 8627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3719999999999999e-05,
+      "loss": 0.0661,
+      "step": 8628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3710000000000001e-05,
+      "loss": 0.3233,
+      "step": 8629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3700000000000001e-05,
+      "loss": 0.111,
+      "step": 8630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.369e-05,
+      "loss": 0.9761,
+      "step": 8631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 0.9186,
+      "step": 8632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.367e-05,
+      "loss": 0.719,
+      "step": 8633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3660000000000001e-05,
+      "loss": 0.9656,
+      "step": 8634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3650000000000001e-05,
+      "loss": 1.0575,
+      "step": 8635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.364e-05,
+      "loss": 1.2884,
+      "step": 8636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3630000000000002e-05,
+      "loss": 1.3295,
+      "step": 8637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.362e-05,
+      "loss": 1.9169,
+      "step": 8638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.361e-05,
+      "loss": 1.297,
+      "step": 8639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.1524,
+      "step": 8640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.359e-05,
+      "loss": 0.1913,
+      "step": 8641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.358e-05,
+      "loss": 0.2053,
+      "step": 8642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3569999999999999e-05,
+      "loss": 0.6046,
+      "step": 8643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.356e-05,
+      "loss": 1.5646,
+      "step": 8644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3550000000000002e-05,
+      "loss": 1.5127,
+      "step": 8645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3539999999999999e-05,
+      "loss": 0.864,
+      "step": 8646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3530000000000001e-05,
+      "loss": 1.1602,
+      "step": 8647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.352e-05,
+      "loss": 1.2221,
+      "step": 8648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3510000000000001e-05,
+      "loss": 1.1167,
+      "step": 8649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 1.601,
+      "step": 8650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.349e-05,
+      "loss": 1.1547,
+      "step": 8651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3480000000000001e-05,
+      "loss": 0.1454,
+      "step": 8652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.347e-05,
+      "loss": 0.0025,
+      "step": 8653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.346e-05,
+      "loss": 0.299,
+      "step": 8654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3450000000000002e-05,
+      "loss": 0.3327,
+      "step": 8655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.344e-05,
+      "loss": 1.1162,
+      "step": 8656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.343e-05,
+      "loss": 1.1272,
+      "step": 8657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3420000000000002e-05,
+      "loss": 1.427,
+      "step": 8658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.341e-05,
+      "loss": 1.6227,
+      "step": 8659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 0.985,
+      "step": 8660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.339e-05,
+      "loss": 1.3897,
+      "step": 8661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.338e-05,
+      "loss": 1.506,
+      "step": 8662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3370000000000002e-05,
+      "loss": 1.9561,
+      "step": 8663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.336e-05,
+      "loss": 0.3666,
+      "step": 8664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3350000000000001e-05,
+      "loss": 0.2366,
+      "step": 8665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.334e-05,
+      "loss": 0.2504,
+      "step": 8666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3330000000000001e-05,
+      "loss": 0.0857,
+      "step": 8667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 0.09,
+      "step": 8668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.331e-05,
+      "loss": 1.1021,
+      "step": 8669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3300000000000001e-05,
+      "loss": 1.49,
+      "step": 8670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.329e-05,
+      "loss": 1.2454,
+      "step": 8671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3280000000000002e-05,
+      "loss": 1.3418,
+      "step": 8672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3270000000000002e-05,
+      "loss": 1.2335,
+      "step": 8673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.326e-05,
+      "loss": 1.3227,
+      "step": 8674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3250000000000002e-05,
+      "loss": 0.96,
+      "step": 8675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.324e-05,
+      "loss": 0.765,
+      "step": 8676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.323e-05,
+      "loss": 0.1554,
+      "step": 8677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3220000000000002e-05,
+      "loss": 0.1906,
+      "step": 8678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.321e-05,
+      "loss": 0.1036,
+      "step": 8679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.32e-05,
+      "loss": 0.094,
+      "step": 8680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3189999999999999e-05,
+      "loss": 0.7429,
+      "step": 8681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 1.2489,
+      "step": 8682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3170000000000001e-05,
+      "loss": 1.0647,
+      "step": 8683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.316e-05,
+      "loss": 0.7484,
+      "step": 8684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3150000000000001e-05,
+      "loss": 1.2443,
+      "step": 8685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.314e-05,
+      "loss": 0.8503,
+      "step": 8686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3130000000000001e-05,
+      "loss": 1.2361,
+      "step": 8687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3120000000000001e-05,
+      "loss": 1.1203,
+      "step": 8688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.311e-05,
+      "loss": 1.1099,
+      "step": 8689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3100000000000002e-05,
+      "loss": 0.3879,
+      "step": 8690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.309e-05,
+      "loss": 0.2529,
+      "step": 8691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.308e-05,
+      "loss": 0.2483,
+      "step": 8692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3070000000000002e-05,
+      "loss": 0.7346,
+      "step": 8693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.306e-05,
+      "loss": 0.6068,
+      "step": 8694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.305e-05,
+      "loss": 1.3152,
+      "step": 8695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3039999999999999e-05,
+      "loss": 1.1076,
+      "step": 8696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.303e-05,
+      "loss": 1.2773,
+      "step": 8697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3020000000000002e-05,
+      "loss": 1.2789,
+      "step": 8698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.301e-05,
+      "loss": 1.3415,
+      "step": 8699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.9392,
+      "step": 8700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.299e-05,
+      "loss": 1.7609,
+      "step": 8701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.5736,
+      "step": 8702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2970000000000001e-05,
+      "loss": 0.1253,
+      "step": 8703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.296e-05,
+      "loss": 0.0255,
+      "step": 8704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2950000000000001e-05,
+      "loss": 0.1349,
+      "step": 8705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.294e-05,
+      "loss": 0.0718,
+      "step": 8706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.293e-05,
+      "loss": 0.8483,
+      "step": 8707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2920000000000002e-05,
+      "loss": 1.2763,
+      "step": 8708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.291e-05,
+      "loss": 0.9172,
+      "step": 8709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.29e-05,
+      "loss": 1.1362,
+      "step": 8710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2889999999999999e-05,
+      "loss": 1.18,
+      "step": 8711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.288e-05,
+      "loss": 1.2803,
+      "step": 8712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2870000000000002e-05,
+      "loss": 1.3895,
+      "step": 8713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.286e-05,
+      "loss": 0.5903,
+      "step": 8714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.285e-05,
+      "loss": 0.0839,
+      "step": 8715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2839999999999999e-05,
+      "loss": 0.4144,
+      "step": 8716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.283e-05,
+      "loss": 0.5204,
+      "step": 8717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2820000000000001e-05,
+      "loss": 1.2729,
+      "step": 8718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.281e-05,
+      "loss": 1.0715,
+      "step": 8719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.9918,
+      "step": 8720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2790000000000001e-05,
+      "loss": 1.2227,
+      "step": 8721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.278e-05,
+      "loss": 0.9775,
+      "step": 8722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2770000000000001e-05,
+      "loss": 0.8321,
+      "step": 8723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.276e-05,
+      "loss": 1.1726,
+      "step": 8724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2750000000000002e-05,
+      "loss": 1.2453,
+      "step": 8725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2740000000000002e-05,
+      "loss": 1.0715,
+      "step": 8726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.273e-05,
+      "loss": 0.3339,
+      "step": 8727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2720000000000002e-05,
+      "loss": 0.1115,
+      "step": 8728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.271e-05,
+      "loss": 0.0184,
+      "step": 8729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.27e-05,
+      "loss": 0.1088,
+      "step": 8730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2690000000000002e-05,
+      "loss": 1.6469,
+      "step": 8731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.268e-05,
+      "loss": 1.3644,
+      "step": 8732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.267e-05,
+      "loss": 1.1369,
+      "step": 8733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 0.9168,
+      "step": 8734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2650000000000001e-05,
+      "loss": 1.1533,
+      "step": 8735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2640000000000003e-05,
+      "loss": 1.2413,
+      "step": 8736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.263e-05,
+      "loss": 0.6918,
+      "step": 8737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2620000000000001e-05,
+      "loss": 1.5316,
+      "step": 8738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.261e-05,
+      "loss": 0.3865,
+      "step": 8739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 0.1748,
+      "step": 8740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2590000000000001e-05,
+      "loss": 0.197,
+      "step": 8741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.258e-05,
+      "loss": 0.0218,
+      "step": 8742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2570000000000002e-05,
+      "loss": 0.1258,
+      "step": 8743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.256e-05,
+      "loss": 0.9454,
+      "step": 8744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.255e-05,
+      "loss": 0.944,
+      "step": 8745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2540000000000002e-05,
+      "loss": 1.2176,
+      "step": 8746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.253e-05,
+      "loss": 1.0133,
+      "step": 8747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.252e-05,
+      "loss": 1.1361,
+      "step": 8748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2509999999999999e-05,
+      "loss": 0.8354,
+      "step": 8749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.25e-05,
+      "loss": 0.9645,
+      "step": 8750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.249e-05,
+      "loss": 1.4013,
+      "step": 8751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.248e-05,
+      "loss": 0.1568,
+      "step": 8752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2470000000000001e-05,
+      "loss": 0.2638,
+      "step": 8753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2460000000000001e-05,
+      "loss": 0.0342,
+      "step": 8754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2450000000000001e-05,
+      "loss": 0.3448,
+      "step": 8755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.244e-05,
+      "loss": 1.167,
+      "step": 8756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.243e-05,
+      "loss": 1.4476,
+      "step": 8757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2420000000000001e-05,
+      "loss": 1.5176,
+      "step": 8758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2410000000000001e-05,
+      "loss": 1.1331,
+      "step": 8759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.24e-05,
+      "loss": 0.958,
+      "step": 8760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.239e-05,
+      "loss": 1.2558,
+      "step": 8761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.238e-05,
+      "loss": 1.2419,
+      "step": 8762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2370000000000002e-05,
+      "loss": 1.1773,
+      "step": 8763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.236e-05,
+      "loss": 0.6278,
+      "step": 8764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.235e-05,
+      "loss": 0.0309,
+      "step": 8765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.234e-05,
+      "loss": 0.1557,
+      "step": 8766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.233e-05,
+      "loss": 0.4084,
+      "step": 8767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.232e-05,
+      "loss": 0.4911,
+      "step": 8768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.231e-05,
+      "loss": 0.8444,
+      "step": 8769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.23e-05,
+      "loss": 1.1116,
+      "step": 8770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2290000000000001e-05,
+      "loss": 1.1499,
+      "step": 8771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2280000000000001e-05,
+      "loss": 0.7949,
+      "step": 8772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2270000000000001e-05,
+      "loss": 1.008,
+      "step": 8773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2260000000000001e-05,
+      "loss": 1.0043,
+      "step": 8774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.225e-05,
+      "loss": 1.4286,
+      "step": 8775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.224e-05,
+      "loss": 0.9843,
+      "step": 8776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2230000000000001e-05,
+      "loss": 0.2989,
+      "step": 8777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2220000000000002e-05,
+      "loss": 0.1814,
+      "step": 8778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.221e-05,
+      "loss": 0.1349,
+      "step": 8779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.22e-05,
+      "loss": 0.5243,
+      "step": 8780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.219e-05,
+      "loss": 0.697,
+      "step": 8781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 1.074,
+      "step": 8782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.217e-05,
+      "loss": 1.2621,
+      "step": 8783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.216e-05,
+      "loss": 1.321,
+      "step": 8784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.215e-05,
+      "loss": 1.1357,
+      "step": 8785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.214e-05,
+      "loss": 0.9902,
+      "step": 8786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.213e-05,
+      "loss": 0.8635,
+      "step": 8787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 2.2021,
+      "step": 8788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2110000000000001e-05,
+      "loss": 0.9799,
+      "step": 8789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2100000000000001e-05,
+      "loss": 0.2663,
+      "step": 8790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.209e-05,
+      "loss": 0.3051,
+      "step": 8791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2080000000000001e-05,
+      "loss": 0.2792,
+      "step": 8792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2070000000000001e-05,
+      "loss": 0.5817,
+      "step": 8793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.206e-05,
