diff --git "a/checkpoint-2400/trainer_state.json" "b/checkpoint-2400/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2400/trainer_state.json"
@@ -0,0 +1,14416 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.09584313673288052,
+  "global_step": 2400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 1.0488,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 1.1027,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 1.2865,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.0681,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 1.1824,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.1611,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 1.0395,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.1625,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 1.1697,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.233,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1e-06,
+      "loss": 1.1196,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.028,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3e-06,
+      "loss": 1.079,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 1.161,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 1.1258,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.1793,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 1.2195,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 1.0552,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 0.9736,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.2038,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 1.1958,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-06,
+      "loss": 0.9723,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 1.0598,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.8501,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 0.8728,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-06,
+      "loss": 1.1256,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 0.9702,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.1258,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9e-06,
+      "loss": 1.0507,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 0.9789,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 0.8359,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.9653,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 0.8996,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 1.0018,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-06,
+      "loss": 0.8889,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.9844,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7e-06,
+      "loss": 0.9551,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.8327,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 0.8914,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.8989,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1e-06,
+      "loss": 0.703,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.884,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3e-06,
+      "loss": 0.7519,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-06,
+      "loss": 0.8547,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-06,
+      "loss": 0.8231,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.7351,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7e-06,
+      "loss": 0.7906,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.7365,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 0.7611,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 0.6853,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.1e-06,
+      "loss": 0.6458,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2e-06,
+      "loss": 0.712,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 0.6755,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 0.6172,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.6864,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.6047,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.7e-06,
+      "loss": 0.5952,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.8e-06,
+      "loss": 0.6368,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9e-06,
+      "loss": 0.5363,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.5781,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.1e-06,
+      "loss": 0.6068,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 0.5856,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 0.5719,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.5862,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.5424,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 0.4901,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 0.4572,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.4476,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.9e-06,
+      "loss": 0.5266,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-06,
+      "loss": 0.5051,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 0.4577,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.4483,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.3e-06,
+      "loss": 0.5268,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4e-06,
+      "loss": 0.4774,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.3889,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.5319,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.7e-06,
+      "loss": 0.4035,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 0.5946,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.9e-06,
+      "loss": 0.4474,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.4996,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.1e-06,
+      "loss": 0.6067,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.2e-06,
+      "loss": 0.4301,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.3e-06,
+      "loss": 0.4119,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.4311,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.5e-06,
+      "loss": 0.4479,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.6e-06,
+      "loss": 0.3555,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 0.4877,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.8e-06,
+      "loss": 0.3912,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 0.3578,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 0.3706,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 0.3889,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.4256,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.3e-06,
+      "loss": 0.3372,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.4e-06,
+      "loss": 0.4404,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-06,
+      "loss": 0.3325,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.4104,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.7e-06,
+      "loss": 0.4069,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.3977,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.9e-06,
+      "loss": 0.47,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.3292,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999866702212744e-06,
+      "loss": 0.4137,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999733404425488e-06,
+      "loss": 0.3597,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999600106638231e-06,
+      "loss": 0.3905,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999466808850974e-06,
+      "loss": 0.3795,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999333511063717e-06,
+      "loss": 0.3564,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999200213276461e-06,
+      "loss": 0.4117,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999066915489204e-06,
+      "loss": 0.3322,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.998933617701947e-06,
+      "loss": 0.3209,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99880031991469e-06,
+      "loss": 0.2899,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.998667022127434e-06,
+      "loss": 0.3324,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.998533724340177e-06,
+      "loss": 0.4101,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99840042655292e-06,
+      "loss": 0.4075,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.998267128765663e-06,
+      "loss": 0.2901,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.998133830978407e-06,
+      "loss": 0.3618,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99800053319115e-06,
+      "loss": 0.3754,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.997867235403893e-06,
+      "loss": 0.2869,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.997733937616636e-06,
+      "loss": 0.4456,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99760063982938e-06,
+      "loss": 0.3589,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.997467342042123e-06,
+      "loss": 0.3046,
+      "step": 119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.997334044254866e-06,
+      "loss": 0.3358,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.997200746467608e-06,
+      "loss": 0.3456,
+      "step": 121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.997067448680353e-06,
+      "loss": 0.3451,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.996934150893096e-06,
+      "loss": 0.3562,
+      "step": 123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.996800853105839e-06,
+      "loss": 0.301,
+      "step": 124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.996667555318581e-06,
+      "loss": 0.379,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996534257531326e-06,
+      "loss": 0.3841,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996400959744069e-06,
+      "loss": 0.3027,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996267661956812e-06,
+      "loss": 0.3442,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996134364169554e-06,
+      "loss": 0.3396,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996001066382299e-06,
+      "loss": 0.3589,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995867768595042e-06,
+      "loss": 0.3797,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995734470807785e-06,
+      "loss": 0.3317,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995601173020527e-06,
+      "loss": 0.3804,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995467875233272e-06,
+      "loss": 0.3244,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995334577446015e-06,
+      "loss": 0.3035,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995201279658758e-06,
+      "loss": 0.3823,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995067981871502e-06,
+      "loss": 0.3087,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994934684084245e-06,
+      "loss": 0.3275,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994801386296988e-06,
+      "loss": 0.318,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99466808850973e-06,
+      "loss": 0.3507,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994534790722475e-06,
+      "loss": 0.3125,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994401492935218e-06,
+      "loss": 0.28,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99426819514796e-06,
+      "loss": 0.2708,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994134897360705e-06,
+      "loss": 0.2941,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994001599573448e-06,
+      "loss": 0.339,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99386830178619e-06,
+      "loss": 0.2508,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.993735003998934e-06,
+      "loss": 0.2983,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.993601706211678e-06,
+      "loss": 0.3023,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.993468408424421e-06,
+      "loss": 0.3689,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.993335110637164e-06,
+      "loss": 0.2918,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.993201812849908e-06,
+      "loss": 0.3157,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.993068515062651e-06,
+      "loss": 0.3471,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.992935217275394e-06,
+      "loss": 0.2485,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.992801919488137e-06,
+      "loss": 0.2765,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.992668621700881e-06,
+      "loss": 0.2795,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.992535323913624e-06,
+      "loss": 0.3035,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.992402026126367e-06,
+      "loss": 0.3065,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.992268728339111e-06,
+      "loss": 0.2798,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.992135430551854e-06,
+      "loss": 0.2647,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.992002132764597e-06,
+      "loss": 0.3191,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99186883497734e-06,
+      "loss": 0.2611,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.991735537190084e-06,
+      "loss": 0.2834,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.991602239402827e-06,
+      "loss": 0.3579,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99146894161557e-06,
+      "loss": 0.2927,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.991335643828315e-06,
+      "loss": 0.3214,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.991202346041057e-06,
+      "loss": 0.2955,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9910690482538e-06,
+      "loss": 0.304,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.990935750466543e-06,
+      "loss": 0.3134,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.990802452679288e-06,
+      "loss": 0.2993,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99066915489203e-06,
+      "loss": 0.2541,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.990535857104773e-06,
+      "loss": 0.2284,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.990402559317516e-06,
+      "loss": 0.2952,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99026926153026e-06,
+      "loss": 0.258,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.990135963743003e-06,
+      "loss": 0.2711,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.990002665955746e-06,
+      "loss": 0.2751,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.989869368168489e-06,
+      "loss": 0.2949,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.989736070381233e-06,
+      "loss": 0.2647,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.989602772593976e-06,
+      "loss": 0.2864,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.989469474806719e-06,
+      "loss": 0.2175,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.989336177019462e-06,
+      "loss": 0.2628,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.989202879232206e-06,
+      "loss": 0.3199,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98906958144495e-06,
+      "loss": 0.2391,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.988936283657692e-06,