+      "loss": 1.3244,
+      "step": 8794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.205e-05,
+      "loss": 1.4533,
+      "step": 8795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.204e-05,
+      "loss": 1.1578,
+      "step": 8796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2030000000000002e-05,
+      "loss": 1.166,
+      "step": 8797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.202e-05,
+      "loss": 1.1605,
+      "step": 8798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.201e-05,
+      "loss": 0.9465,
+      "step": 8799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2e-05,
+      "loss": 1.3668,
+      "step": 8800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.199e-05,
+      "loss": 1.1491,
+      "step": 8801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.198e-05,
+      "loss": 0.4803,
+      "step": 8802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.197e-05,
+      "loss": 0.3658,
+      "step": 8803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.196e-05,
+      "loss": 0.2061,
+      "step": 8804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.195e-05,
+      "loss": 0.0077,
+      "step": 8805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 1.0201,
+      "step": 8806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1930000000000001e-05,
+      "loss": 1.1462,
+      "step": 8807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1920000000000001e-05,
+      "loss": 1.5198,
+      "step": 8808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1910000000000001e-05,
+      "loss": 1.4086,
+      "step": 8809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.19e-05,
+      "loss": 1.0506,
+      "step": 8810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1890000000000001e-05,
+      "loss": 1.2642,
+      "step": 8811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 0.9922,
+      "step": 8812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.187e-05,
+      "loss": 1.021,
+      "step": 8813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.186e-05,
+      "loss": 0.7107,
+      "step": 8814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.185e-05,
+      "loss": 0.2712,
+      "step": 8815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1840000000000002e-05,
+      "loss": 0.0148,
+      "step": 8816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.183e-05,
+      "loss": 0.2875,
+      "step": 8817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.182e-05,
+      "loss": 0.16,
+      "step": 8818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.181e-05,
+      "loss": 0.7117,
+      "step": 8819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.18e-05,
+      "loss": 1.3356,
+      "step": 8820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.179e-05,
+      "loss": 1.2215,
+      "step": 8821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.178e-05,
+      "loss": 0.884,
+      "step": 8822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.177e-05,
+      "loss": 0.9303,
+      "step": 8823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 1.2564,
+      "step": 8824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.175e-05,
+      "loss": 1.2199,
+      "step": 8825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1740000000000001e-05,
+      "loss": 1.3975,
+      "step": 8826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1730000000000001e-05,
+      "loss": 0.4336,
+      "step": 8827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.172e-05,
+      "loss": 0.2009,
+      "step": 8828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.171e-05,
+      "loss": 0.3605,
+      "step": 8829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 0.261,
+      "step": 8830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1690000000000002e-05,
+      "loss": 0.0076,
+      "step": 8831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.168e-05,
+      "loss": 0.9893,
+      "step": 8832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.167e-05,
+      "loss": 1.2663,
+      "step": 8833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.166e-05,
+      "loss": 1.1436,
+      "step": 8834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1650000000000002e-05,
+      "loss": 1.1054,
+      "step": 8835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.164e-05,
+      "loss": 1.3147,
+      "step": 8836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.163e-05,
+      "loss": 1.0484,
+      "step": 8837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.162e-05,
+      "loss": 1.5077,
+      "step": 8838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.161e-05,
+      "loss": 0.6378,
+      "step": 8839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.16e-05,
+      "loss": 0.1631,
+      "step": 8840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.159e-05,
+      "loss": 0.0823,
+      "step": 8841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.4594,
+      "step": 8842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1570000000000001e-05,
+      "loss": 1.0957,
+      "step": 8843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.156e-05,
+      "loss": 1.4969,
+      "step": 8844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1550000000000001e-05,
+      "loss": 0.8662,
+      "step": 8845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1540000000000001e-05,
+      "loss": 1.263,
+      "step": 8846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.153e-05,
+      "loss": 1.2168,
+      "step": 8847
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.152e-05,
+      "loss": 1.1056,
+      "step": 8848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.151e-05,
+      "loss": 0.8684,
+      "step": 8849
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 1.3496,
+      "step": 8850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.149e-05,
+      "loss": 0.7939,
+      "step": 8851
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.148e-05,
+      "loss": 0.0768,
+      "step": 8852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.147e-05,
+      "loss": 0.1753,
+      "step": 8853
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.146e-05,
+      "loss": 0.2409,
+      "step": 8854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.145e-05,
+      "loss": 0.3219,
+      "step": 8855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.144e-05,
+      "loss": 1.3649,
+      "step": 8856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.143e-05,
+      "loss": 1.1355,
+      "step": 8857
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.142e-05,
+      "loss": 0.9928,
+      "step": 8858
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.141e-05,
+      "loss": 1.1869,
+      "step": 8859
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 1.0131,
+      "step": 8860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1390000000000001e-05,
+      "loss": 1.1975,
+      "step": 8861
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1380000000000001e-05,
+      "loss": 0.5418,
+      "step": 8862
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.137e-05,
+      "loss": 1.8251,
+      "step": 8863
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1360000000000001e-05,
+      "loss": 0.9027,
+      "step": 8864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1350000000000001e-05,
+      "loss": 0.1923,
+      "step": 8865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.134e-05,
+      "loss": 0.0077,
+      "step": 8866
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.133e-05,
+      "loss": 0.0156,
+      "step": 8867
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.132e-05,
+      "loss": 0.6432,
+      "step": 8868
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1310000000000002e-05,
+      "loss": 0.8209,
+      "step": 8869
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.13e-05,
+      "loss": 1.1973,
+      "step": 8870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.129e-05,
+      "loss": 1.2285,
+      "step": 8871
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.128e-05,
+      "loss": 1.5391,
+      "step": 8872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.127e-05,
+      "loss": 0.9902,
+      "step": 8873
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.126e-05,
+      "loss": 0.996,
+      "step": 8874
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.125e-05,
+      "loss": 1.2465,
+      "step": 8875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.124e-05,
+      "loss": 1.4468,
+      "step": 8876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1230000000000001e-05,
+      "loss": 0.7777,
+      "step": 8877
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.122e-05,
+      "loss": 0.0094,
+      "step": 8878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1210000000000001e-05,
+      "loss": 0.0262,
+      "step": 8879
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.0334,
+      "step": 8880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1190000000000001e-05,
+      "loss": 0.565,
+      "step": 8881
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.118e-05,
+      "loss": 0.8032,
+      "step": 8882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.117e-05,
+      "loss": 1.2811,
+      "step": 8883
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1160000000000002e-05,
+      "loss": 1.1651,
+      "step": 8884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.115e-05,
+      "loss": 0.7434,
+      "step": 8885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.114e-05,
+      "loss": 1.2841,
+      "step": 8886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.113e-05,
+      "loss": 0.8164,
+      "step": 8887
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.112e-05,
+      "loss": 1.4554,
+      "step": 8888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.111e-05,
+      "loss": 0.6778,
+      "step": 8889
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.11e-05,
+      "loss": 0.709,
+      "step": 8890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.109e-05,
+      "loss": 0.0645,
+      "step": 8891
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.108e-05,
+      "loss": 0.1546,
+      "step": 8892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.107e-05,
+      "loss": 0.5588,
+      "step": 8893
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.106e-05,
+      "loss": 0.6248,
+      "step": 8894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1050000000000001e-05,
+      "loss": 1.1937,
+      "step": 8895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1040000000000001e-05,
+      "loss": 1.1693,
+      "step": 8896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.103e-05,
+      "loss": 1.0955,
+      "step": 8897
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1020000000000001e-05,
+      "loss": 1.1895,
+      "step": 8898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1010000000000001e-05,
+      "loss": 1.2459,
+      "step": 8899
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.0145,
+      "step": 8900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.099e-05,
+      "loss": 1.1469,
+      "step": 8901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.098e-05,
+      "loss": 0.3714,
+      "step": 8902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0970000000000002e-05,
+      "loss": 0.2733,
+      "step": 8903
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.096e-05,
+      "loss": 0.4814,
+      "step": 8904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.095e-05,
+      "loss": 0.5541,
+      "step": 8905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.094e-05,
+      "loss": 0.8203,
+      "step": 8906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.093e-05,
+      "loss": 1.0257,
+      "step": 8907
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.092e-05,
+      "loss": 1.0157,
+      "step": 8908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.091e-05,
+      "loss": 0.5883,
+      "step": 8909
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.09e-05,
+      "loss": 1.3595,
+      "step": 8910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.089e-05,
+      "loss": 1.2821,
+      "step": 8911
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.088e-05,
+      "loss": 0.996,
+      "step": 8912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0870000000000001e-05,
+      "loss": 1.645,
+      "step": 8913
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0860000000000001e-05,
+      "loss": 0.5686,
+      "step": 8914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0850000000000001e-05,
+      "loss": 0.3445,
+      "step": 8915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.084e-05,
+      "loss": 0.0761,
+      "step": 8916
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.083e-05,
+      "loss": 0.0466,
+      "step": 8917
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0820000000000001e-05,
+      "loss": 0.5529,
+      "step": 8918
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.081e-05,
+      "loss": 0.8599,
+      "step": 8919
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.08e-05,
+      "loss": 1.1354,
+      "step": 8920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.079e-05,
+      "loss": 1.1927,
+      "step": 8921
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.791,
+      "step": 8922
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.077e-05,
+      "loss": 1.5645,
+      "step": 8923
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.076e-05,
+      "loss": 0.898,
+      "step": 8924
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.075e-05,
+      "loss": 1.3271,
+      "step": 8925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.074e-05,
+      "loss": 1.3911,
+      "step": 8926
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.073e-05,
+      "loss": 0.661,
+      "step": 8927
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.072e-05,
+      "loss": 0.4176,
+      "step": 8928
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.071e-05,
+      "loss": 0.0155,
+      "step": 8929
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0700000000000001e-05,
+      "loss": 0.2575,
+      "step": 8930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.069e-05,
+      "loss": 1.2086,
+      "step": 8931
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0680000000000001e-05,
+      "loss": 0.5268,
+      "step": 8932
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0670000000000001e-05,
+      "loss": 0.9826,
+      "step": 8933
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0660000000000001e-05,
+      "loss": 1.4885,
+      "step": 8934
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.065e-05,
+      "loss": 1.2878,
+      "step": 8935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.064e-05,
+      "loss": 1.0394,
+      "step": 8936
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0630000000000002e-05,
+      "loss": 0.7332,
+      "step": 8937
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.062e-05,
+      "loss": 1.2627,
+      "step": 8938
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.061e-05,
+      "loss": 0.7234,
+      "step": 8939
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.06e-05,
+      "loss": 0.0377,
+      "step": 8940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.059e-05,
+      "loss": 0.0528,
+      "step": 8941
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.058e-05,
+      "loss": 0.2954,
+      "step": 8942
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.057e-05,
+      "loss": 0.2178,
+      "step": 8943
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.056e-05,
+      "loss": 0.875,
+      "step": 8944
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.055e-05,
+      "loss": 0.8384,
+      "step": 8945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0539999999999999e-05,
+      "loss": 1.0424,
+      "step": 8946
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.053e-05,
+      "loss": 0.6959,
+      "step": 8947
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0520000000000001e-05,
+      "loss": 0.8922,
+      "step": 8948
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0510000000000001e-05,
+      "loss": 1.1842,
+      "step": 8949
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.05e-05,
+      "loss": 1.2474,
+      "step": 8950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.049e-05,
+      "loss": 1.1817,
+      "step": 8951
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0480000000000001e-05,
+      "loss": 0.2418,
+      "step": 8952
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0470000000000001e-05,
+      "loss": 0.25,
+      "step": 8953
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.046e-05,
+      "loss": 0.1119,
+      "step": 8954
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.045e-05,
+      "loss": 0.9404,
+      "step": 8955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0440000000000002e-05,
+      "loss": 0.5985,
+      "step": 8956
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.043e-05,
+      "loss": 1.3291,
+      "step": 8957
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.042e-05,
+      "loss": 1.1641,
+      "step": 8958
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.041e-05,
+      "loss": 1.0186,
+      "step": 8959
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.04e-05,
+      "loss": 1.2423,
+      "step": 8960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.039e-05,
+      "loss": 1.0721,
+      "step": 8961
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.038e-05,
+      "loss": 1.176,
+      "step": 8962
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.037e-05,
+      "loss": 1.2327,
+      "step": 8963
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.036e-05,
+      "loss": 0.612,
+      "step": 8964
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.035e-05,
+      "loss": 0.2404,
+      "step": 8965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0340000000000001e-05,
+      "loss": 0.6585,
+      "step": 8966
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0330000000000001e-05,
+      "loss": 0.1739,
+      "step": 8967
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0320000000000001e-05,
+      "loss": 0.4164,
+      "step": 8968
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.031e-05,
+      "loss": 0.5238,
+      "step": 8969
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.03e-05,
+      "loss": 1.4072,
+      "step": 8970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0290000000000001e-05,
+      "loss": 0.827,
+      "step": 8971
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0280000000000002e-05,
+      "loss": 1.2538,
+      "step": 8972
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.027e-05,
+      "loss": 0.9728,
+      "step": 8973
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.026e-05,
+      "loss": 1.3465,
+      "step": 8974
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.025e-05,
+      "loss": 0.7429,
+      "step": 8975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.024e-05,
+      "loss": 0.6645,
+      "step": 8976
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.023e-05,
+      "loss": 0.1182,
+      "step": 8977
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.022e-05,
+      "loss": 0.2403,
+      "step": 8978
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.021e-05,
+      "loss": 0.0775,
+      "step": 8979
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.02e-05,
+      "loss": 0.6076,
+      "step": 8980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.019e-05,
+      "loss": 0.8956,
+      "step": 8981
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.018e-05,
+      "loss": 1.3875,
+      "step": 8982
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0170000000000001e-05,
+      "loss": 1.0818,
+      "step": 8983
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.016e-05,
+      "loss": 1.4318,
+      "step": 8984
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0150000000000001e-05,
+      "loss": 1.4289,
+      "step": 8985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 1.5648,
+      "step": 8986
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0130000000000001e-05,
+      "loss": 1.1604,
+      "step": 8987
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.012e-05,
+      "loss": 1.509,
+      "step": 8988
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.011e-05,
+      "loss": 0.3864,
+      "step": 8989
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0100000000000002e-05,
+      "loss": 0.2933,
+      "step": 8990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0090000000000002e-05,
+      "loss": 0.0023,
+      "step": 8991
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.008e-05,
+      "loss": 0.0355,
+      "step": 8992
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.007e-05,
+      "loss": 0.925,
+      "step": 8993
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.006e-05,
+      "loss": 0.4965,
+      "step": 8994
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.005e-05,
+      "loss": 1.4809,
+      "step": 8995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.004e-05,
+      "loss": 1.1089,
+      "step": 8996
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.003e-05,
+      "loss": 1.1451,
+      "step": 8997
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.002e-05,
+      "loss": 1.3522,
+      "step": 8998
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.001e-05,
+      "loss": 0.9589,
+      "step": 8999
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 1.2298,
+      "step": 9000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.990000000000001e-06,
+      "loss": 1.1212,
+      "step": 9001
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.4435,
+      "step": 9002
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97e-06,
+      "loss": 0.3982,
+      "step": 9003
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96e-06,
+      "loss": 0.0457,
+      "step": 9004
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.950000000000001e-06,
+      "loss": 0.3645,
+      "step": 9005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 0.1441,
+      "step": 9006
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.93e-06,
+      "loss": 1.6647,
+      "step": 9007
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92e-06,
+      "loss": 0.9996,
+      "step": 9008
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91e-06,
+      "loss": 1.0135,
+      "step": 9009
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900000000000002e-06,
+      "loss": 0.9664,
+      "step": 9010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89e-06,
+      "loss": 1.2895,
+      "step": 9011
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.88e-06,
+      "loss": 0.8661,
+      "step": 9012
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.87e-06,
+      "loss": 1.6382,
+      "step": 9013
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.86e-06,
+      "loss": 0.9724,
+      "step": 9014
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.85e-06,
+      "loss": 0.2579,
+      "step": 9015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.84e-06,
+      "loss": 0.1268,
+      "step": 9016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.83e-06,
+      "loss": 0.1962,
+      "step": 9017
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 0.268,
+      "step": 9018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810000000000001e-06,
+      "loss": 0.5511,
+      "step": 9019
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.5402,
+      "step": 9020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.790000000000001e-06,
+      "loss": 1.1797,
+      "step": 9021
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78e-06,
+      "loss": 1.5561,
+      "step": 9022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.77e-06,
+      "loss": 1.3856,
+      "step": 9023
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 1.1156,
+      "step": 9024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.750000000000002e-06,
+      "loss": 1.3707,
+      "step": 9025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.74e-06,
+      "loss": 0.978,
+      "step": 9026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.73e-06,
+      "loss": 0.134,
+      "step": 9027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72e-06,
+      "loss": 0.1492,
+      "step": 9028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71e-06,
+      "loss": 0.0215,
+      "step": 9029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.7e-06,
+      "loss": 0.0297,
+      "step": 9030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.69e-06,
+      "loss": 0.539,
+      "step": 9031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.68e-06,
+      "loss": 1.0943,
+      "step": 9032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.67e-06,
+      "loss": 1.2388,
+      "step": 9033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.66e-06,
+      "loss": 1.1196,
+      "step": 9034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.65e-06,
+      "loss": 1.236,
+      "step": 9035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 1.0431,
+      "step": 9036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.630000000000001e-06,
+      "loss": 0.8436,
+      "step": 9037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.62e-06,
+      "loss": 1.6103,
+      "step": 9038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.610000000000001e-06,
+      "loss": 0.8361,
+      "step": 9039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.2613,
+      "step": 9040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.59e-06,
+      "loss": 0.3021,
+      "step": 9041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.58e-06,
+      "loss": 0.1518,
+      "step": 9042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.57e-06,
+      "loss": 0.0684,
+      "step": 9043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.560000000000002e-06,
+      "loss": 0.6407,
+      "step": 9044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.55e-06,
+      "loss": 0.762,
+      "step": 9045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.54e-06,
+      "loss": 1.3802,
+      "step": 9046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.53e-06,
+      "loss": 0.9305,
+      "step": 9047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.52e-06,
+      "loss": 0.9756,
+      "step": 9048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.51e-06,
+      "loss": 1.2563,
+      "step": 9049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.5e-06,
+      "loss": 1.0087,
+      "step": 9050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.49e-06,
+      "loss": 1.009,
+      "step": 9051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.48e-06,
+      "loss": 0.4823,
+      "step": 9052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.47e-06,
+      "loss": 0.1873,
+      "step": 9053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 0.0105,
+      "step": 9054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.450000000000001e-06,
+      "loss": 0.1768,
+      "step": 9055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.44e-06,
+      "loss": 0.5939,
+      "step": 9056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.43e-06,
+      "loss": 1.1324,
+      "step": 9057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.420000000000001e-06,
+      "loss": 1.2305,
+      "step": 9058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.410000000000001e-06,
+      "loss": 0.809,
+      "step": 9059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.4e-06,
+      "loss": 1.4144,
+      "step": 9060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.39e-06,
+      "loss": 1.2908,
+      "step": 9061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.38e-06,
+      "loss": 1.08,
+      "step": 9062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.370000000000002e-06,
+      "loss": 1.5512,
+      "step": 9063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.36e-06,
+      "loss": 1.0278,
+      "step": 9064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.35e-06,
+      "loss": 0.0176,
+      "step": 9065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.34e-06,
+      "loss": 0.2656,
+      "step": 9066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.33e-06,
+      "loss": 0.3017,
+      "step": 9067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.32e-06,
+      "loss": 0.2691,
+      "step": 9068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.31e-06,
+      "loss": 0.6199,
+      "step": 9069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.3e-06,
+      "loss": 1.2943,
+      "step": 9070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.29e-06,
+      "loss": 0.9131,
+      "step": 9071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.28e-06,
+      "loss": 1.1101,
+      "step": 9072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.270000000000001e-06,
+      "loss": 1.1707,
+      "step": 9073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 0.9883,
+      "step": 9074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.25e-06,
+      "loss": 0.6343,
+      "step": 9075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.24e-06,
+      "loss": 1.14,
+      "step": 9076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.23e-06,
+      "loss": 0.3203,
+      "step": 9077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 0.0587,
+      "step": 9078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.21e-06,
+      "loss": 0.5765,
+      "step": 9079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.2e-06,
+      "loss": 0.3819,
+      "step": 9080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.19e-06,
+      "loss": 0.3055,
+      "step": 9081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 1.0407,
+      "step": 9082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.17e-06,
+      "loss": 0.9788,
+      "step": 9083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.16e-06,
+      "loss": 1.1338,
+      "step": 9084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.15e-06,
+      "loss": 1.1311,
+      "step": 9085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.14e-06,
+      "loss": 1.1857,
+      "step": 9086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.13e-06,
+      "loss": 1.0759,
+      "step": 9087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.12e-06,
+      "loss": 1.0086,
+      "step": 9088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.110000000000001e-06,
+      "loss": 0.8172,
+      "step": 9089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 0.3383,
+      "step": 9090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.09e-06,
+      "loss": 0.0157,
+      "step": 9091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 0.7139,
+      "step": 9092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.070000000000001e-06,
+      "loss": 0.652,
+      "step": 9093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.06e-06,
+      "loss": 1.1221,
+      "step": 9094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.05e-06,
+      "loss": 1.2626,
+      "step": 9095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.04e-06,
+      "loss": 1.044,
+      "step": 9096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.030000000000002e-06,
+      "loss": 1.2074,
+      "step": 9097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.02e-06,
+      "loss": 0.9351,
+      "step": 9098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.01e-06,
+      "loss": 1.0331,
+      "step": 9099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9e-06,
+      "loss": 1.4624,
+      "step": 9100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.99e-06,
+      "loss": 1.3172,
+      "step": 9101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.98e-06,
+      "loss": 0.3975,
+      "step": 9102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.97e-06,
+      "loss": 0.0614,
+      "step": 9103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.96e-06,
+      "loss": 0.1135,
+      "step": 9104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.95e-06,
+      "loss": 0.4417,
+      "step": 9105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.939999999999999e-06,
+      "loss": 0.3226,
+      "step": 9106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.930000000000001e-06,
+      "loss": 1.3098,
+      "step": 9107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 1.2155,
+      "step": 9108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.910000000000001e-06,
+      "loss": 0.8638,
+      "step": 9109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.9e-06,
+      "loss": 0.6958,
+      "step": 9110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.890000000000001e-06,
+      "loss": 0.9253,
+      "step": 9111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 0.9871,
+      "step": 9112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.87e-06,
+      "loss": 1.6358,
+      "step": 9113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.86e-06,
+      "loss": 0.6114,
+      "step": 9114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.85e-06,
+      "loss": 0.2049,
+      "step": 9115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.840000000000002e-06,
+      "loss": 0.0222,
+      "step": 9116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.83e-06,
+      "loss": 0.2525,
+      "step": 9117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.82e-06,
+      "loss": 0.5165,
+      "step": 9118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.81e-06,
+      "loss": 1.117,
+      "step": 9119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.8e-06,
+      "loss": 1.4894,
+      "step": 9120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.79e-06,
+      "loss": 1.5274,
+      "step": 9121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.78e-06,
+      "loss": 1.1452,
+      "step": 9122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.77e-06,
+      "loss": 0.688,
+      "step": 9123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.76e-06,
+      "loss": 1.2945,
+      "step": 9124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.75e-06,
+      "loss": 1.6924,
+      "step": 9125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 1.2706,
+      "step": 9126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.730000000000001e-06,
+      "loss": 0.1178,
+      "step": 9127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 0.256,
+      "step": 9128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.71e-06,
+      "loss": 0.0808,
+      "step": 9129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.7e-06,
+      "loss": 0.0838,
+      "step": 9130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.690000000000002e-06,
+      "loss": 1.6484,
+      "step": 9131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.68e-06,
+      "loss": 1.3042,
+      "step": 9132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.67e-06,
+      "loss": 0.8447,
+      "step": 9133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.66e-06,