+      "loss": 0.2495,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.988802985870435e-06,
+      "loss": 0.2704,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98866968808318e-06,
+      "loss": 0.2108,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.988536390295922e-06,
+      "loss": 0.3196,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.988403092508665e-06,
+      "loss": 0.276,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.988269794721408e-06,
+      "loss": 0.3185,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.988136496934152e-06,
+      "loss": 0.234,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.988003199146895e-06,
+      "loss": 0.2377,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.987869901359638e-06,
+      "loss": 0.2593,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98773660357238e-06,
+      "loss": 0.3055,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.987603305785125e-06,
+      "loss": 0.2497,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.987470007997868e-06,
+      "loss": 0.2585,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.987336710210611e-06,
+      "loss": 0.2356,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.987203412423354e-06,
+      "loss": 0.2526,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.987070114636098e-06,
+      "loss": 0.306,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.986936816848841e-06,
+      "loss": 0.2766,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.986803519061584e-06,
+      "loss": 0.2694,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.986670221274327e-06,
+      "loss": 0.2712,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.986536923487071e-06,
+      "loss": 0.2296,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.986403625699814e-06,
+      "loss": 0.2322,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.986270327912557e-06,
+      "loss": 0.308,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9861370301253e-06,
+      "loss": 0.2734,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.986003732338044e-06,
+      "loss": 0.373,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.985870434550787e-06,
+      "loss": 0.2614,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98573713676353e-06,
+      "loss": 0.2577,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.985603838976273e-06,
+      "loss": 0.2611,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.985470541189017e-06,
+      "loss": 0.2841,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98533724340176e-06,
+      "loss": 0.2529,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.985203945614503e-06,
+      "loss": 0.3231,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.985070647827246e-06,
+      "loss": 0.2378,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98493735003999e-06,
+      "loss": 0.2943,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.984804052252733e-06,
+      "loss": 0.2543,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.984670754465476e-06,
+      "loss": 0.2852,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98453745667822e-06,
+      "loss": 0.2654,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.984404158890963e-06,
+      "loss": 0.2501,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.984270861103706e-06,
+      "loss": 0.2566,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.984137563316449e-06,
+      "loss": 0.2683,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.984004265529193e-06,
+      "loss": 0.2632,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.983870967741936e-06,
+      "loss": 0.2654,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.983737669954679e-06,
+      "loss": 0.2296,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.983604372167422e-06,
+      "loss": 0.2424,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.983471074380166e-06,
+      "loss": 0.2829,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.983337776592909e-06,
+      "loss": 0.2626,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.983204478805652e-06,
+      "loss": 0.2232,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.983071181018395e-06,
+      "loss": 0.2274,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98293788323114e-06,
+      "loss": 0.314,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.982804585443882e-06,
+      "loss": 0.2423,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.982671287656625e-06,
+      "loss": 0.2245,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98253798986937e-06,
+      "loss": 0.2669,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.982404692082112e-06,
+      "loss": 0.2575,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.982271394294855e-06,
+      "loss": 0.2434,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.982138096507598e-06,
+      "loss": 0.2384,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.982004798720342e-06,
+      "loss": 0.2143,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.981871500933085e-06,
+      "loss": 0.2146,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.981738203145828e-06,
+      "loss": 0.2148,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.981604905358573e-06,
+      "loss": 0.2109,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.981471607571315e-06,
+      "loss": 0.207,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.981338309784058e-06,
+      "loss": 0.2262,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.981205011996801e-06,
+      "loss": 0.2663,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.981071714209545e-06,
+      "loss": 0.2131,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.980938416422288e-06,
+      "loss": 0.2106,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.980805118635031e-06,
+      "loss": 0.2212,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.980671820847776e-06,
+      "loss": 0.2108,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.980538523060518e-06,
+      "loss": 0.3081,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.980405225273261e-06,
+      "loss": 0.2827,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.980271927486004e-06,
+      "loss": 0.2791,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.980138629698749e-06,
+      "loss": 0.2734,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.980005331911491e-06,
+      "loss": 0.244,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.979872034124234e-06,
+      "loss": 0.2329,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.979738736336979e-06,
+      "loss": 0.2721,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.979605438549722e-06,
+      "loss": 0.1978,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.979472140762464e-06,
+      "loss": 0.2301,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.979338842975207e-06,
+      "loss": 0.171,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.979205545187952e-06,
+      "loss": 0.2484,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.979072247400695e-06,
+      "loss": 0.1941,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.978938949613437e-06,
+      "loss": 0.298,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97880565182618e-06,
+      "loss": 0.2461,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.978672354038925e-06,
+      "loss": 0.2526,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.978539056251668e-06,
+      "loss": 0.1747,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97840575846441e-06,
+      "loss": 0.2062,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.978272460677153e-06,
+      "loss": 0.2439,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.978139162889898e-06,
+      "loss": 0.273,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97800586510264e-06,
+      "loss": 0.2838,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.977872567315383e-06,
+      "loss": 0.217,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.977739269528126e-06,
+      "loss": 0.2484,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97760597174087e-06,
+      "loss": 0.2254,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.977472673953613e-06,
+      "loss": 0.2451,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.977339376166356e-06,
+      "loss": 0.2048,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9772060783791e-06,
+      "loss": 0.3033,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.977072780591844e-06,
+      "loss": 0.2393,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.976939482804586e-06,
+      "loss": 0.2174,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97680618501733e-06,
+      "loss": 0.2839,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.976672887230074e-06,
+      "loss": 0.2567,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.976539589442817e-06,
+      "loss": 0.2696,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97640629165556e-06,
+      "loss": 0.265,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.976272993868302e-06,
+      "loss": 0.2521,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.976139696081047e-06,
+      "loss": 0.2795,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97600639829379e-06,
+      "loss": 0.2528,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.975873100506532e-06,
+      "loss": 0.2622,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.975739802719275e-06,
+      "loss": 0.2223,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97560650493202e-06,
+      "loss": 0.1791,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.975473207144763e-06,
+      "loss": 0.1852,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.975339909357505e-06,
+      "loss": 0.208,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.975206611570248e-06,
+      "loss": 0.2116,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.975073313782993e-06,
+      "loss": 0.2073,
+      "step": 287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.974940015995735e-06,
+      "loss": 0.2499,
+      "step": 288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.974806718208478e-06,
+      "loss": 0.2036,
+      "step": 289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.974673420421221e-06,
+      "loss": 0.2002,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.974540122633966e-06,
+      "loss": 0.2174,
+      "step": 291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.974406824846708e-06,
+      "loss": 0.2696,
+      "step": 292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.974273527059451e-06,
+      "loss": 0.2361,
+      "step": 293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.974140229272194e-06,
+      "loss": 0.2007,
+      "step": 294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.974006931484939e-06,
+      "loss": 0.2188,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.973873633697681e-06,
+      "loss": 0.2078,
+      "step": 296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.973740335910424e-06,
+      "loss": 0.2195,
+      "step": 297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.973607038123167e-06,
+      "loss": 0.1892,
+      "step": 298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.973473740335912e-06,
+      "loss": 0.2222,
+      "step": 299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.973340442548654e-06,
+      "loss": 0.2315,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.973207144761397e-06,
+      "loss": 0.2185,
+      "step": 301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97307384697414e-06,
+      "loss": 0.2148,
+      "step": 302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.972940549186885e-06,
+      "loss": 0.2228,
+      "step": 303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.972807251399627e-06,
+      "loss": 0.1759,
+      "step": 304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97267395361237e-06,
+      "loss": 0.1981,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.972540655825113e-06,
+      "loss": 0.2366,
+      "step": 306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.972407358037858e-06,
+      "loss": 0.2247,
+      "step": 307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9722740602506e-06,
+      "loss": 0.2328,
+      "step": 308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.972140762463343e-06,
+      "loss": 0.2712,
+      "step": 309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.972007464676086e-06,
+      "loss": 0.2111,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97187416688883e-06,
+      "loss": 0.2068,
+      "step": 311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.971740869101573e-06,
+      "loss": 0.1966,
+      "step": 312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.971607571314316e-06,
+      "loss": 0.2331,
+      "step": 313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.971474273527059e-06,
+      "loss": 0.1854,
+      "step": 314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.971340975739803e-06,
+      "loss": 0.2479,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.971207677952546e-06,
+      "loss": 0.2219,
+      "step": 316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.971074380165289e-06,
+      "loss": 0.1735,
+      "step": 317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.970941082378034e-06,
+      "loss": 0.2171,
+      "step": 318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.970807784590776e-06,
+      "loss": 0.1955,
+      "step": 319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97067448680352e-06,
+      "loss": 0.1766,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.970541189016262e-06,
+      "loss": 0.2598,
+      "step": 321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.970407891229007e-06,
+      "loss": 0.1983,
+      "step": 322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.97027459344175e-06,
+      "loss": 0.2088,
+      "step": 323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.970141295654492e-06,
+      "loss": 0.2033,
+      "step": 324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.970007997867237e-06,
+      "loss": 0.2359,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.96987470007998e-06,
+      "loss": 0.2355,
+      "step": 326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.969741402292722e-06,
+      "loss": 0.2323,
+      "step": 327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.969608104505465e-06,
+      "loss": 0.2265,
+      "step": 328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.96947480671821e-06,
+      "loss": 0.2111,
+      "step": 329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.969341508930953e-06,
+      "loss": 0.2175,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.969208211143695e-06,
+      "loss": 0.2193,
+      "step": 331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.96907491335644e-06,
+      "loss": 0.2515,
+      "step": 332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.968941615569183e-06,