+      "loss": 1.0493,
+      "step": 9134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.65e-06,
+      "loss": 1.2723,
+      "step": 9135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.64e-06,
+      "loss": 1.0904,
+      "step": 9136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.63e-06,
+      "loss": 1.1346,
+      "step": 9137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.62e-06,
+      "loss": 1.3207,
+      "step": 9138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.61e-06,
+      "loss": 1.0153,
+      "step": 9139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 0.4938,
+      "step": 9140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.59e-06,
+      "loss": 0.0442,
+      "step": 9141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.2046,
+      "step": 9142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.570000000000001e-06,
+      "loss": 0.9083,
+      "step": 9143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.56e-06,
+      "loss": 1.2679,
+      "step": 9144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.550000000000001e-06,
+      "loss": 1.1047,
+      "step": 9145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 0.8381,
+      "step": 9146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.53e-06,
+      "loss": 1.4453,
+      "step": 9147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.52e-06,
+      "loss": 1.0744,
+      "step": 9148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.51e-06,
+      "loss": 0.9264,
+      "step": 9149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 1.169,
+      "step": 9150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.49e-06,
+      "loss": 1.137,
+      "step": 9151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.48e-06,
+      "loss": 0.1413,
+      "step": 9152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.47e-06,
+      "loss": 0.3888,
+      "step": 9153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.46e-06,
+      "loss": 0.1493,
+      "step": 9154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.45e-06,
+      "loss": 0.2455,
+      "step": 9155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.44e-06,
+      "loss": 0.4766,
+      "step": 9156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.43e-06,
+      "loss": 1.2278,
+      "step": 9157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.42e-06,
+      "loss": 1.5308,
+      "step": 9158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.409999999999999e-06,
+      "loss": 1.0765,
+      "step": 9159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.9664,
+      "step": 9160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.390000000000001e-06,
+      "loss": 1.2857,
+      "step": 9161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.6676,
+      "step": 9162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.37e-06,
+      "loss": 1.7137,
+      "step": 9163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.36e-06,
+      "loss": 0.6834,
+      "step": 9164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.350000000000001e-06,
+      "loss": 0.1545,
+      "step": 9165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.34e-06,
+      "loss": 0.2959,
+      "step": 9166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.33e-06,
+      "loss": 0.0947,
+      "step": 9167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.32e-06,
+      "loss": 0.7126,
+      "step": 9168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.31e-06,
+      "loss": 0.9879,
+      "step": 9169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.3e-06,
+      "loss": 1.17,
+      "step": 9170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.29e-06,
+      "loss": 1.3426,
+      "step": 9171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.28e-06,
+      "loss": 1.0589,
+      "step": 9172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.27e-06,
+      "loss": 1.5209,
+      "step": 9173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.26e-06,
+      "loss": 0.8943,
+      "step": 9174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.25e-06,
+      "loss": 1.5103,
+      "step": 9175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.24e-06,
+      "loss": 1.1995,
+      "step": 9176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.23e-06,
+      "loss": 0.3928,
+      "step": 9177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.22e-06,
+      "loss": 0.1115,
+      "step": 9178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.210000000000001e-06,
+      "loss": 0.2667,
+      "step": 9179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 0.1169,
+      "step": 9180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.190000000000001e-06,
+      "loss": 0.6046,
+      "step": 9181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.18e-06,
+      "loss": 1.3369,
+      "step": 9182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.17e-06,
+      "loss": 0.8647,
+      "step": 9183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.160000000000001e-06,
+      "loss": 1.1186,
+      "step": 9184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.15e-06,
+      "loss": 1.4034,
+      "step": 9185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.14e-06,
+      "loss": 0.985,
+      "step": 9186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.13e-06,
+      "loss": 0.9089,
+      "step": 9187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.12e-06,
+      "loss": 1.012,
+      "step": 9188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.11e-06,
+      "loss": 0.688,
+      "step": 9189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.1e-06,
+      "loss": 0.1135,
+      "step": 9190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.09e-06,
+      "loss": 0.0559,
+      "step": 9191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.08e-06,
+      "loss": 0.2387,
+      "step": 9192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.069999999999999e-06,
+      "loss": 0.1306,
+      "step": 9193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.06e-06,
+      "loss": 0.6419,
+      "step": 9194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.050000000000001e-06,
+      "loss": 0.7693,
+      "step": 9195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 1.1027,
+      "step": 9196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.03e-06,
+      "loss": 0.7846,
+      "step": 9197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.02e-06,
+      "loss": 0.9138,
+      "step": 9198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.010000000000001e-06,
+      "loss": 1.0668,
+      "step": 9199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.8282,
+      "step": 9200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.99e-06,
+      "loss": 1.1161,
+      "step": 9201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.98e-06,
+      "loss": 0.4034,
+      "step": 9202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.97e-06,
+      "loss": 0.0122,
+      "step": 9203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.96e-06,
+      "loss": 0.0767,
+      "step": 9204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.95e-06,
+      "loss": 1.0141,
+      "step": 9205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.94e-06,
+      "loss": 1.0845,
+      "step": 9206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.93e-06,
+      "loss": 0.9617,
+      "step": 9207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.92e-06,
+      "loss": 0.9491,
+      "step": 9208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.91e-06,
+      "loss": 1.3047,
+      "step": 9209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.9e-06,
+      "loss": 0.96,
+      "step": 9210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.89e-06,
+      "loss": 1.3428,
+      "step": 9211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.879999999999999e-06,
+      "loss": 0.9119,
+      "step": 9212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.870000000000001e-06,
+      "loss": 1.6642,
+      "step": 9213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 0.9977,
+      "step": 9214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.850000000000001e-06,
+      "loss": 0.0313,
+      "step": 9215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.84e-06,
+      "loss": 0.1155,
+      "step": 9216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.83e-06,
+      "loss": 0.0952,
+      "step": 9217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 0.6737,
+      "step": 9218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.810000000000001e-06,
+      "loss": 0.5008,
+      "step": 9219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.8e-06,
+      "loss": 0.9697,
+      "step": 9220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.79e-06,
+      "loss": 1.3001,
+      "step": 9221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.78e-06,
+      "loss": 1.7141,
+      "step": 9222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.77e-06,
+      "loss": 1.2476,
+      "step": 9223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.76e-06,
+      "loss": 0.9622,
+      "step": 9224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.75e-06,
+      "loss": 0.5103,
+      "step": 9225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.74e-06,
+      "loss": 1.4008,
+      "step": 9226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.73e-06,
+      "loss": 0.3768,
+      "step": 9227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.72e-06,
+      "loss": 0.243,
+      "step": 9228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.71e-06,
+      "loss": 0.05,
+      "step": 9229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.7e-06,
+      "loss": 0.355,
+      "step": 9230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.69e-06,
+      "loss": 0.9911,
+      "step": 9231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.68e-06,
+      "loss": 1.0412,
+      "step": 9232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.670000000000001e-06,
+      "loss": 1.4688,
+      "step": 9233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 1.3916,
+      "step": 9234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.65e-06,
+      "loss": 1.1853,
+      "step": 9235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.64e-06,
+      "loss": 1.4791,
+      "step": 9236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.630000000000001e-06,
+      "loss": 1.3106,
+      "step": 9237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 1.523,
+      "step": 9238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.610000000000001e-06,
+      "loss": 0.4371,
+      "step": 9239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.6e-06,
+      "loss": 0.1286,
+      "step": 9240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.59e-06,
+      "loss": 0.0574,
+      "step": 9241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.580000000000001e-06,
+      "loss": 0.3603,
+      "step": 9242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.57e-06,
+      "loss": 0.2256,
+      "step": 9243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 0.7025,
+      "step": 9244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.55e-06,
+      "loss": 1.3485,
+      "step": 9245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.54e-06,
+      "loss": 1.3015,
+      "step": 9246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.530000000000001e-06,
+      "loss": 0.8301,
+      "step": 9247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 1.3893,
+      "step": 9248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.51e-06,
+      "loss": 1.0887,
+      "step": 9249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.5e-06,
+      "loss": 1.0289,
+      "step": 9250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.4899999999999994e-06,
+      "loss": 1.3692,
+      "step": 9251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.480000000000001e-06,
+      "loss": 0.4842,
+      "step": 9252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.4700000000000005e-06,
+      "loss": 0.1557,
+      "step": 9253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 0.0974,
+      "step": 9254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.45e-06,
+      "loss": 0.2424,
+      "step": 9255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.44e-06,
+      "loss": 0.8509,
+      "step": 9256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.430000000000001e-06,
+      "loss": 0.9718,
+      "step": 9257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 1.183,
+      "step": 9258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.41e-06,
+      "loss": 1.3301,
+      "step": 9259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.4e-06,
+      "loss": 1.2358,
+      "step": 9260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.3899999999999995e-06,
+      "loss": 1.34,
+      "step": 9261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.883,
+      "step": 9262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.370000000000001e-06,
+      "loss": 1.1593,
+      "step": 9263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.36e-06,
+      "loss": 0.998,
+      "step": 9264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.35e-06,
+      "loss": 0.1051,
+      "step": 9265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 0.5916,
+      "step": 9266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.330000000000001e-06,
+      "loss": 0.1857,
+      "step": 9267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.32e-06,
+      "loss": 0.3499,
+      "step": 9268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.31e-06,
+      "loss": 0.8502,
+      "step": 9269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.2999999999999996e-06,
+      "loss": 1.3894,
+      "step": 9270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.290000000000001e-06,
+      "loss": 1.1426,
+      "step": 9271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 0.658,
+      "step": 9272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.270000000000001e-06,
+      "loss": 0.9729,
+      "step": 9273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.26e-06,
+      "loss": 0.7465,
+      "step": 9274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.25e-06,
+      "loss": 1.1162,
+      "step": 9275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.240000000000001e-06,
+      "loss": 1.0719,
+      "step": 9276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.230000000000001e-06,
+      "loss": 0.0395,
+      "step": 9277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.22e-06,
+      "loss": 0.0444,
+      "step": 9278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.2100000000000004e-06,
+      "loss": 0.0584,
+      "step": 9279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.2e-06,
+      "loss": 0.629,
+      "step": 9280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.190000000000001e-06,
+      "loss": 1.0895,
+      "step": 9281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.8235,
+      "step": 9282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.17e-06,
+      "loss": 1.1972,
+      "step": 9283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.16e-06,
+      "loss": 0.9588,
+      "step": 9284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.15e-06,
+      "loss": 0.9842,
+      "step": 9285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.140000000000001e-06,
+      "loss": 1.148,
+      "step": 9286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.13e-06,
+      "loss": 1.212,
+      "step": 9287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 0.8871,
+      "step": 9288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.11e-06,
+      "loss": 0.2441,
+      "step": 9289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.1e-06,
+      "loss": 0.4461,
+      "step": 9290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.090000000000001e-06,
+      "loss": 0.0375,
+      "step": 9291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.080000000000001e-06,
+      "loss": 0.359,
+      "step": 9292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.07e-06,
+      "loss": 0.9043,
+      "step": 9293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.06e-06,
+      "loss": 1.442,
+      "step": 9294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.049999999999999e-06,
+      "loss": 1.3265,
+      "step": 9295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.04e-06,
+      "loss": 1.3084,
+      "step": 9296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.0300000000000005e-06,
+      "loss": 0.9888,
+      "step": 9297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 1.03,
+      "step": 9298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.01e-06,
+      "loss": 1.0585,
+      "step": 9299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.8203,
+      "step": 9300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.990000000000001e-06,
+      "loss": 1.0938,
+      "step": 9301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.98e-06,
+      "loss": 0.6571,
+      "step": 9302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.97e-06,
+      "loss": 0.0129,
+      "step": 9303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.9599999999999994e-06,
+      "loss": 0.0314,
+      "step": 9304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.950000000000001e-06,
+      "loss": 0.3645,
+      "step": 9305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 0.6765,
+      "step": 9306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.9300000000000006e-06,
+      "loss": 1.2494,
+      "step": 9307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.92e-06,
+      "loss": 1.1829,
+      "step": 9308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.91e-06,
+      "loss": 1.5681,
+      "step": 9309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 0.9834,
+      "step": 9310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.890000000000001e-06,
+      "loss": 0.8072,
+      "step": 9311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.88e-06,
+      "loss": 1.2137,
+      "step": 9312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.87e-06,
+      "loss": 1.3338,
+      "step": 9313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.8599999999999995e-06,
+      "loss": 0.5861,
+      "step": 9314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.8500000000000005e-06,
+      "loss": 0.3873,
+      "step": 9315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 0.2564,
+      "step": 9316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.830000000000001e-06,
+      "loss": 0.2028,
+      "step": 9317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.82e-06,
+      "loss": 0.4038,
+      "step": 9318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.81e-06,
+      "loss": 1.4787,
+      "step": 9319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.3331,
+      "step": 9320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.79e-06,
+      "loss": 1.3337,
+      "step": 9321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.78e-06,
+      "loss": 1.2481,
+      "step": 9322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.7699999999999996e-06,
+      "loss": 1.0703,
+      "step": 9323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.76e-06,
+      "loss": 1.0136,
+      "step": 9324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 1.2092,
+      "step": 9325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 1.21,
+      "step": 9326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.73e-06,
+      "loss": 0.328,
+      "step": 9327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.72e-06,
+      "loss": 0.1331,
+      "step": 9328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.710000000000001e-06,
+      "loss": 0.0694,
+      "step": 9329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 0.5027,
+      "step": 9330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.69e-06,
+      "loss": 0.7879,
+      "step": 9331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.68e-06,
+      "loss": 1.0729,
+      "step": 9332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.67e-06,
+      "loss": 0.8605,
+      "step": 9333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 0.6729,
+      "step": 9334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.650000000000001e-06,
+      "loss": 1.2001,
+      "step": 9335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 1.1282,
+      "step": 9336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.63e-06,
+      "loss": 1.1929,
+      "step": 9337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.62e-06,
+      "loss": 1.6573,
+      "step": 9338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.610000000000001e-06,
+      "loss": 0.7417,
+      "step": 9339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.6e-06,
+      "loss": 0.1528,
+      "step": 9340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.5900000000000004e-06,
+      "loss": 0.1574,
+      "step": 9341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.58e-06,
+      "loss": 0.1408,
+      "step": 9342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.57e-06,
+      "loss": 0.552,
+      "step": 9343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 0.9379,
+      "step": 9344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.550000000000001e-06,
+      "loss": 1.2062,
+      "step": 9345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.54e-06,
+      "loss": 1.0262,
+      "step": 9346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.53e-06,
+      "loss": 1.4361,
+      "step": 9347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.519999999999999e-06,
+      "loss": 0.6651,
+      "step": 9348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.510000000000001e-06,
+      "loss": 1.2873,
+      "step": 9349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 1.3931,
+      "step": 9350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.4900000000000005e-06,
+      "loss": 1.2813,
+      "step": 9351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.48e-06,
+      "loss": 0.5694,
+      "step": 9352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.47e-06,
+      "loss": 0.1196,
+      "step": 9353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 0.2299,
+      "step": 9354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.45e-06,
+      "loss": 0.4676,
+      "step": 9355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.44e-06,
+      "loss": 1.0627,
+      "step": 9356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.43e-06,
+      "loss": 1.3079,
+      "step": 9357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.4199999999999995e-06,
+      "loss": 0.9292,
+      "step": 9358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.4100000000000005e-06,
+      "loss": 0.8918,
+      "step": 9359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.0126,
+      "step": 9360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.39e-06,
+      "loss": 1.2144,
+      "step": 9361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.38e-06,
+      "loss": 1.0761,
+      "step": 9362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.370000000000001e-06,
+      "loss": 1.4981,
+      "step": 9363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 0.8046,
+      "step": 9364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.35e-06,
+      "loss": 0.0632,
+      "step": 9365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.34e-06,
+      "loss": 0.021,
+      "step": 9366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.3299999999999995e-06,
+      "loss": 0.0988,
+      "step": 9367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.320000000000001e-06,
+      "loss": 0.6189,
+      "step": 9368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.3100000000000006e-06,
+      "loss": 1.2076,
+      "step": 9369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 1.2006,
+      "step": 9370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.29e-06,
+      "loss": 1.3145,
+      "step": 9371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.28e-06,
+      "loss": 1.0331,
+      "step": 9372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.270000000000001e-06,
+      "loss": 1.2152,
+      "step": 9373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.26e-06,
+      "loss": 1.2781,
+      "step": 9374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.25e-06,
+      "loss": 1.3541,
+      "step": 9375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.24e-06,
+      "loss": 0.8207,
+      "step": 9376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.2300000000000005e-06,
+      "loss": 0.2547,
+      "step": 9377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.22e-06,
+      "loss": 0.1391,
+      "step": 9378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.210000000000001e-06,
+      "loss": 0.2055,
+      "step": 9379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.2e-06,
+      "loss": 0.6358,
+      "step": 9380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.19e-06,
+      "loss": 0.7535,
+      "step": 9381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.18e-06,
+      "loss": 0.8816,
+      "step": 9382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.17e-06,
+      "loss": 1.0721,
+      "step": 9383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.16e-06,
+      "loss": 1.2359,
+      "step": 9384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.15e-06,
+      "loss": 1.2635,
+      "step": 9385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 1.6325,
+      "step": 9386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.130000000000001e-06,
+      "loss": 1.2254,
+      "step": 9387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.12e-06,
+      "loss": 1.4128,
+      "step": 9388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.110000000000001e-06,
+      "loss": 0.9734,
+      "step": 9389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.1e-06,
+      "loss": 0.3017,
+      "step": 9390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.090000000000001e-06,
+      "loss": 0.1204,
+      "step": 9391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.08e-06,
+      "loss": 0.2069,
+      "step": 9392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.07e-06,
+      "loss": 0.4958,
+      "step": 9393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 1.3427,
+      "step": 9394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.0500000000000005e-06,
+      "loss": 1.316,
+      "step": 9395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 1.0999,
+      "step": 9396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.03e-06,
+      "loss": 1.2692,
+      "step": 9397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.02e-06,
+      "loss": 0.7737,
+      "step": 9398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.01e-06,
+      "loss": 1.0379,
+      "step": 9399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6e-06,
+      "loss": 0.8529,
+      "step": 9400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99e-06,
+      "loss": 1.4835,
+      "step": 9401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.98e-06,
+      "loss": 0.3358,
+      "step": 9402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9700000000000004e-06,
+      "loss": 0.1945,
+      "step": 9403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 0.0187,
+      "step": 9404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.95e-06,
+      "loss": 0.0046,
+      "step": 9405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.940000000000001e-06,
+      "loss": 1.0547,
+      "step": 9406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.93e-06,
+      "loss": 1.421,
+      "step": 9407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.920000000000001e-06,
+      "loss": 1.0147,
+      "step": 9408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.91e-06,
+      "loss": 1.0109,
+      "step": 9409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9e-06,
+      "loss": 1.3779,
+      "step": 9410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.89e-06,
+      "loss": 1.1825,
+      "step": 9411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 0.7139,
+      "step": 9412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.8700000000000005e-06,
+      "loss": 1.2911,
+      "step": 9413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.86e-06,
+      "loss": 0.5299,
+      "step": 9414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.850000000000001e-06,
+      "loss": 0.092,
+      "step": 9415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.84e-06,
+      "loss": 0.1195,
+      "step": 9416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.83e-06,
+      "loss": 0.3381,
+      "step": 9417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.82e-06,
+      "loss": 0.919,
+      "step": 9418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.81e-06,
+      "loss": 1.3924,
+      "step": 9419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.8e-06,
+      "loss": 1.046,
+      "step": 9420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.7900000000000005e-06,
+      "loss": 1.2561,
+      "step": 9421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.78e-06,
+      "loss": 1.3404,
+      "step": 9422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.770000000000001e-06,
+      "loss": 1.1153,
+      "step": 9423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.76e-06,
+      "loss": 0.8992,
+      "step": 9424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.750000000000001e-06,
+      "loss": 1.1885,
+      "step": 9425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.74e-06,
+      "loss": 1.1545,
+      "step": 9426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.73e-06,
+      "loss": 0.2467,
+      "step": 9427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.72e-06,
+      "loss": 0.1775,
+      "step": 9428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.71e-06,
+      "loss": 0.053,
+      "step": 9429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 0.3844,
+      "step": 9430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.690000000000001e-06,
+      "loss": 0.0363,
+      "step": 9431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.680000000000001e-06,
+      "loss": 1.1411,
+      "step": 9432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.67e-06,
+      "loss": 1.3372,
+      "step": 9433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.66e-06,
+      "loss": 0.9467,
+      "step": 9434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.65e-06,
+      "loss": 1.0917,
+      "step": 9435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.64e-06,
+      "loss": 1.3011,
+      "step": 9436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.63e-06,
+      "loss": 0.6959,
+      "step": 9437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.62e-06,
+      "loss": 1.2736,
+      "step": 9438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.61e-06,
+      "loss": 0.8117,
+      "step": 9439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.2596,
+      "step": 9440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.59e-06,
+      "loss": 0.2427,
+      "step": 9441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.3826,
+      "step": 9442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.57e-06,
+      "loss": 0.1864,
+      "step": 9443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.56e-06,
+      "loss": 1.3609,
+      "step": 9444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.55e-06,
+      "loss": 0.8822,
+      "step": 9445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.54e-06,
+      "loss": 1.5571,
+      "step": 9446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.53e-06,
+      "loss": 1.2244,
+      "step": 9447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 0.8672,
+      "step": 9448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.510000000000001e-06,
+      "loss": 1.3941,
+      "step": 9449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 1.2571,
+      "step": 9450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.49e-06,
+      "loss": 1.6936,
+      "step": 9451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.48e-06,
+      "loss": 0.5284,
+      "step": 9452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.47e-06,
+      "loss": 0.2583,
+      "step": 9453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.46e-06,
+      "loss": 0.0184,
+      "step": 9454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.45e-06,
+      "loss": 0.2325,
+      "step": 9455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.44e-06,
+      "loss": 0.7285,
+      "step": 9456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.4300000000000005e-06,
+      "loss": 0.8934,
+      "step": 9457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.42e-06,
+      "loss": 1.4488,
+      "step": 9458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.410000000000001e-06,
+      "loss": 1.2091,
+      "step": 9459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.4e-06,
+      "loss": 0.9613,
+      "step": 9460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.390000000000001e-06,
+      "loss": 1.0766,
+      "step": 9461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.38e-06,
+      "loss": 1.0963,
+      "step": 9462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.37e-06,
+      "loss": 1.1446,
+      "step": 9463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.36e-06,
+      "loss": 1.1262,
+      "step": 9464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.3500000000000004e-06,
+      "loss": 0.2041,
+      "step": 9465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 0.2461,
+      "step": 9466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.330000000000001e-06,
+      "loss": 0.0975,
+      "step": 9467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.32e-06,
+      "loss": 0.1884,
+      "step": 9468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.31e-06,
+      "loss": 0.5831,
+      "step": 9469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.3e-06,
+      "loss": 1.1814,
+      "step": 9470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.29e-06,
+      "loss": 1.3025,
+      "step": 9471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.28e-06,
+      "loss": 1.3519,
+      "step": 9472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.2699999999999995e-06,
+      "loss": 1.0944,
+      "step": 9473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 0.8551,
+      "step": 9474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.25e-06,
+      "loss": 1.4025,
+      "step": 9475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 1.1712,
+      "step": 9476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.23e-06,
+      "loss": 0.3878,
+      "step": 9477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 0.2377,
+      "step": 9478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.21e-06,
+      "loss": 0.077,
+      "step": 9479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.2e-06,
+      "loss": 0.0228,