+      "loss": 0.1859,
+      "step": 333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.968808317781925e-06,
+      "loss": 0.236,
+      "step": 334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.968675019994668e-06,
+      "loss": 0.2299,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.968541722207413e-06,
+      "loss": 0.2232,
+      "step": 336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.968408424420156e-06,
+      "loss": 0.1819,
+      "step": 337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.968275126632898e-06,
+      "loss": 0.1866,
+      "step": 338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.968141828845643e-06,
+      "loss": 0.2574,
+      "step": 339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.968008531058386e-06,
+      "loss": 0.2325,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.967875233271129e-06,
+      "loss": 0.2212,
+      "step": 341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.967741935483871e-06,
+      "loss": 0.2192,
+      "step": 342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.967608637696616e-06,
+      "loss": 0.1918,
+      "step": 343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.967475339909359e-06,
+      "loss": 0.2243,
+      "step": 344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.967342042122102e-06,
+      "loss": 0.2391,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.967208744334846e-06,
+      "loss": 0.2287,
+      "step": 346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.967075446547589e-06,
+      "loss": 0.192,
+      "step": 347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.966942148760332e-06,
+      "loss": 0.2215,
+      "step": 348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.966808850973075e-06,
+      "loss": 0.1705,
+      "step": 349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.966675553185819e-06,
+      "loss": 0.1775,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.966542255398562e-06,
+      "loss": 0.2048,
+      "step": 351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.966408957611305e-06,
+      "loss": 0.2141,
+      "step": 352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.966275659824048e-06,
+      "loss": 0.2265,
+      "step": 353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.966142362036792e-06,
+      "loss": 0.2002,
+      "step": 354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.966009064249535e-06,
+      "loss": 0.2152,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.965875766462278e-06,
+      "loss": 0.1991,
+      "step": 356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.96574246867502e-06,
+      "loss": 0.1985,
+      "step": 357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.965609170887765e-06,
+      "loss": 0.2076,
+      "step": 358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.965475873100508e-06,
+      "loss": 0.2167,
+      "step": 359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.96534257531325e-06,
+      "loss": 0.239,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.965209277525993e-06,
+      "loss": 0.1731,
+      "step": 361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.965075979738738e-06,
+      "loss": 0.2006,
+      "step": 362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.96494268195148e-06,
+      "loss": 0.1882,
+      "step": 363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.964809384164224e-06,
+      "loss": 0.2155,
+      "step": 364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.964676086376966e-06,
+      "loss": 0.1793,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.964542788589711e-06,
+      "loss": 0.2205,
+      "step": 366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.964409490802454e-06,
+      "loss": 0.1717,
+      "step": 367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.964276193015197e-06,
+      "loss": 0.1876,
+      "step": 368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.96414289522794e-06,
+      "loss": 0.1951,
+      "step": 369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.964009597440684e-06,
+      "loss": 0.1272,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.963876299653427e-06,
+      "loss": 0.1715,
+      "step": 371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.96374300186617e-06,
+      "loss": 0.1494,
+      "step": 372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.963609704078912e-06,
+      "loss": 0.1957,
+      "step": 373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.963476406291657e-06,
+      "loss": 0.2448,
+      "step": 374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9633431085044e-06,
+      "loss": 0.1905,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.963209810717143e-06,
+      "loss": 0.2125,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.963076512929885e-06,
+      "loss": 0.2128,
+      "step": 377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.96294321514263e-06,
+      "loss": 0.2136,
+      "step": 378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.962809917355373e-06,
+      "loss": 0.216,
+      "step": 379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.962676619568116e-06,
+      "loss": 0.1928,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.962543321780858e-06,
+      "loss": 0.2432,
+      "step": 381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.962410023993603e-06,
+      "loss": 0.1724,
+      "step": 382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.962276726206346e-06,
+      "loss": 0.2409,
+      "step": 383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.962143428419088e-06,
+      "loss": 0.1632,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.962010130631833e-06,
+      "loss": 0.1845,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.961876832844576e-06,
+      "loss": 0.2127,
+      "step": 386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.961743535057319e-06,
+      "loss": 0.2163,
+      "step": 387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.961610237270061e-06,
+      "loss": 0.2409,
+      "step": 388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.961476939482806e-06,
+      "loss": 0.2279,
+      "step": 389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.961343641695549e-06,
+      "loss": 0.2303,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.961210343908292e-06,
+      "loss": 0.2285,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.961077046121034e-06,
+      "loss": 0.1884,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.960943748333779e-06,
+      "loss": 0.1826,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.960810450546522e-06,
+      "loss": 0.192,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.960677152759265e-06,
+      "loss": 0.2114,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.960543854972007e-06,
+      "loss": 0.1949,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.960410557184752e-06,
+      "loss": 0.1871,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.960277259397495e-06,
+      "loss": 0.1848,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.960143961610238e-06,
+      "loss": 0.1901,
+      "step": 399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.96001066382298e-06,
+      "loss": 0.2432,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.959877366035725e-06,
+      "loss": 0.2295,
+      "step": 401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.959744068248468e-06,
+      "loss": 0.2045,
+      "step": 402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95961077046121e-06,
+      "loss": 0.2009,
+      "step": 403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.959477472673953e-06,
+      "loss": 0.2044,
+      "step": 404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.959344174886698e-06,
+      "loss": 0.2251,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95921087709944e-06,
+      "loss": 0.2486,
+      "step": 406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.959077579312183e-06,
+      "loss": 0.1897,
+      "step": 407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.958944281524926e-06,
+      "loss": 0.2208,
+      "step": 408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95881098373767e-06,
+      "loss": 0.2308,
+      "step": 409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.958677685950414e-06,
+      "loss": 0.2256,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.958544388163156e-06,
+      "loss": 0.1759,
+      "step": 411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.958411090375901e-06,
+      "loss": 0.2234,
+      "step": 412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.958277792588644e-06,
+      "loss": 0.2309,
+      "step": 413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.958144494801387e-06,
+      "loss": 0.1627,
+      "step": 414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95801119701413e-06,
+      "loss": 0.1762,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.957877899226874e-06,
+      "loss": 0.2459,
+      "step": 416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.957744601439617e-06,
+      "loss": 0.2292,
+      "step": 417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95761130365236e-06,
+      "loss": 0.1528,
+      "step": 418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.957478005865104e-06,
+      "loss": 0.204,
+      "step": 419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.957344708077847e-06,
+      "loss": 0.2003,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95721141029059e-06,
+      "loss": 0.166,
+      "step": 421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.957078112503333e-06,
+      "loss": 0.1674,
+      "step": 422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.956944814716077e-06,
+      "loss": 0.1642,
+      "step": 423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95681151692882e-06,
+      "loss": 0.1706,
+      "step": 424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.956678219141563e-06,
+      "loss": 0.1557,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.956544921354307e-06,
+      "loss": 0.1716,
+      "step": 426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95641162356705e-06,
+      "loss": 0.1799,
+      "step": 427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.956278325779793e-06,
+      "loss": 0.1879,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.956145027992536e-06,
+      "loss": 0.2326,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95601173020528e-06,
+      "loss": 0.2191,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.955878432418023e-06,
+      "loss": 0.2121,
+      "step": 431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.955745134630766e-06,
+      "loss": 0.2485,
+      "step": 432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95561183684351e-06,
+      "loss": 0.2327,
+      "step": 433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.955478539056253e-06,
+      "loss": 0.1954,
+      "step": 434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.955345241268996e-06,
+      "loss": 0.1432,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.955211943481739e-06,
+      "loss": 0.2018,
+      "step": 436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.955078645694483e-06,
+      "loss": 0.1769,
+      "step": 437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.954945347907226e-06,
+      "loss": 0.205,
+      "step": 438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.954812050119969e-06,
+      "loss": 0.1834,
+      "step": 439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.954678752332713e-06,
+      "loss": 0.1641,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.954545454545456e-06,
+      "loss": 0.2168,
+      "step": 441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.954412156758199e-06,
+      "loss": 0.2191,
+      "step": 442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.954278858970942e-06,
+      "loss": 0.236,
+      "step": 443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.954145561183686e-06,
+      "loss": 0.2504,
+      "step": 444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95401226339643e-06,
+      "loss": 0.1767,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.953878965609172e-06,
+      "loss": 0.2016,
+      "step": 446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.953745667821915e-06,
+      "loss": 0.139,
+      "step": 447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95361237003466e-06,
+      "loss": 0.1341,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.953479072247402e-06,
+      "loss": 0.2284,
+      "step": 449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.953345774460145e-06,
+      "loss": 0.168,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.953212476672888e-06,
+      "loss": 0.1672,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.953079178885632e-06,
+      "loss": 0.238,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.952945881098375e-06,
+      "loss": 0.1706,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.952812583311118e-06,
+      "loss": 0.1689,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95267928552386e-06,
+      "loss": 0.2128,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.952545987736605e-06,
+      "loss": 0.2185,
+      "step": 456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.952412689949348e-06,
+      "loss": 0.1851,
+      "step": 457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.952279392162091e-06,
+      "loss": 0.1793,
+      "step": 458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.952146094374834e-06,
+      "loss": 0.1969,
+      "step": 459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.952012796587578e-06,
+      "loss": 0.17,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.951879498800321e-06,
+      "loss": 0.1812,
+      "step": 461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.951746201013064e-06,
+      "loss": 0.1882,
+      "step": 462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.951612903225807e-06,
+      "loss": 0.1678,
+      "step": 463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.951479605438551e-06,
+      "loss": 0.2108,
+      "step": 464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.951346307651294e-06,
+      "loss": 0.1847,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.951213009864037e-06,
+      "loss": 0.1964,
+      "step": 466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95107971207678e-06,
+      "loss": 0.2409,
+      "step": 467
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.950946414289524e-06,
+      "loss": 0.1728,
+      "step": 468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.950813116502267e-06,
+      "loss": 0.1888,
+      "step": 469
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95067981871501e-06,
+      "loss": 0.201,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.950546520927753e-06,