+      "step": 9480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.19e-06,
+      "loss": 1.0568,
+      "step": 9481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.18e-06,
+      "loss": 1.4108,
+      "step": 9482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.1700000000000005e-06,
+      "loss": 1.0805,
+      "step": 9483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 0.8758,
+      "step": 9484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.15e-06,
+      "loss": 1.0039,
+      "step": 9485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 0.9052,
+      "step": 9486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.13e-06,
+      "loss": 1.1451,
+      "step": 9487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.12e-06,
+      "loss": 1.6597,
+      "step": 9488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.11e-06,
+      "loss": 0.6817,
+      "step": 9489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.1e-06,
+      "loss": 0.0324,
+      "step": 9490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.09e-06,
+      "loss": 0.2281,
+      "step": 9491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.08e-06,
+      "loss": 0.1057,
+      "step": 9492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.070000000000001e-06,
+      "loss": 0.2692,
+      "step": 9493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.06e-06,
+      "loss": 1.3191,
+      "step": 9494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.050000000000001e-06,
+      "loss": 1.1299,
+      "step": 9495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.04e-06,
+      "loss": 0.8608,
+      "step": 9496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.03e-06,
+      "loss": 1.0194,
+      "step": 9497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.02e-06,
+      "loss": 1.021,
+      "step": 9498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.01e-06,
+      "loss": 0.9389,
+      "step": 9499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 1.7205,
+      "step": 9500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9900000000000005e-06,
+      "loss": 1.3018,
+      "step": 9501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.98e-06,
+      "loss": 0.7002,
+      "step": 9502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.970000000000001e-06,
+      "loss": 0.2937,
+      "step": 9503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.96e-06,
+      "loss": 0.1504,
+      "step": 9504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.950000000000001e-06,
+      "loss": 0.2121,
+      "step": 9505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.94e-06,
+      "loss": 0.677,
+      "step": 9506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.93e-06,
+      "loss": 0.926,
+      "step": 9507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.92e-06,
+      "loss": 0.8874,
+      "step": 9508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9100000000000004e-06,
+      "loss": 1.1351,
+      "step": 9509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 1.5312,
+      "step": 9510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.89e-06,
+      "loss": 0.9529,
+      "step": 9511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 1.5059,
+      "step": 9512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.87e-06,
+      "loss": 1.4399,
+      "step": 9513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.86e-06,
+      "loss": 1.2416,
+      "step": 9514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.85e-06,
+      "loss": 0.2295,
+      "step": 9515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.84e-06,
+      "loss": 0.1866,
+      "step": 9516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.83e-06,
+      "loss": 0.1738,
+      "step": 9517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 0.1405,
+      "step": 9518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.81e-06,
+      "loss": 1.1445,
+      "step": 9519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.0268,
+      "step": 9520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.79e-06,
+      "loss": 1.066,
+      "step": 9521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.780000000000001e-06,
+      "loss": 1.2441,
+      "step": 9522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.77e-06,
+      "loss": 0.9849,
+      "step": 9523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.76e-06,
+      "loss": 1.5934,
+      "step": 9524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.75e-06,
+      "loss": 1.4279,
+      "step": 9525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.74e-06,
+      "loss": 0.9563,
+      "step": 9526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7300000000000005e-06,
+      "loss": 0.298,
+      "step": 9527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.72e-06,
+      "loss": 0.0369,
+      "step": 9528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.710000000000001e-06,
+      "loss": 0.2531,
+      "step": 9529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7e-06,
+      "loss": 0.3488,
+      "step": 9530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.69e-06,
+      "loss": 1.1318,
+      "step": 9531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.68e-06,
+      "loss": 1.2951,
+      "step": 9532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.67e-06,
+      "loss": 1.0039,
+      "step": 9533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.66e-06,
+      "loss": 1.247,
+      "step": 9534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.65e-06,
+      "loss": 1.3483,
+      "step": 9535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.64e-06,
+      "loss": 1.1398,
+      "step": 9536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6300000000000006e-06,
+      "loss": 1.2365,
+      "step": 9537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.62e-06,
+      "loss": 2.0361,
+      "step": 9538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.610000000000001e-06,
+      "loss": 0.5404,
+      "step": 9539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6e-06,
+      "loss": 0.3246,
+      "step": 9540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.590000000000001e-06,
+      "loss": 0.2358,
+      "step": 9541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.58e-06,
+      "loss": 0.0396,
+      "step": 9542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.57e-06,
+      "loss": 0.2297,
+      "step": 9543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.56e-06,
+      "loss": 1.3555,
+      "step": 9544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5500000000000005e-06,
+      "loss": 1.5116,
+      "step": 9545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 1.5182,
+      "step": 9546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.53e-06,
+      "loss": 0.9905,
+      "step": 9547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.52e-06,
+      "loss": 1.0342,
+      "step": 9548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.51e-06,
+      "loss": 1.2864,
+      "step": 9549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5e-06,
+      "loss": 1.069,
+      "step": 9550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.49e-06,
+      "loss": 1.4527,
+      "step": 9551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.48e-06,
+      "loss": 0.3237,
+      "step": 9552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4699999999999996e-06,
+      "loss": 0.0718,
+      "step": 9553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 0.0673,
+      "step": 9554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.45e-06,
+      "loss": 0.2912,
+      "step": 9555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 0.6337,
+      "step": 9556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.43e-06,
+      "loss": 1.1896,
+      "step": 9557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.420000000000001e-06,
+      "loss": 1.1467,
+      "step": 9558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.41e-06,
+      "loss": 0.9315,
+      "step": 9559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4e-06,
+      "loss": 1.0472,
+      "step": 9560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.39e-06,
+      "loss": 0.8513,
+      "step": 9561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.38e-06,
+      "loss": 0.9456,
+      "step": 9562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.3700000000000005e-06,
+      "loss": 1.2204,
+      "step": 9563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 0.4768,
+      "step": 9564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.35e-06,
+      "loss": 0.3766,
+      "step": 9565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.34e-06,
+      "loss": 0.007,
+      "step": 9566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.33e-06,
+      "loss": 0.1405,
+      "step": 9567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.32e-06,
+      "loss": 0.0811,
+      "step": 9568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.31e-06,
+      "loss": 1.4327,
+      "step": 9569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.2999999999999995e-06,
+      "loss": 1.3682,
+      "step": 9570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.2900000000000004e-06,
+      "loss": 1.155,
+      "step": 9571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.28e-06,
+      "loss": 0.8706,
+      "step": 9572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.270000000000001e-06,
+      "loss": 1.3015,
+      "step": 9573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.26e-06,
+      "loss": 1.4678,
+      "step": 9574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.250000000000001e-06,
+      "loss": 1.3998,
+      "step": 9575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.24e-06,
+      "loss": 1.1331,
+      "step": 9576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.23e-06,
+      "loss": 0.3694,
+      "step": 9577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.22e-06,
+      "loss": 0.1237,
+      "step": 9578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.21e-06,
+      "loss": 0.1307,
+      "step": 9579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.0314,
+      "step": 9580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1900000000000005e-06,
+      "loss": 0.6712,
+      "step": 9581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.18e-06,
+      "loss": 1.1191,
+      "step": 9582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.17e-06,
+      "loss": 0.767,
+      "step": 9583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.16e-06,
+      "loss": 1.1867,
+      "step": 9584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.15e-06,
+      "loss": 1.0443,
+      "step": 9585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.14e-06,
+      "loss": 1.4233,
+      "step": 9586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.13e-06,
+      "loss": 1.885,
+      "step": 9587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.12e-06,
+      "loss": 1.0021,
+      "step": 9588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.11e-06,
+      "loss": 0.892,
+      "step": 9589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1000000000000006e-06,
+      "loss": 0.0696,
+      "step": 9590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.09e-06,
+      "loss": 0.0456,
+      "step": 9591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.080000000000001e-06,
+      "loss": 0.4547,
+      "step": 9592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.07e-06,
+      "loss": 0.4177,
+      "step": 9593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.06e-06,
+      "loss": 1.2332,
+      "step": 9594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.05e-06,
+      "loss": 0.9642,
+      "step": 9595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.04e-06,
+      "loss": 1.1997,
+      "step": 9596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.03e-06,
+      "loss": 1.3496,
+      "step": 9597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 1.0407,
+      "step": 9598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.01e-06,
+      "loss": 1.4505,
+      "step": 9599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.8302,
+      "step": 9600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.99e-06,
+      "loss": 1.1573,
+      "step": 9601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.98e-06,
+      "loss": 0.2873,
+      "step": 9602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.97e-06,
+      "loss": 0.1186,
+      "step": 9603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.96e-06,
+      "loss": 0.3548,
+      "step": 9604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.95e-06,
+      "loss": 0.376,
+      "step": 9605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9399999999999995e-06,
+      "loss": 0.9503,
+      "step": 9606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9300000000000005e-06,
+      "loss": 0.9704,
+      "step": 9607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.92e-06,
+      "loss": 1.1432,
+      "step": 9608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.910000000000001e-06,
+      "loss": 1.3446,
+      "step": 9609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9e-06,
+      "loss": 0.9283,
+      "step": 9610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.89e-06,
+      "loss": 0.8842,
+      "step": 9611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.88e-06,
+      "loss": 0.5418,
+      "step": 9612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.87e-06,
+      "loss": 1.0888,
+      "step": 9613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.86e-06,
+      "loss": 0.5499,
+      "step": 9614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.85e-06,
+      "loss": 0.1553,
+      "step": 9615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.84e-06,
+      "loss": 0.2437,
+      "step": 9616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.830000000000001e-06,
+      "loss": 0.1139,
+      "step": 9617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.82e-06,
+      "loss": 0.7241,
+      "step": 9618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8100000000000004e-06,
+      "loss": 0.8383,
+      "step": 9619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8e-06,
+      "loss": 1.3906,
+      "step": 9620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7900000000000006e-06,
+      "loss": 1.1345,
+      "step": 9621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.9587,
+      "step": 9622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.77e-06,
+      "loss": 1.2152,
+      "step": 9623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 1.3722,
+      "step": 9624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.75e-06,
+      "loss": 1.0066,
+      "step": 9625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7400000000000006e-06,
+      "loss": 1.447,
+      "step": 9626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7300000000000003e-06,
+      "loss": 0.6507,
+      "step": 9627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.72e-06,
+      "loss": 0.4423,
+      "step": 9628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7100000000000005e-06,
+      "loss": 0.0679,
+      "step": 9629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7e-06,
+      "loss": 0.4208,
+      "step": 9630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.6900000000000002e-06,
+      "loss": 0.6233,
+      "step": 9631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.68e-06,
+      "loss": 1.56,
+      "step": 9632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.6700000000000004e-06,
+      "loss": 1.1249,
+      "step": 9633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.66e-06,
+      "loss": 0.8391,
+      "step": 9634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.6499999999999998e-06,
+      "loss": 0.9291,
+      "step": 9635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 0.9864,
+      "step": 9636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.63e-06,
+      "loss": 1.3961,
+      "step": 9637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.6200000000000005e-06,
+      "loss": 1.213,
+      "step": 9638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.61e-06,
+      "loss": 0.6785,
+      "step": 9639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.6e-06,
+      "loss": 0.0907,
+      "step": 9640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.5900000000000004e-06,
+      "loss": 0.0635,
+      "step": 9641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.58e-06,
+      "loss": 0.2088,
+      "step": 9642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.5700000000000005e-06,
+      "loss": 0.5069,
+      "step": 9643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 0.8248,
+      "step": 9644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.55e-06,
+      "loss": 1.3117,
+      "step": 9645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.5400000000000004e-06,
+      "loss": 1.037,
+      "step": 9646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.53e-06,
+      "loss": 1.4913,
+      "step": 9647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.52e-06,
+      "loss": 0.8107,
+      "step": 9648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.5100000000000003e-06,
+      "loss": 1.3071,
+      "step": 9649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 1.1276,
+      "step": 9650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.49e-06,
+      "loss": 1.0381,
+      "step": 9651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4799999999999997e-06,
+      "loss": 0.3107,
+      "step": 9652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4700000000000002e-06,
+      "loss": 0.1225,
+      "step": 9653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.46e-06,
+      "loss": 0.2573,
+      "step": 9654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4500000000000004e-06,
+      "loss": 0.4674,
+      "step": 9655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.44e-06,
+      "loss": 1.3041,
+      "step": 9656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4299999999999998e-06,
+      "loss": 1.0679,
+      "step": 9657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4200000000000003e-06,
+      "loss": 1.3346,
+      "step": 9658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.41e-06,
+      "loss": 0.9182,
+      "step": 9659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.8446,
+      "step": 9660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.39e-06,
+      "loss": 1.0342,
+      "step": 9661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.38e-06,
+      "loss": 0.9519,
+      "step": 9662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3700000000000003e-06,
+      "loss": 1.3901,
+      "step": 9663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.36e-06,
+      "loss": 0.2017,
+      "step": 9664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3500000000000005e-06,
+      "loss": 0.229,
+      "step": 9665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.34e-06,
+      "loss": 0.2134,
+      "step": 9666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3300000000000003e-06,
+      "loss": 0.2402,
+      "step": 9667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 0.7899,
+      "step": 9668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.31e-06,
+      "loss": 1.2052,
+      "step": 9669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3e-06,
+      "loss": 1.0291,
+      "step": 9670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.29e-06,
+      "loss": 1.0299,
+      "step": 9671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 1.1969,
+      "step": 9672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.27e-06,
+      "loss": 1.1508,
+      "step": 9673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2599999999999997e-06,
+      "loss": 1.3378,
+      "step": 9674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 1.0924,
+      "step": 9675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.24e-06,
+      "loss": 1.0878,
+      "step": 9676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2300000000000004e-06,
+      "loss": 0.3748,
+      "step": 9677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.22e-06,
+      "loss": 0.0764,
+      "step": 9678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2099999999999998e-06,
+      "loss": 0.1465,
+      "step": 9679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.5068,
+      "step": 9680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.19e-06,
+      "loss": 0.2986,
+      "step": 9681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 1.1886,
+      "step": 9682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.17e-06,
+      "loss": 1.0533,
+      "step": 9683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1600000000000007e-06,
+      "loss": 1.2566,
+      "step": 9684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1500000000000003e-06,
+      "loss": 1.2125,
+      "step": 9685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.14e-06,
+      "loss": 1.215,
+      "step": 9686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.13e-06,
+      "loss": 0.7076,
+      "step": 9687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.12e-06,
+      "loss": 1.633,
+      "step": 9688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.11e-06,
+      "loss": 0.466,
+      "step": 9689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1e-06,
+      "loss": 0.2291,
+      "step": 9690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.09e-06,
+      "loss": 0.009,
+      "step": 9691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.08e-06,
+      "loss": 0.0921,
+      "step": 9692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0700000000000003e-06,
+      "loss": 0.4659,
+      "step": 9693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.06e-06,
+      "loss": 0.9787,
+      "step": 9694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.05e-06,
+      "loss": 1.1609,
+      "step": 9695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.04e-06,
+      "loss": 0.9138,
+      "step": 9696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0300000000000002e-06,
+      "loss": 1.1143,
+      "step": 9697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 1.1079,
+      "step": 9698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.01e-06,
+      "loss": 1.1799,
+      "step": 9699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3e-06,
+      "loss": 0.9636,
+      "step": 9700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.99e-06,
+      "loss": 1.0582,
+      "step": 9701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.309,
+      "step": 9702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9700000000000004e-06,
+      "loss": 0.1222,
+      "step": 9703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9600000000000005e-06,
+      "loss": 0.1407,
+      "step": 9704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.95e-06,
+      "loss": 0.0297,
+      "step": 9705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 0.6291,
+      "step": 9706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.93e-06,
+      "loss": 1.3753,
+      "step": 9707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.92e-06,
+      "loss": 1.2292,
+      "step": 9708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.91e-06,
+      "loss": 1.2066,
+      "step": 9709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9e-06,
+      "loss": 1.2158,
+      "step": 9710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.89e-06,
+      "loss": 0.8492,
+      "step": 9711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.88e-06,
+      "loss": 0.9411,
+      "step": 9712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.87e-06,
+      "loss": 1.3354,
+      "step": 9713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.86e-06,
+      "loss": 0.3543,
+      "step": 9714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8500000000000002e-06,
+      "loss": 0.0861,
+      "step": 9715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8400000000000003e-06,
+      "loss": 0.2885,
+      "step": 9716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.83e-06,
+      "loss": 0.1708,
+      "step": 9717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.82e-06,
+      "loss": 0.7727,
+      "step": 9718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.81e-06,
+      "loss": 1.0949,
+      "step": 9719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.941,
+      "step": 9720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7900000000000004e-06,
+      "loss": 1.1043,
+      "step": 9721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.78e-06,
+      "loss": 0.8587,
+      "step": 9722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.77e-06,
+      "loss": 0.9478,
+      "step": 9723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 1.2509,
+      "step": 9724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 1.0436,
+      "step": 9725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.74e-06,
+      "loss": 0.9649,
+      "step": 9726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.73e-06,
+      "loss": 0.2303,
+      "step": 9727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.72e-06,
+      "loss": 0.0297,
+      "step": 9728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.71e-06,
+      "loss": 0.4852,
+      "step": 9729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7e-06,
+      "loss": 0.4984,
+      "step": 9730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.69e-06,
+      "loss": 0.3058,
+      "step": 9731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.68e-06,
+      "loss": 0.5587,
+      "step": 9732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6700000000000003e-06,
+      "loss": 0.9134,
+      "step": 9733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.66e-06,
+      "loss": 1.123,
+      "step": 9734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.65e-06,
+      "loss": 1.3245,
+      "step": 9735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.64e-06,
+      "loss": 0.8053,
+      "step": 9736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6300000000000002e-06,
+      "loss": 1.7133,
+      "step": 9737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 1.6808,
+      "step": 9738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6100000000000004e-06,
+      "loss": 0.3095,
+      "step": 9739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6e-06,
+      "loss": 0.0337,
+      "step": 9740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.59e-06,
+      "loss": 0.3474,
+      "step": 9741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.2686,
+      "step": 9742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5700000000000004e-06,
+      "loss": 0.2242,
+      "step": 9743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.56e-06,
+      "loss": 0.6842,
+      "step": 9744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.55e-06,
+      "loss": 1.4974,
+      "step": 9745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.54e-06,
+      "loss": 1.0065,
+      "step": 9746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.53e-06,
+      "loss": 0.725,
+      "step": 9747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.52e-06,
+      "loss": 1.1639,
+      "step": 9748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.51e-06,
+      "loss": 0.9965,
+      "step": 9749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5e-06,
+      "loss": 0.7701,
+      "step": 9750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.49e-06,
+      "loss": 1.2026,
+      "step": 9751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.48e-06,
+      "loss": 0.365,
+      "step": 9752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.47e-06,
+      "loss": 0.0113,
+      "step": 9753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.46e-06,
+      "loss": 0.4402,
+      "step": 9754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 0.4634,
+      "step": 9755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 0.7723,
+      "step": 9756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.43e-06,
+      "loss": 0.8435,
+      "step": 9757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.42e-06,
+      "loss": 1.4133,
+      "step": 9758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4100000000000002e-06,
+      "loss": 1.3758,
+      "step": 9759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.082,
+      "step": 9760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.3900000000000004e-06,
+      "loss": 1.1886,
+      "step": 9761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.38e-06,
+      "loss": 1.0468,
+      "step": 9762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.37e-06,
+      "loss": 1.0225,
+      "step": 9763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.36e-06,
+      "loss": 1.206,
+      "step": 9764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.35e-06,
+      "loss": 0.1404,
+      "step": 9765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.34e-06,
+      "loss": 0.0515,
+      "step": 9766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.33e-06,
+      "loss": 0.0372,
+      "step": 9767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.32e-06,
+      "loss": 0.3037,
+      "step": 9768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.31e-06,
+      "loss": 1.2617,
+      "step": 9769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.3e-06,
+      "loss": 1.1386,
+      "step": 9770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.29e-06,
+      "loss": 1.3093,
+      "step": 9771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.28e-06,
+      "loss": 0.8519,
+      "step": 9772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.2700000000000003e-06,
+      "loss": 1.0167,
+      "step": 9773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.26e-06,
+      "loss": 1.2149,
+      "step": 9774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.25e-06,
+      "loss": 0.9092,
+      "step": 9775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.24e-06,
+      "loss": 0.991,
+      "step": 9776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.2300000000000002e-06,
+      "loss": 0.1918,
+      "step": 9777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 0.1526,
+      "step": 9778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.2100000000000004e-06,
+      "loss": 0.2901,
+      "step": 9779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.2e-06,
+      "loss": 0.2008,
+      "step": 9780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.19e-06,
+      "loss": 0.8404,
+      "step": 9781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.9908,
+      "step": 9782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17e-06,
+      "loss": 1.1769,
+      "step": 9783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.16e-06,
+      "loss": 1.0274,
+      "step": 9784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1499999999999997e-06,
+      "loss": 1.2468,
+      "step": 9785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.14e-06,
+      "loss": 1.0173,
+      "step": 9786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.13e-06,
+      "loss": 1.0704,
+      "step": 9787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.12e-06,
+      "loss": 1.2944,
+      "step": 9788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.11e-06,
+      "loss": 1.1637,
+      "step": 9789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 0.6834,
+      "step": 9790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.09e-06,
+      "loss": 0.0406,
+      "step": 9791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.08e-06,
+      "loss": 0.1131,
+      "step": 9792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.07e-06,
+      "loss": 0.2311,
+      "step": 9793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.06e-06,
+      "loss": 1.1289,
+      "step": 9794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.0500000000000003e-06,
+      "loss": 1.2953,
+      "step": 9795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 1.2537,
+      "step": 9796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.03e-06,
+      "loss": 0.9668,
+      "step": 9797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.02e-06,
+      "loss": 0.9528,
+      "step": 9798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.0100000000000002e-06,
+      "loss": 1.4501,
+      "step": 9799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.4518,
+      "step": 9800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99e-06,
+      "loss": 1.1404,
+      "step": 9801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.98e-06,
+      "loss": 0.6356,
+      "step": 9802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9699999999999998e-06,
+      "loss": 0.1418,
+      "step": 9803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.96e-06,
+      "loss": 0.1662,
+      "step": 9804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95e-06,
+      "loss": 0.1637,
+      "step": 9805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.94e-06,
+      "loss": 1.0421,
+      "step": 9806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.93e-06,
+      "loss": 1.1861,
+      "step": 9807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.92e-06,
+      "loss": 1.1278,
+      "step": 9808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.91e-06,
+      "loss": 1.2508,
+      "step": 9809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9e-06,
+      "loss": 0.9613,
+      "step": 9810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8900000000000001e-06,
+      "loss": 0.9637,
+      "step": 9811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 0.7191,
+      "step": 9812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8700000000000003e-06,
+      "loss": 1.4433,
+      "step": 9813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.86e-06,
+      "loss": 0.6594,
+      "step": 9814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.85e-06,
+      "loss": 0.0379,
+      "step": 9815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.84e-06,
+      "loss": 0.3455,
+      "step": 9816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.83e-06,
+      "loss": 0.3167,
+      "step": 9817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 0.2702,
+      "step": 9818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8100000000000002e-06,
+      "loss": 1.2694,
+      "step": 9819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8e-06,
+      "loss": 0.9651,
+      "step": 9820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.79e-06,
+      "loss": 1.3737,
+      "step": 9821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 1.6668,
+      "step": 9822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7700000000000002e-06,
+      "loss": 0.7892,
+      "step": 9823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.76e-06,
+      "loss": 1.137,
+      "step": 9824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7500000000000002e-06,
+      "loss": 1.2604,
+      "step": 9825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7399999999999999e-06,