+      "loss": 0.1965,
+      "step": 471
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.950413223140497e-06,
+      "loss": 0.1726,
+      "step": 472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95027992535324e-06,
+      "loss": 0.1448,
+      "step": 473
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.950146627565983e-06,
+      "loss": 0.1819,
+      "step": 474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.950013329778726e-06,
+      "loss": 0.1928,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.94988003199147e-06,
+      "loss": 0.1982,
+      "step": 476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.949746734204213e-06,
+      "loss": 0.2022,
+      "step": 477
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.949613436416956e-06,
+      "loss": 0.1753,
+      "step": 478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.949480138629699e-06,
+      "loss": 0.2047,
+      "step": 479
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.949346840842443e-06,
+      "loss": 0.1592,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.949213543055186e-06,
+      "loss": 0.2255,
+      "step": 481
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.949080245267929e-06,
+      "loss": 0.1797,
+      "step": 482
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.948946947480672e-06,
+      "loss": 0.2411,
+      "step": 483
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.948813649693416e-06,
+      "loss": 0.2061,
+      "step": 484
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.948680351906159e-06,
+      "loss": 0.2177,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.948547054118902e-06,
+      "loss": 0.2384,
+      "step": 486
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.948413756331645e-06,
+      "loss": 0.1846,
+      "step": 487
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.948280458544389e-06,
+      "loss": 0.241,
+      "step": 488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.948147160757132e-06,
+      "loss": 0.2319,
+      "step": 489
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.948013862969875e-06,
+      "loss": 0.1587,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.947880565182618e-06,
+      "loss": 0.1702,
+      "step": 491
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.947747267395362e-06,
+      "loss": 0.183,
+      "step": 492
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.947613969608105e-06,
+      "loss": 0.1297,
+      "step": 493
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.947480671820848e-06,
+      "loss": 0.2208,
+      "step": 494
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.947347374033592e-06,
+      "loss": 0.2033,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.947214076246335e-06,
+      "loss": 0.179,
+      "step": 496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.947080778459078e-06,
+      "loss": 0.204,
+      "step": 497
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.94694748067182e-06,
+      "loss": 0.1981,
+      "step": 498
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.946814182884565e-06,
+      "loss": 0.175,
+      "step": 499
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.946680885097308e-06,
+      "loss": 0.239,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.94654758731005e-06,
+      "loss": 0.1766,
+      "step": 501
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.946414289522794e-06,
+      "loss": 0.2208,
+      "step": 502
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.946280991735538e-06,
+      "loss": 0.1757,
+      "step": 503
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.946147693948281e-06,
+      "loss": 0.1888,
+      "step": 504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.946014396161024e-06,
+      "loss": 0.1662,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.945881098373768e-06,
+      "loss": 0.167,
+      "step": 506
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.945747800586511e-06,
+      "loss": 0.179,
+      "step": 507
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.945614502799254e-06,
+      "loss": 0.2478,
+      "step": 508
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.945481205011997e-06,
+      "loss": 0.1778,
+      "step": 509
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.945347907224741e-06,
+      "loss": 0.1596,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.945214609437484e-06,
+      "loss": 0.1743,
+      "step": 511
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.945081311650227e-06,
+      "loss": 0.1909,
+      "step": 512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.944948013862971e-06,
+      "loss": 0.2093,
+      "step": 513
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.944814716075714e-06,
+      "loss": 0.2196,
+      "step": 514
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.944681418288457e-06,
+      "loss": 0.1832,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9445481205012e-06,
+      "loss": 0.1884,
+      "step": 516
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.944414822713944e-06,
+      "loss": 0.1987,
+      "step": 517
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.944281524926687e-06,
+      "loss": 0.1826,
+      "step": 518
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.94414822713943e-06,
+      "loss": 0.2217,
+      "step": 519
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.944014929352175e-06,
+      "loss": 0.1992,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.943881631564917e-06,
+      "loss": 0.1945,
+      "step": 521
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.94374833377766e-06,
+      "loss": 0.1686,
+      "step": 522
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.943615035990403e-06,
+      "loss": 0.1931,
+      "step": 523
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.943481738203148e-06,
+      "loss": 0.1533,
+      "step": 524
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.94334844041589e-06,
+      "loss": 0.211,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.943215142628633e-06,
+      "loss": 0.2226,
+      "step": 526
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.943081844841378e-06,
+      "loss": 0.1583,
+      "step": 527
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.94294854705412e-06,
+      "loss": 0.2328,
+      "step": 528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.942815249266863e-06,
+      "loss": 0.2064,
+      "step": 529
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.942681951479606e-06,
+      "loss": 0.2281,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.94254865369235e-06,
+      "loss": 0.1782,
+      "step": 531
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.942415355905093e-06,
+      "loss": 0.1613,
+      "step": 532
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.942282058117836e-06,
+      "loss": 0.1661,
+      "step": 533
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.942148760330579e-06,
+      "loss": 0.1514,
+      "step": 534
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.942015462543324e-06,
+      "loss": 0.2058,
+      "step": 535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.941882164756066e-06,
+      "loss": 0.2076,
+      "step": 536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.94174886696881e-06,
+      "loss": 0.1974,
+      "step": 537
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.941615569181552e-06,
+      "loss": 0.2201,
+      "step": 538
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.941482271394297e-06,
+      "loss": 0.1856,
+      "step": 539
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.94134897360704e-06,
+      "loss": 0.2188,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.941215675819782e-06,
+      "loss": 0.1779,
+      "step": 541
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.941082378032525e-06,
+      "loss": 0.1967,
+      "step": 542
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.94094908024527e-06,
+      "loss": 0.1733,
+      "step": 543
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.940815782458012e-06,
+      "loss": 0.1181,
+      "step": 544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.940682484670755e-06,
+      "loss": 0.2046,
+      "step": 545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.940549186883498e-06,
+      "loss": 0.1776,
+      "step": 546
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.940415889096243e-06,
+      "loss": 0.1466,
+      "step": 547
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.940282591308985e-06,
+      "loss": 0.2233,
+      "step": 548
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.940149293521728e-06,
+      "loss": 0.1821,
+      "step": 549
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.940015995734471e-06,
+      "loss": 0.1532,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.939882697947215e-06,
+      "loss": 0.1437,
+      "step": 551
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.939749400159958e-06,
+      "loss": 0.1371,
+      "step": 552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.939616102372701e-06,
+      "loss": 0.1858,
+      "step": 553
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.939482804585446e-06,
+      "loss": 0.1922,
+      "step": 554
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.939349506798188e-06,
+      "loss": 0.1766,
+      "step": 555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.939216209010931e-06,
+      "loss": 0.1887,
+      "step": 556
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.939082911223674e-06,
+      "loss": 0.1168,
+      "step": 557
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.938949613436419e-06,
+      "loss": 0.2044,
+      "step": 558
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.938816315649161e-06,
+      "loss": 0.1898,
+      "step": 559
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.938683017861904e-06,
+      "loss": 0.1608,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.938549720074647e-06,
+      "loss": 0.1969,
+      "step": 561
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.938416422287392e-06,
+      "loss": 0.2167,
+      "step": 562
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.938283124500134e-06,
+      "loss": 0.1404,
+      "step": 563
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.938149826712877e-06,
+      "loss": 0.1834,
+      "step": 564
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.93801652892562e-06,
+      "loss": 0.183,
+      "step": 565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.937883231138365e-06,
+      "loss": 0.1832,
+      "step": 566
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.937749933351107e-06,
+      "loss": 0.1732,
+      "step": 567
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.93761663556385e-06,
+      "loss": 0.1467,
+      "step": 568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.937483337776593e-06,
+      "loss": 0.1815,
+      "step": 569
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.937350039989338e-06,
+      "loss": 0.2087,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.93721674220208e-06,
+      "loss": 0.1751,
+      "step": 571
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.937083444414823e-06,
+      "loss": 0.1982,
+      "step": 572
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.936950146627566e-06,
+      "loss": 0.1867,
+      "step": 573
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.93681684884031e-06,
+      "loss": 0.2222,
+      "step": 574
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.936683551053053e-06,
+      "loss": 0.1978,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.936550253265796e-06,
+      "loss": 0.1563,
+      "step": 576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.936416955478539e-06,
+      "loss": 0.1833,
+      "step": 577
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.936283657691283e-06,
+      "loss": 0.1777,
+      "step": 578
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.936150359904026e-06,
+      "loss": 0.1659,
+      "step": 579
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.936017062116769e-06,
+      "loss": 0.1671,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.935883764329512e-06,
+      "loss": 0.1773,
+      "step": 581
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.935750466542256e-06,
+      "loss": 0.1646,
+      "step": 582
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.935617168755e-06,
+      "loss": 0.1771,
+      "step": 583
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.935483870967742e-06,
+      "loss": 0.2103,
+      "step": 584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.935350573180485e-06,
+      "loss": 0.1426,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.93521727539323e-06,
+      "loss": 0.2256,
+      "step": 586
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.935083977605972e-06,
+      "loss": 0.1647,
+      "step": 587
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.934950679818715e-06,
+      "loss": 0.1421,
+      "step": 588
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.934817382031458e-06,
+      "loss": 0.1872,
+      "step": 589
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.934684084244202e-06,
+      "loss": 0.2479,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.934550786456945e-06,
+      "loss": 0.1702,
+      "step": 591
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.934417488669688e-06,
+      "loss": 0.1582,
+      "step": 592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.934284190882433e-06,
+      "loss": 0.1654,
+      "step": 593
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.934150893095175e-06,
+      "loss": 0.1743,
+      "step": 594
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.934017595307918e-06,
+      "loss": 0.1397,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.933884297520661e-06,
+      "loss": 0.1599,
+      "step": 596
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.933750999733405e-06,
+      "loss": 0.2328,
+      "step": 597
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.933617701946148e-06,
+      "loss": 0.2086,
+      "step": 598
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.933484404158891e-06,
+      "loss": 0.154,
+      "step": 599
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.933351106371636e-06,
+      "loss": 0.155,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.933217808584378e-06,
+      "loss": 0.1941,
+      "step": 601
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.933084510797121e-06,
+      "loss": 0.1988,
+      "step": 602
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.932951213009864e-06,
+      "loss": 0.1939,
+      "step": 603
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.932817915222609e-06,
+      "loss": 0.2051,
+      "step": 604
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.932684617435351e-06,
+      "loss": 0.1781,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.932551319648094e-06,
+      "loss": 0.1732,
+      "step": 606
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.932418021860839e-06,
+      "loss": 0.1651,
+      "step": 607
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.932284724073582e-06,
+      "loss": 0.1776,
+      "step": 608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.932151426286324e-06,
+      "loss": 0.1871,
+      "step": 609
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.932018128499067e-06,
+      "loss": 0.1868,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.931884830711812e-06,
+      "loss": 0.2196,
+      "step": 611
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.931751532924555e-06,
+      "loss": 0.198,
+      "step": 612
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.931618235137297e-06,
+      "loss": 0.1652,
+      "step": 613
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.931484937350042e-06,
+      "loss": 0.1611,
+      "step": 614
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.931351639562785e-06,
+      "loss": 0.1824,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.931218341775528e-06,
+      "loss": 0.2027,
+      "step": 616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.93108504398827e-06,
+      "loss": 0.1958,
+      "step": 617
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.930951746201015e-06,
+      "loss": 0.237,
+      "step": 618
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.930818448413758e-06,
+      "loss": 0.1859,
+      "step": 619
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9306851506265e-06,
+      "loss": 0.1719,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.930551852839245e-06,
+      "loss": 0.2071,
+      "step": 621
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.930418555051988e-06,
+      "loss": 0.1644,
+      "step": 622
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.93028525726473e-06,
+      "loss": 0.1656,
+      "step": 623
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.930151959477473e-06,
+      "loss": 0.1571,
+      "step": 624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.930018661690218e-06,
+      "loss": 0.1605,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.92988536390296e-06,
+      "loss": 0.1395,
+      "step": 626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.929752066115704e-06,
+      "loss": 0.1533,
+      "step": 627
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.929618768328446e-06,
+      "loss": 0.1287,
+      "step": 628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.929485470541191e-06,
+      "loss": 0.2039,
+      "step": 629
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.929352172753934e-06,
+      "loss": 0.22,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.929218874966677e-06,
+      "loss": 0.1955,
+      "step": 631
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92908557717942e-06,
+      "loss": 0.1811,
+      "step": 632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.928952279392164e-06,
+      "loss": 0.2012,
+      "step": 633
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.928818981604907e-06,
+      "loss": 0.1704,
+      "step": 634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92868568381765e-06,
+      "loss": 0.1996,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.928552386030392e-06,
+      "loss": 0.2665,
+      "step": 636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.928419088243137e-06,
+      "loss": 0.2115,
+      "step": 637
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92828579045588e-06,
+      "loss": 0.1573,
+      "step": 638
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.928152492668623e-06,
+      "loss": 0.1969,
+      "step": 639
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.928019194881365e-06,
+      "loss": 0.2111,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92788589709411e-06,
+      "loss": 0.1911,
+      "step": 641
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.927752599306853e-06,
+      "loss": 0.1601,
+      "step": 642
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.927619301519595e-06,
+      "loss": 0.163,
+      "step": 643
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.927486003732338e-06,
+      "loss": 0.1691,
+      "step": 644
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.927352705945083e-06,
+      "loss": 0.1781,
+      "step": 645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.927219408157826e-06,
+      "loss": 0.166,
+      "step": 646
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.927086110370568e-06,
+      "loss": 0.1948,
+      "step": 647
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.926952812583311e-06,
+      "loss": 0.1901,
+      "step": 648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.926819514796056e-06,
+      "loss": 0.2003,
+      "step": 649
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.926686217008799e-06,
+      "loss": 0.1498,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.926552919221541e-06,
+      "loss": 0.2387,
+      "step": 651
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.926419621434284e-06,
+      "loss": 0.1618,
+      "step": 652
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.926286323647029e-06,
+      "loss": 0.2223,
+      "step": 653
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.926153025859772e-06,
+      "loss": 0.1542,
+      "step": 654
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.926019728072514e-06,
+      "loss": 0.1616,
+      "step": 655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.925886430285257e-06,
+      "loss": 0.1934,
+      "step": 656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.925753132498002e-06,
+      "loss": 0.1437,
+      "step": 657
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.925619834710745e-06,
+      "loss": 0.172,
+      "step": 658
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.925486536923487e-06,
+      "loss": 0.2211,
+      "step": 659
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92535323913623e-06,
+      "loss": 0.1705,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.925219941348975e-06,
+      "loss": 0.2009,
+      "step": 661
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.925086643561718e-06,
+      "loss": 0.2043,
+      "step": 662
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92495334577446e-06,
+      "loss": 0.1469,
+      "step": 663
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.924820047987205e-06,
+      "loss": 0.1422,
+      "step": 664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.924686750199948e-06,
+      "loss": 0.1584,
+      "step": 665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92455345241269e-06,
+      "loss": 0.1479,
+      "step": 666
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.924420154625433e-06,
+      "loss": 0.1627,
+      "step": 667
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.924286856838178e-06,
+      "loss": 0.165,
+      "step": 668
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92415355905092e-06,
+      "loss": 0.1532,
+      "step": 669
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.924020261263663e-06,
+      "loss": 0.1558,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.923886963476406e-06,
+      "loss": 0.1519,
+      "step": 671
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92375366568915e-06,
+      "loss": 0.1666,
+      "step": 672
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.923620367901894e-06,
+      "loss": 0.1732,
+      "step": 673
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.923487070114636e-06,
+      "loss": 0.1691,
+      "step": 674
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92335377232738e-06,
+      "loss": 0.1647,
+      "step": 675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.923220474540124e-06,
+      "loss": 0.1806,
+      "step": 676
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.923087176752867e-06,
+      "loss": 0.1578,
+      "step": 677
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92295387896561e-06,
+      "loss": 0.1852,
+      "step": 678
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.922820581178352e-06,
+      "loss": 0.1855,
+      "step": 679
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.922687283391097e-06,
+      "loss": 0.1756,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92255398560384e-06,
+      "loss": 0.1588,
+      "step": 681
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.922420687816582e-06,
+      "loss": 0.1507,
+      "step": 682
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.922287390029325e-06,
+      "loss": 0.2001,
+      "step": 683
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92215409224207e-06,
+      "loss": 0.1605,
+      "step": 684
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.922020794454813e-06,
+      "loss": 0.1402,
+      "step": 685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.921887496667555e-06,
+      "loss": 0.1552,
+      "step": 686
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.921754198880298e-06,
+      "loss": 0.2262,
+      "step": 687
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.921620901093043e-06,
+      "loss": 0.169,
+      "step": 688
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.921487603305785e-06,
+      "loss": 0.2193,
+      "step": 689
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.921354305518528e-06,
+      "loss": 0.2318,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.921221007731273e-06,
+      "loss": 0.2113,
+      "step": 691
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.921087709944016e-06,
+      "loss": 0.1774,
+      "step": 692
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.920954412156758e-06,
+      "loss": 0.1587,
+      "step": 693
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.920821114369503e-06,
+      "loss": 0.1783,
+      "step": 694
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.920687816582246e-06,
+      "loss": 0.1698,
+      "step": 695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.920554518794989e-06,
+      "loss": 0.158,
+      "step": 696
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.920421221007731e-06,
+      "loss": 0.1701,
+      "step": 697
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.920287923220476e-06,
+      "loss": 0.1478,
+      "step": 698
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.920154625433219e-06,
+      "loss": 0.1531,
+      "step": 699
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.920021327645962e-06,
+      "loss": 0.1941,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.919888029858706e-06,
+      "loss": 0.194,
+      "step": 701
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.919754732071449e-06,
+      "loss": 0.1539,
+      "step": 702
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.919621434284192e-06,
+      "loss": 0.175,
+      "step": 703
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.919488136496935e-06,
+      "loss": 0.2185,
+      "step": 704
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.919354838709679e-06,
+      "loss": 0.1409,
+      "step": 705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.919221540922422e-06,
+      "loss": 0.1573,
+      "step": 706
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.919088243135165e-06,
+      "loss": 0.1512,
+      "step": 707
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.91895494534791e-06,
+      "loss": 0.1903,
+      "step": 708
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.918821647560652e-06,
+      "loss": 0.1657,
+      "step": 709
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.918688349773395e-06,
+      "loss": 0.173,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.918555051986138e-06,
+      "loss": 0.1924,
+      "step": 711
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.918421754198882e-06,
+      "loss": 0.198,
+      "step": 712
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.918288456411625e-06,
+      "loss": 0.1581,
+      "step": 713
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.918155158624368e-06,
+      "loss": 0.1938,
+      "step": 714
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.91802186083711e-06,
+      "loss": 0.1719,
+      "step": 715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.917888563049855e-06,
+      "loss": 0.1976,
+      "step": 716
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.917755265262598e-06,
+      "loss": 0.1244,
+      "step": 717
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.91762196747534e-06,
+      "loss": 0.1582,
+      "step": 718
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.917488669688084e-06,
+      "loss": 0.1243,
+      "step": 719
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.917355371900828e-06,
+      "loss": 0.1607,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.917222074113571e-06,
+      "loss": 0.1691,
+      "step": 721
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.917088776326314e-06,
+      "loss": 0.1509,
+      "step": 722
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.916955478539058e-06,
+      "loss": 0.1372,
+      "step": 723
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.916822180751801e-06,
+      "loss": 0.1876,
+      "step": 724
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.916688882964544e-06,
+      "loss": 0.192,
+      "step": 725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.916555585177287e-06,
+      "loss": 0.1273,
+      "step": 726
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.916422287390031e-06,
+      "loss": 0.1551,
+      "step": 727
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.916288989602774e-06,
+      "loss": 0.195,
+      "step": 728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.916155691815517e-06,
+      "loss": 0.1152,
+      "step": 729
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.91602239402826e-06,
+      "loss": 0.1634,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.915889096241004e-06,
+      "loss": 0.1675,
+      "step": 731
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.915755798453747e-06,
+      "loss": 0.1427,
+      "step": 732
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.91562250066649e-06,
+      "loss": 0.1708,
+      "step": 733
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.915489202879233e-06,
+      "loss": 0.1793,
+      "step": 734
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.915355905091977e-06,
+      "loss": 0.1736,
+      "step": 735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.91522260730472e-06,
+      "loss": 0.168,
+      "step": 736
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.915089309517463e-06,