+      "loss": 1.4637,
+      "step": 9826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.73e-06,
+      "loss": 0.3092,
+      "step": 9827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.72e-06,
+      "loss": 0.0051,
+      "step": 9828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7100000000000001e-06,
+      "loss": 0.002,
+      "step": 9829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 0.2041,
+      "step": 9830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.69e-06,
+      "loss": 1.2555,
+      "step": 9831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.68e-06,
+      "loss": 0.9517,
+      "step": 9832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.67e-06,
+      "loss": 0.9885,
+      "step": 9833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 1.0312,
+      "step": 9834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.65e-06,
+      "loss": 0.7782,
+      "step": 9835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 1.098,
+      "step": 9836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6299999999999999e-06,
+      "loss": 0.9364,
+      "step": 9837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.62e-06,
+      "loss": 1.6189,
+      "step": 9838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.61e-06,
+      "loss": 1.0262,
+      "step": 9839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.0773,
+      "step": 9840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.5900000000000002e-06,
+      "loss": 0.0492,
+      "step": 9841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.5800000000000003e-06,
+      "loss": 0.4906,
+      "step": 9842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.57e-06,
+      "loss": 0.6813,
+      "step": 9843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.56e-06,
+      "loss": 0.8912,
+      "step": 9844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.55e-06,
+      "loss": 1.1461,
+      "step": 9845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.54e-06,
+      "loss": 1.0821,
+      "step": 9846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.53e-06,
+      "loss": 0.8335,
+      "step": 9847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.52e-06,
+      "loss": 0.8184,
+      "step": 9848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.5100000000000002e-06,
+      "loss": 1.1578,
+      "step": 9849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.5e-06,
+      "loss": 1.0492,
+      "step": 9850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4900000000000001e-06,
+      "loss": 1.5963,
+      "step": 9851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4800000000000002e-06,
+      "loss": 0.3266,
+      "step": 9852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4700000000000001e-06,
+      "loss": 0.4175,
+      "step": 9853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.46e-06,
+      "loss": 0.003,
+      "step": 9854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.45e-06,
+      "loss": 0.0136,
+      "step": 9855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.44e-06,
+      "loss": 0.4092,
+      "step": 9856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.43e-06,
+      "loss": 1.1228,
+      "step": 9857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4200000000000002e-06,
+      "loss": 1.1773,
+      "step": 9858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.41e-06,
+      "loss": 0.9175,
+      "step": 9859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 0.9514,
+      "step": 9860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.39e-06,
+      "loss": 1.1415,
+      "step": 9861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.8094,
+      "step": 9862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.37e-06,
+      "loss": 1.2437,
+      "step": 9863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.36e-06,
+      "loss": 0.7208,
+      "step": 9864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.35e-06,
+      "loss": 0.1911,
+      "step": 9865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.34e-06,
+      "loss": 0.0409,
+      "step": 9866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.33e-06,
+      "loss": 0.3889,
+      "step": 9867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.32e-06,
+      "loss": 0.7122,
+      "step": 9868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3100000000000002e-06,
+      "loss": 1.4292,
+      "step": 9869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3e-06,
+      "loss": 1.428,
+      "step": 9870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2900000000000001e-06,
+      "loss": 1.2219,
+      "step": 9871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.28e-06,
+      "loss": 1.6046,
+      "step": 9872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.27e-06,
+      "loss": 0.9181,
+      "step": 9873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.26e-06,
+      "loss": 0.892,
+      "step": 9874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.25e-06,
+      "loss": 0.9583,
+      "step": 9875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.24e-06,
+      "loss": 1.2444,
+      "step": 9876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.23e-06,
+      "loss": 0.4074,
+      "step": 9877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 0.013,
+      "step": 9878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.21e-06,
+      "loss": 0.027,
+      "step": 9879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.8428,
+      "step": 9880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.19e-06,
+      "loss": 0.9138,
+      "step": 9881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.18e-06,
+      "loss": 1.2915,
+      "step": 9882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.17e-06,
+      "loss": 1.354,
+      "step": 9883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.16e-06,
+      "loss": 1.0811,
+      "step": 9884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.15e-06,
+      "loss": 1.2835,
+      "step": 9885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.14e-06,
+      "loss": 1.2888,
+      "step": 9886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.13e-06,
+      "loss": 0.6575,
+      "step": 9887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.12e-06,
+      "loss": 1.2568,
+      "step": 9888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1100000000000002e-06,
+      "loss": 0.6993,
+      "step": 9889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1e-06,
+      "loss": 0.2475,
+      "step": 9890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0900000000000002e-06,
+      "loss": 0.0884,
+      "step": 9891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.08e-06,
+      "loss": 0.2315,
+      "step": 9892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.07e-06,
+      "loss": 0.3096,
+      "step": 9893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.06e-06,
+      "loss": 0.4196,
+      "step": 9894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 1.1115,
+      "step": 9895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.04e-06,
+      "loss": 1.333,
+      "step": 9896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.03e-06,
+      "loss": 1.2262,
+      "step": 9897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 1.3016,
+      "step": 9898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.01e-06,
+      "loss": 0.7424,
+      "step": 9899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.9443,
+      "step": 9900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9e-07,
+      "loss": 0.8115,
+      "step": 9901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8e-07,
+      "loss": 0.0991,
+      "step": 9902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7e-07,
+      "loss": 0.1165,
+      "step": 9903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.6e-07,
+      "loss": 0.0166,
+      "step": 9904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.5e-07,
+      "loss": 0.0749,
+      "step": 9905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 1.2857,
+      "step": 9906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.3e-07,
+      "loss": 1.2339,
+      "step": 9907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.2e-07,
+      "loss": 1.0454,
+      "step": 9908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.100000000000001e-07,
+      "loss": 1.0852,
+      "step": 9909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9e-07,
+      "loss": 1.149,
+      "step": 9910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.900000000000001e-07,
+      "loss": 0.8029,
+      "step": 9911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.8e-07,
+      "loss": 1.3525,
+      "step": 9912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.699999999999999e-07,
+      "loss": 1.2548,
+      "step": 9913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.6e-07,
+      "loss": 0.3179,
+      "step": 9914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 0.0321,
+      "step": 9915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.4e-07,
+      "loss": 0.0047,
+      "step": 9916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.300000000000001e-07,
+      "loss": 0.0828,
+      "step": 9917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 0.958,
+      "step": 9918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.1e-07,
+      "loss": 1.3567,
+      "step": 9919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.3552,
+      "step": 9920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.900000000000002e-07,
+      "loss": 1.3835,
+      "step": 9921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.8e-07,
+      "loss": 1.4341,
+      "step": 9922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.7e-07,
+      "loss": 0.9566,
+      "step": 9923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.6e-07,
+      "loss": 1.526,
+      "step": 9924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.5e-07,
+      "loss": 0.9833,
+      "step": 9925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.400000000000001e-07,
+      "loss": 0.9907,
+      "step": 9926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.3e-07,
+      "loss": 0.1815,
+      "step": 9927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.2e-07,
+      "loss": 0.1178,
+      "step": 9928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.100000000000001e-07,
+      "loss": 0.1483,
+      "step": 9929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 0.0799,
+      "step": 9930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.900000000000001e-07,
+      "loss": 0.6464,
+      "step": 9931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.8e-07,
+      "loss": 0.7083,
+      "step": 9932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.7e-07,
+      "loss": 0.918,
+      "step": 9933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.6e-07,
+      "loss": 1.5812,
+      "step": 9934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.5e-07,
+      "loss": 1.0719,
+      "step": 9935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.4e-07,
+      "loss": 0.9446,
+      "step": 9936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.3e-07,
+      "loss": 0.6562,
+      "step": 9937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.2e-07,
+      "loss": 1.2158,
+      "step": 9938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.100000000000001e-07,
+      "loss": 0.4098,
+      "step": 9939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 0.2176,
+      "step": 9940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9e-07,
+      "loss": 0.1448,
+      "step": 9941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.8e-07,
+      "loss": 0.0342,
+      "step": 9942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.7e-07,
+      "loss": 0.7055,
+      "step": 9943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.6e-07,
+      "loss": 1.4103,
+      "step": 9944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5e-07,
+      "loss": 1.4279,
+      "step": 9945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.4e-07,
+      "loss": 0.9794,
+      "step": 9946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.3e-07,
+      "loss": 1.0483,
+      "step": 9947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.2e-07,
+      "loss": 1.4308,
+      "step": 9948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.100000000000001e-07,
+      "loss": 1.1283,
+      "step": 9949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.5558,
+      "step": 9950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9e-07,
+      "loss": 1.4049,
+      "step": 9951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8e-07,
+      "loss": 0.1415,
+      "step": 9952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7000000000000005e-07,
+      "loss": 0.0532,
+      "step": 9953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6e-07,
+      "loss": 0.099,
+      "step": 9954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5e-07,
+      "loss": 0.0017,
+      "step": 9955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4e-07,
+      "loss": 0.2194,
+      "step": 9956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.3e-07,
+      "loss": 1.1154,
+      "step": 9957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.2e-07,
+      "loss": 1.3125,
+      "step": 9958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1000000000000004e-07,
+      "loss": 1.4924,
+      "step": 9959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.0606,
+      "step": 9960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9e-07,
+      "loss": 1.4376,
+      "step": 9961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8e-07,
+      "loss": 0.6661,
+      "step": 9962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7000000000000006e-07,
+      "loss": 0.9533,
+      "step": 9963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.6e-07,
+      "loss": 0.8728,
+      "step": 9964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 0.4738,
+      "step": 9965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4e-07,
+      "loss": 0.0994,
+      "step": 9966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3e-07,
+      "loss": 0.1524,
+      "step": 9967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2e-07,
+      "loss": 0.7164,
+      "step": 9968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1e-07,
+      "loss": 1.2366,
+      "step": 9969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 0.9918,
+      "step": 9970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9e-07,
+      "loss": 1.4258,
+      "step": 9971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8e-07,
+      "loss": 1.4093,
+      "step": 9972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7e-07,
+      "loss": 1.1264,
+      "step": 9973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6e-07,
+      "loss": 1.2581,
+      "step": 9974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 1.1629,
+      "step": 9975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4e-07,
+      "loss": 1.0256,
+      "step": 9976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.3e-07,
+      "loss": 0.4866,
+      "step": 9977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.2e-07,
+      "loss": 0.2356,
+      "step": 9978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1e-07,
+      "loss": 0.0773,
+      "step": 9979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 0.3254,
+      "step": 9980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9e-07,
+      "loss": 0.4097,
+      "step": 9981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8e-07,
+      "loss": 0.9222,
+      "step": 9982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7e-07,
+      "loss": 1.0312,
+      "step": 9983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6e-07,
+      "loss": 1.3766,
+      "step": 9984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 0.8438,
+      "step": 9985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4e-07,
+      "loss": 1.2565,
+      "step": 9986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3e-07,
+      "loss": 1.0697,
+      "step": 9987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2e-07,
+      "loss": 1.3603,
+      "step": 9988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1e-07,
+      "loss": 0.7208,
+      "step": 9989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 0.1299,
+      "step": 9990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9e-08,
+      "loss": 0.1473,
+      "step": 9991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8e-08,
+      "loss": 0.6326,
+      "step": 9992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7e-08,
+      "loss": 1.1753,
+      "step": 9993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6e-08,
+      "loss": 0.8024,
+      "step": 9994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 1.4001,
+      "step": 9995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4e-08,
+      "loss": 1.0274,
+      "step": 9996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3e-08,
+      "loss": 1.1238,
+      "step": 9997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-08,
+      "loss": 0.8862,
+      "step": 9998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-08,
+      "loss": 0.8835,
+      "step": 9999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0,
+      "loss": 0.9368,
+      "step": 10000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 10000,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "total_flos": 1.2439635835496694e+18,
+  "trial_name": null,
+  "trial_params": null
+}