+      "loss": 0.1774,
+      "step": 737
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.914956011730206e-06,
+      "loss": 0.1505,
+      "step": 738
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.91482271394295e-06,
+      "loss": 0.1394,
+      "step": 739
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.914689416155693e-06,
+      "loss": 0.147,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.914556118368436e-06,
+      "loss": 0.1287,
+      "step": 741
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.914422820581179e-06,
+      "loss": 0.1769,
+      "step": 742
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.914289522793923e-06,
+      "loss": 0.2391,
+      "step": 743
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.914156225006666e-06,
+      "loss": 0.2356,
+      "step": 744
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.914022927219409e-06,
+      "loss": 0.1548,
+      "step": 745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.913889629432152e-06,
+      "loss": 0.199,
+      "step": 746
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.913756331644896e-06,
+      "loss": 0.155,
+      "step": 747
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.913623033857639e-06,
+      "loss": 0.1913,
+      "step": 748
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.913489736070382e-06,
+      "loss": 0.1378,
+      "step": 749
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.913356438283125e-06,
+      "loss": 0.1195,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.913223140495869e-06,
+      "loss": 0.1736,
+      "step": 751
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.913089842708612e-06,
+      "loss": 0.1635,
+      "step": 752
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.912956544921355e-06,
+      "loss": 0.1774,
+      "step": 753
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.912823247134098e-06,
+      "loss": 0.1656,
+      "step": 754
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.912689949346842e-06,
+      "loss": 0.1455,
+      "step": 755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.912556651559585e-06,
+      "loss": 0.1396,
+      "step": 756
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.912423353772328e-06,
+      "loss": 0.2479,
+      "step": 757
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.91229005598507e-06,
+      "loss": 0.1565,
+      "step": 758
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.912156758197815e-06,
+      "loss": 0.1911,
+      "step": 759
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.912023460410558e-06,
+      "loss": 0.2161,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9118901626233e-06,
+      "loss": 0.1451,
+      "step": 761
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.911756864836043e-06,
+      "loss": 0.2208,
+      "step": 762
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.911623567048788e-06,
+      "loss": 0.1458,
+      "step": 763
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.91149026926153e-06,
+      "loss": 0.202,
+      "step": 764
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.911356971474274e-06,
+      "loss": 0.1551,
+      "step": 765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.911223673687016e-06,
+      "loss": 0.176,
+      "step": 766
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.911090375899761e-06,
+      "loss": 0.1734,
+      "step": 767
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.910957078112504e-06,
+      "loss": 0.1802,
+      "step": 768
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.910823780325247e-06,
+      "loss": 0.1511,
+      "step": 769
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.91069048253799e-06,
+      "loss": 0.1885,
+      "step": 770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.910557184750734e-06,
+      "loss": 0.1912,
+      "step": 771
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.910423886963477e-06,
+      "loss": 0.1953,
+      "step": 772
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.91029058917622e-06,
+      "loss": 0.1476,
+      "step": 773
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.910157291388962e-06,
+      "loss": 0.1713,
+      "step": 774
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.910023993601707e-06,
+      "loss": 0.1687,
+      "step": 775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90989069581445e-06,
+      "loss": 0.1525,
+      "step": 776
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.909757398027193e-06,
+      "loss": 0.1545,
+      "step": 777
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.909624100239937e-06,
+      "loss": 0.1789,
+      "step": 778
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90949080245268e-06,
+      "loss": 0.1804,
+      "step": 779
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.909357504665423e-06,
+      "loss": 0.1265,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.909224206878165e-06,
+      "loss": 0.1676,
+      "step": 781
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90909090909091e-06,
+      "loss": 0.2089,
+      "step": 782
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.908957611303653e-06,
+      "loss": 0.196,
+      "step": 783
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.908824313516396e-06,
+      "loss": 0.1953,
+      "step": 784
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90869101572914e-06,
+      "loss": 0.2139,
+      "step": 785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.908557717941883e-06,
+      "loss": 0.1778,
+      "step": 786
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.908424420154626e-06,
+      "loss": 0.1574,
+      "step": 787
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.908291122367369e-06,
+      "loss": 0.1877,
+      "step": 788
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.908157824580113e-06,
+      "loss": 0.1747,
+      "step": 789
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.908024526792856e-06,
+      "loss": 0.1891,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.907891229005599e-06,
+      "loss": 0.1897,
+      "step": 791
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.907757931218343e-06,
+      "loss": 0.1134,
+      "step": 792
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.907624633431086e-06,
+      "loss": 0.15,
+      "step": 793
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.907491335643829e-06,
+      "loss": 0.1782,
+      "step": 794
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.907358037856572e-06,
+      "loss": 0.1685,
+      "step": 795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.907224740069316e-06,
+      "loss": 0.1636,
+      "step": 796
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.907091442282059e-06,
+      "loss": 0.1961,
+      "step": 797
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.906958144494802e-06,
+      "loss": 0.1867,
+      "step": 798
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.906824846707546e-06,
+      "loss": 0.1404,
+      "step": 799
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90669154892029e-06,
+      "loss": 0.1646,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.906558251133032e-06,
+      "loss": 0.1574,
+      "step": 801
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.906424953345775e-06,
+      "loss": 0.1505,
+      "step": 802
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90629165555852e-06,
+      "loss": 0.1298,
+      "step": 803
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.906158357771262e-06,
+      "loss": 0.1465,
+      "step": 804
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.906025059984005e-06,
+      "loss": 0.1591,
+      "step": 805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90589176219675e-06,
+      "loss": 0.1406,
+      "step": 806
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.905758464409492e-06,
+      "loss": 0.2217,
+      "step": 807
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.905625166622235e-06,
+      "loss": 0.1988,
+      "step": 808
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.905491868834978e-06,
+      "loss": 0.152,
+      "step": 809
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.905358571047722e-06,
+      "loss": 0.1957,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.905225273260465e-06,
+      "loss": 0.1404,
+      "step": 811
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.905091975473208e-06,
+      "loss": 0.137,
+      "step": 812
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.904958677685951e-06,
+      "loss": 0.1345,
+      "step": 813
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.904825379898695e-06,
+      "loss": 0.1551,
+      "step": 814
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.904692082111438e-06,
+      "loss": 0.2091,
+      "step": 815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.904558784324181e-06,
+      "loss": 0.1576,
+      "step": 816
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.904425486536924e-06,
+      "loss": 0.137,
+      "step": 817
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.904292188749668e-06,
+      "loss": 0.1705,
+      "step": 818
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.904158890962411e-06,
+      "loss": 0.1848,
+      "step": 819
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.904025593175154e-06,
+      "loss": 0.1729,
+      "step": 820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.903892295387897e-06,
+      "loss": 0.1907,
+      "step": 821
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.903758997600641e-06,
+      "loss": 0.1956,
+      "step": 822
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.903625699813384e-06,
+      "loss": 0.1829,
+      "step": 823
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.903492402026127e-06,
+      "loss": 0.1181,
+      "step": 824
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90335910423887e-06,
+      "loss": 0.1403,
+      "step": 825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.903225806451614e-06,
+      "loss": 0.165,
+      "step": 826
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.903092508664357e-06,
+      "loss": 0.1747,
+      "step": 827
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9029592108771e-06,
+      "loss": 0.1495,
+      "step": 828
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.902825913089843e-06,
+      "loss": 0.1512,
+      "step": 829
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.902692615302587e-06,
+      "loss": 0.1567,
+      "step": 830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90255931751533e-06,
+      "loss": 0.1674,
+      "step": 831
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.902426019728073e-06,
+      "loss": 0.1558,
+      "step": 832
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.902292721940817e-06,
+      "loss": 0.143,
+      "step": 833
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90215942415356e-06,
+      "loss": 0.1314,
+      "step": 834
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.902026126366303e-06,
+      "loss": 0.1732,
+      "step": 835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.901892828579046e-06,
+      "loss": 0.1208,
+      "step": 836
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90175953079179e-06,
+      "loss": 0.1235,
+      "step": 837
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.901626233004533e-06,
+      "loss": 0.1178,
+      "step": 838
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.901492935217276e-06,
+      "loss": 0.1831,
+      "step": 839
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.901359637430019e-06,
+      "loss": 0.1846,
+      "step": 840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.901226339642763e-06,
+      "loss": 0.1738,
+      "step": 841
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.901093041855506e-06,
+      "loss": 0.1703,
+      "step": 842
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.900959744068249e-06,
+      "loss": 0.1429,
+      "step": 843
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.900826446280992e-06,
+      "loss": 0.1382,
+      "step": 844
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.900693148493736e-06,
+      "loss": 0.154,
+      "step": 845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90055985070648e-06,
+      "loss": 0.1789,
+      "step": 846
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.900426552919222e-06,
+      "loss": 0.1653,
+      "step": 847
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.900293255131965e-06,
+      "loss": 0.1556,
+      "step": 848
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90015995734471e-06,
+      "loss": 0.1255,
+      "step": 849
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.900026659557452e-06,
+      "loss": 0.1751,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.899893361770195e-06,
+      "loss": 0.1888,
+      "step": 851
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.899760063982938e-06,
+      "loss": 0.1519,
+      "step": 852
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.899626766195682e-06,
+      "loss": 0.1854,
+      "step": 853
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.899493468408425e-06,
+      "loss": 0.1454,
+      "step": 854
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.899360170621168e-06,
+      "loss": 0.1543,
+      "step": 855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.89922687283391e-06,
+      "loss": 0.1302,
+      "step": 856
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.899093575046655e-06,
+      "loss": 0.1612,
+      "step": 857
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.898960277259398e-06,
+      "loss": 0.1413,
+      "step": 858
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.898826979472141e-06,
+      "loss": 0.1887,
+      "step": 859
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.898693681684884e-06,
+      "loss": 0.1204,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.898560383897628e-06,
+      "loss": 0.2046,
+      "step": 861
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.898427086110371e-06,
+      "loss": 0.1633,
+      "step": 862
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.898293788323114e-06,
+      "loss": 0.1758,
+      "step": 863
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.898160490535857e-06,
+      "loss": 0.1626,
+      "step": 864
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.898027192748601e-06,
+      "loss": 0.1279,
+      "step": 865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.897893894961344e-06,
+      "loss": 0.1581,
+      "step": 866
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.897760597174087e-06,
+      "loss": 0.177,
+      "step": 867
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.89762729938683e-06,
+      "loss": 0.157,
+      "step": 868
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.897494001599574e-06,
+      "loss": 0.1084,
+      "step": 869
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.897360703812317e-06,
+      "loss": 0.1831,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.89722740602506e-06,
+      "loss": 0.1663,
+      "step": 871
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.897094108237804e-06,
+      "loss": 0.176,
+      "step": 872
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.896960810450547e-06,
+      "loss": 0.1734,
+      "step": 873
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.89682751266329e-06,
+      "loss": 0.1528,
+      "step": 874
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.896694214876033e-06,
+      "loss": 0.1932,
+      "step": 875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.896560917088777e-06,
+      "loss": 0.1835,
+      "step": 876
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.89642761930152e-06,
+      "loss": 0.1523,
+      "step": 877
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.896294321514263e-06,
+      "loss": 0.1522,
+      "step": 878
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.896161023727007e-06,
+      "loss": 0.1885,
+      "step": 879
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.89602772593975e-06,
+      "loss": 0.1921,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.895894428152493e-06,
+      "loss": 0.2162,
+      "step": 881
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.895761130365236e-06,
+      "loss": 0.1617,
+      "step": 882
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.89562783257798e-06,
+      "loss": 0.1266,
+      "step": 883
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.895494534790723e-06,
+      "loss": 0.1402,
+      "step": 884
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.895361237003466e-06,
+      "loss": 0.1738,
+      "step": 885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.89522793921621e-06,
+      "loss": 0.1417,
+      "step": 886
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.895094641428953e-06,
+      "loss": 0.2088,
+      "step": 887
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.894961343641696e-06,
+      "loss": 0.166,
+      "step": 888
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.894828045854439e-06,
+      "loss": 0.165,
+      "step": 889
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.894694748067184e-06,
+      "loss": 0.1643,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.894561450279926e-06,
+      "loss": 0.1648,
+      "step": 891
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.89442815249267e-06,
+      "loss": 0.154,
+      "step": 892
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.894294854705414e-06,
+      "loss": 0.1594,
+      "step": 893
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.894161556918157e-06,
+      "loss": 0.1798,
+      "step": 894
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.8940282591309e-06,
+      "loss": 0.1141,
+      "step": 895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.893894961343642e-06,
+      "loss": 0.1377,
+      "step": 896
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.893761663556387e-06,
+      "loss": 0.1692,
+      "step": 897
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.89362836576913e-06,
+      "loss": 0.1552,
+      "step": 898
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.893495067981872e-06,
+      "loss": 0.167,
+      "step": 899
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.893361770194617e-06,
+      "loss": 0.146,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.89322847240736e-06,
+      "loss": 0.1355,
+      "step": 901
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.893095174620103e-06,
+      "loss": 0.1307,
+      "step": 902
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.892961876832845e-06,
+      "loss": 0.1999,
+      "step": 903
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.89282857904559e-06,
+      "loss": 0.1594,
+      "step": 904
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.892695281258333e-06,
+      "loss": 0.1921,
+      "step": 905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.892561983471075e-06,
+      "loss": 0.1769,
+      "step": 906
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.892428685683818e-06,
+      "loss": 0.1411,
+      "step": 907
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.892295387896563e-06,
+      "loss": 0.1889,
+      "step": 908
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.892162090109306e-06,
+      "loss": 0.1375,
+      "step": 909
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.892028792322048e-06,
+      "loss": 0.1348,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.891895494534791e-06,
+      "loss": 0.1877,
+      "step": 911
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.891762196747536e-06,
+      "loss": 0.1247,
+      "step": 912
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.891628898960279e-06,
+      "loss": 0.126,
+      "step": 913
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.891495601173021e-06,
+      "loss": 0.1373,
+      "step": 914
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.891362303385764e-06,
+      "loss": 0.1543,
+      "step": 915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.891229005598509e-06,
+      "loss": 0.1688,
+      "step": 916
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.891095707811252e-06,
+      "loss": 0.1385,
+      "step": 917
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.890962410023994e-06,
+      "loss": 0.1391,
+      "step": 918
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.890829112236737e-06,
+      "loss": 0.1521,
+      "step": 919
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.890695814449482e-06,
+      "loss": 0.1645,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.890562516662225e-06,
+      "loss": 0.1725,
+      "step": 921
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.890429218874967e-06,
+      "loss": 0.1651,
+      "step": 922
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.89029592108771e-06,
+      "loss": 0.1621,
+      "step": 923
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.890162623300455e-06,
+      "loss": 0.1441,
+      "step": 924
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.890029325513198e-06,
+      "loss": 0.1976,
+      "step": 925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88989602772594e-06,
+      "loss": 0.1737,
+      "step": 926
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.889762729938683e-06,
+      "loss": 0.139,
+      "step": 927
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.889629432151428e-06,
+      "loss": 0.1533,
+      "step": 928
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88949613436417e-06,
+      "loss": 0.1576,
+      "step": 929
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.889362836576913e-06,
+      "loss": 0.1375,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.889229538789656e-06,
+      "loss": 0.2009,
+      "step": 931
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.8890962410024e-06,
+      "loss": 0.1409,
+      "step": 932
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.888962943215143e-06,
+      "loss": 0.1687,
+      "step": 933
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.888829645427886e-06,
+      "loss": 0.2027,
+      "step": 934
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.888696347640629e-06,
+      "loss": 0.1772,
+      "step": 935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.888563049853374e-06,
+      "loss": 0.1393,
+      "step": 936
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.888429752066116e-06,
+      "loss": 0.1818,
+      "step": 937
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88829645427886e-06,
+      "loss": 0.1454,
+      "step": 938
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.888163156491602e-06,
+      "loss": 0.1536,
+      "step": 939
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.888029858704347e-06,
+      "loss": 0.139,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88789656091709e-06,
+      "loss": 0.1459,
+      "step": 941
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.887763263129832e-06,
+      "loss": 0.1655,
+      "step": 942
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.887629965342575e-06,
+      "loss": 0.1524,
+      "step": 943
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88749666755532e-06,
+      "loss": 0.1415,
+      "step": 944
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.887363369768062e-06,
+      "loss": 0.1269,
+      "step": 945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.887230071980805e-06,
+      "loss": 0.1898,
+      "step": 946
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88709677419355e-06,
+      "loss": 0.1836,
+      "step": 947
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.886963476406293e-06,
+      "loss": 0.137,
+      "step": 948
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.886830178619035e-06,
+      "loss": 0.1741,
+      "step": 949
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.886696880831778e-06,
+      "loss": 0.1577,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.886563583044523e-06,
+      "loss": 0.1221,
+      "step": 951
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.886430285257265e-06,
+      "loss": 0.1395,
+      "step": 952
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.886296987470008e-06,
+      "loss": 0.1467,
+      "step": 953
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.886163689682751e-06,
+      "loss": 0.106,
+      "step": 954
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.886030391895496e-06,
+      "loss": 0.1216,
+      "step": 955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.885897094108238e-06,
+      "loss": 0.1313,
+      "step": 956
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.885763796320981e-06,
+      "loss": 0.1573,
+      "step": 957
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.885630498533724e-06,
+      "loss": 0.1614,
+      "step": 958
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.885497200746469e-06,
+      "loss": 0.1203,
+      "step": 959
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.885363902959211e-06,
+      "loss": 0.1798,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.885230605171954e-06,
+      "loss": 0.1304,
+      "step": 961
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.885097307384697e-06,
+      "loss": 0.1326,
+      "step": 962
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.884964009597442e-06,
+      "loss": 0.1463,
+      "step": 963
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.884830711810184e-06,
+      "loss": 0.2028,
+      "step": 964
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.884697414022927e-06,
+      "loss": 0.1749,
+      "step": 965
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.884564116235672e-06,
+      "loss": 0.1686,
+      "step": 966
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.884430818448415e-06,
+      "loss": 0.1643,
+      "step": 967
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.884297520661157e-06,
+      "loss": 0.1534,
+      "step": 968
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.8841642228739e-06,
+      "loss": 0.165,
+      "step": 969
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.884030925086645e-06,
+      "loss": 0.1495,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.883897627299388e-06,
+      "loss": 0.1731,
+      "step": 971
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88376432951213e-06,
+      "loss": 0.146,
+      "step": 972
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.883631031724875e-06,
+      "loss": 0.2028,
+      "step": 973
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.883497733937618e-06,
+      "loss": 0.1138,
+      "step": 974
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88336443615036e-06,
+      "loss": 0.1541,
+      "step": 975
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.883231138363103e-06,
+      "loss": 0.14,
+      "step": 976
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.883097840575848e-06,
+      "loss": 0.149,
+      "step": 977
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88296454278859e-06,
+      "loss": 0.1504,
+      "step": 978
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.882831245001333e-06,
+      "loss": 0.1369,
+      "step": 979
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.882697947214078e-06,
+      "loss": 0.1189,
+      "step": 980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88256464942682e-06,
+      "loss": 0.156,
+      "step": 981
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.882431351639564e-06,
+      "loss": 0.1359,
+      "step": 982
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.882298053852306e-06,
+      "loss": 0.1437,
+      "step": 983
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.882164756065051e-06,
+      "loss": 0.1457,
+      "step": 984
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.882031458277794e-06,
+      "loss": 0.1843,
+      "step": 985
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.881898160490537e-06,
+      "loss": 0.1888,
+      "step": 986
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.881764862703281e-06,
+      "loss": 0.1687,
+      "step": 987
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.881631564916024e-06,
+      "loss": 0.1469,
+      "step": 988
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.881498267128767e-06,
+      "loss": 0.1636,
+      "step": 989
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88136496934151e-06,
+      "loss": 0.1467,
+      "step": 990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.881231671554254e-06,
+      "loss": 0.1439,
+      "step": 991
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.881098373766997e-06,
+      "loss": 0.1905,
+      "step": 992
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88096507597974e-06,
+      "loss": 0.1454,
+      "step": 993
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.880831778192483e-06,
+      "loss": 0.1379,
+      "step": 994
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.880698480405227e-06,
+      "loss": 0.1412,
+      "step": 995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88056518261797e-06,
+      "loss": 0.121,
+      "step": 996
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.880431884830713e-06,
+      "loss": 0.1289,
+      "step": 997
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.880298587043455e-06,
+      "loss": 0.1393,
+      "step": 998
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.8801652892562e-06,
+      "loss": 0.2252,
+      "step": 999
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.880031991468943e-06,
+      "loss": 0.1656,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.879898693681686e-06,
+      "loss": 0.1667,
+      "step": 1001
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.87976539589443e-06,
+      "loss": 0.1027,
+      "step": 1002
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.879632098107173e-06,
+      "loss": 0.1558,
+      "step": 1003
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.879498800319916e-06,
+      "loss": 0.1078,
+      "step": 1004
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.879365502532659e-06,
+      "loss": 0.1656,
+      "step": 1005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.879232204745403e-06,
+      "loss": 0.1897,
+      "step": 1006
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.879098906958146e-06,
+      "loss": 0.1746,
+      "step": 1007
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.878965609170889e-06,
+      "loss": 0.122,
+      "step": 1008
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.878832311383632e-06,
+      "loss": 0.1619,
+      "step": 1009
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.878699013596376e-06,
+      "loss": 0.1567,
+      "step": 1010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.878565715809119e-06,
+      "loss": 0.17,
+      "step": 1011
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.878432418021862e-06,
+      "loss": 0.1647,
+      "step": 1012
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.878299120234605e-06,
+      "loss": 0.1527,
+      "step": 1013
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.878165822447349e-06,
+      "loss": 0.1189,
+      "step": 1014
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.878032524660092e-06,
+      "loss": 0.1575,
+      "step": 1015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.877899226872835e-06,
+      "loss": 0.1438,
+      "step": 1016
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.877765929085578e-06,
+      "loss": 0.1432,
+      "step": 1017
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.877632631298322e-06,
+      "loss": 0.1305,
+      "step": 1018
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.877499333511065e-06,
+      "loss": 0.146,
+      "step": 1019
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.877366035723808e-06,
+      "loss": 0.1498,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.87723273793655e-06,
+      "loss": 0.2041,
+      "step": 1021
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.877099440149295e-06,
+      "loss": 0.1359,
+      "step": 1022
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.876966142362038e-06,
+      "loss": 0.1605,
+      "step": 1023
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.87683284457478e-06,
+      "loss": 0.1448,
+      "step": 1024
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.876699546787523e-06,
+      "loss": 0.1058,
+      "step": 1025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.876566249000268e-06,
+      "loss": 0.1357,
+      "step": 1026
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.87643295121301e-06,
+      "loss": 0.122,
+      "step": 1027
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.876299653425754e-06,
+      "loss": 0.1558,
+      "step": 1028
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.876166355638496e-06,
+      "loss": 0.1536,
+      "step": 1029
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.876033057851241e-06,
+      "loss": 0.1356,
+      "step": 1030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.875899760063984e-06,
+      "loss": 0.1609,
+      "step": 1031
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.875766462276727e-06,
+      "loss": 0.1555,
+      "step": 1032
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.87563316448947e-06,
+      "loss": 0.1918,
+      "step": 1033
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.875499866702214e-06,
+      "loss": 0.1461,
+      "step": 1034
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.875366568914957e-06,
+      "loss": 0.1577,
+      "step": 1035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.8752332711277e-06,
+      "loss": 0.1437,
+      "step": 1036
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.875099973340442e-06,
+      "loss": 0.1584,
+      "step": 1037
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.874966675553187e-06,
+      "loss": 0.1548,
+      "step": 1038
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.87483337776593e-06,
+      "loss": 0.1649,
+      "step": 1039
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.874700079978673e-06,
+      "loss": 0.1379,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.874566782191415e-06,
+      "loss": 0.1658,
+      "step": 1041
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.87443348440416e-06,
+      "loss": 0.1699,
+      "step": 1042
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.874300186616903e-06,
+      "loss": 0.1507,
+      "step": 1043
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.874166888829645e-06,
+      "loss": 0.1701,
+      "step": 1044
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.874033591042388e-06,
+      "loss": 0.1712,
+      "step": 1045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.873900293255133e-06,
+      "loss": 0.1389,
+      "step": 1046
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.873766995467876e-06,
+      "loss": 0.1768,
+      "step": 1047
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.873633697680618e-06,
+      "loss": 0.1199,
+      "step": 1048
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.873500399893361e-06,
+      "loss": 0.207,
+      "step": 1049
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.873367102106106e-06,
+      "loss": 0.1657,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.873233804318849e-06,
+      "loss": 0.1794,
+      "step": 1051
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.873100506531591e-06,
+      "loss": 0.2125,
+      "step": 1052
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.872967208744336e-06,
+      "loss": 0.1473,
+      "step": 1053
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.872833910957079e-06,
+      "loss": 0.1697,
+      "step": 1054
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.872700613169822e-06,
+      "loss": 0.1718,
+      "step": 1055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.872567315382564e-06,
+      "loss": 0.1392,
+      "step": 1056
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.872434017595309e-06,
+      "loss": 0.1626,
+      "step": 1057
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.872300719808052e-06,
+      "loss": 0.1343,
+      "step": 1058
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.872167422020795e-06,
+      "loss": 0.1896,
+      "step": 1059
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.872034124233539e-06,
+      "loss": 0.1608,
+      "step": 1060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.871900826446282e-06,
+      "loss": 0.1361,
+      "step": 1061
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.871767528659025e-06,
+      "loss": 0.1741,
+      "step": 1062
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.871634230871768e-06,
+      "loss": 0.157,
+      "step": 1063
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.871500933084512e-06,
+      "loss": 0.1663,
+      "step": 1064
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.871367635297255e-06,
+      "loss": 0.1395,
+      "step": 1065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.871234337509998e-06,
+      "loss": 0.1365,
+      "step": 1066
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.871101039722742e-06,
+      "loss": 0.1467,
+      "step": 1067
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.870967741935485e-06,
+      "loss": 0.1983,
+      "step": 1068
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.870834444148228e-06,
+      "loss": 0.1699,
+      "step": 1069
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.87070114636097e-06,
+      "loss": 0.1379,
+      "step": 1070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.870567848573715e-06,
+      "loss": 0.1887,
+      "step": 1071
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.870434550786458e-06,
+      "loss": 0.1458,
+      "step": 1072
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.8703012529992e-06,
+      "loss": 0.1449,
+      "step": 1073
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.870167955211945e-06,
+      "loss": 0.1263,
+      "step": 1074
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.870034657424688e-06,
+      "loss": 0.1219,
+      "step": 1075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.869901359637431e-06,
+      "loss": 0.1675,
+      "step": 1076
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.869768061850174e-06,
+      "loss": 0.1802,
+      "step": 1077
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.869634764062918e-06,
+      "loss": 0.1632,
+      "step": 1078
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.869501466275661e-06,
+      "loss": 0.1478,
+      "step": 1079
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.869368168488404e-06,
+      "loss": 0.2206,
+      "step": 1080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.869234870701148e-06,
+      "loss": 0.1582,
+      "step": 1081
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.869101572913891e-06,
+      "loss": 0.1417,
+      "step": 1082
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.868968275126634e-06,
+      "loss": 0.1686,
+      "step": 1083
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.868834977339377e-06,
+      "loss": 0.1417,
+      "step": 1084
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.868701679552121e-06,
+      "loss": 0.1921,
+      "step": 1085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.868568381764864e-06,
+      "loss": 0.1885,
+      "step": 1086
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.868435083977607e-06,
+      "loss": 0.1454,
+      "step": 1087
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.86830178619035e-06,
+      "loss": 0.1564,
+      "step": 1088
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.868168488403094e-06,
+      "loss": 0.1156,
+      "step": 1089
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.868035190615837e-06,
+      "loss": 0.1387,
+      "step": 1090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.86790189282858e-06,
+      "loss": 0.1473,
+      "step": 1091
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.867768595041323e-06,
+      "loss": 0.1313,
+      "step": 1092
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.867635297254067e-06,
+      "loss": 0.1569,
+      "step": 1093
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.86750199946681e-06,
+      "loss": 0.1378,
+      "step": 1094
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.867368701679553e-06,
+      "loss": 0.1263,
+      "step": 1095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.867235403892296e-06,
+      "loss": 0.1853,
+      "step": 1096
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.86710210610504e-06,
+      "loss": 0.1157,
+      "step": 1097
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.866968808317783e-06,
+      "loss": 0.1615,
+      "step": 1098
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.866835510530526e-06,
+      "loss": 0.1504,
+      "step": 1099
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.866702212743269e-06,
+      "loss": 0.2074,
+      "step": 1100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.866568914956013e-06,
+      "loss": 0.1364,
+      "step": 1101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.866435617168756e-06,
+      "loss": 0.1417,
+      "step": 1102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.866302319381499e-06,
+      "loss": 0.1986,
+      "step": 1103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.866169021594242e-06,
+      "loss": 0.1466,
+      "step": 1104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.866035723806986e-06,
+      "loss": 0.0855,
+      "step": 1105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.865902426019729e-06,
+      "loss": 0.1666,
+      "step": 1106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.865769128232472e-06,
+      "loss": 0.1177,
+      "step": 1107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.865635830445215e-06,
+      "loss": 0.1829,
+      "step": 1108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.86550253265796e-06,
+      "loss": 0.1063,
+      "step": 1109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.865369234870702e-06,
+      "loss": 0.1191,
+      "step": 1110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.865235937083445e-06,
+      "loss": 0.1326,
+      "step": 1111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.865102639296188e-06,
+      "loss": 0.1228,
+      "step": 1112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.864969341508932e-06,
+      "loss": 0.1285,
+      "step": 1113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.864836043721675e-06,
+      "loss": 0.1195,
+      "step": 1114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.864702745934418e-06,
+      "loss": 0.222,
+      "step": 1115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.864569448147162e-06,
+      "loss": 0.1306,
+      "step": 1116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.864436150359905e-06,
+      "loss": 0.1237,
+      "step": 1117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.864302852572648e-06,
+      "loss": 0.1729,
+      "step": 1118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.86416955478539e-06,
+      "loss": 0.1586,
+      "step": 1119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.864036256998135e-06,
+      "loss": 0.1675,
+      "step": 1120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.863902959210878e-06,
+      "loss": 0.1437,
+      "step": 1121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.863769661423621e-06,
+      "loss": 0.143,
+      "step": 1122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.863636363636364e-06,
+      "loss": 0.1511,
+      "step": 1123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.863503065849108e-06,
+      "loss": 0.1275,
+      "step": 1124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.863369768061851e-06,
+      "loss": 0.1726,
+      "step": 1125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.863236470274594e-06,
+      "loss": 0.1357,
+      "step": 1126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.863103172487337e-06,
+      "loss": 0.1299,
+      "step": 1127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.862969874700081e-06,
+      "loss": 0.1288,
+      "step": 1128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.862836576912824e-06,
+      "loss": 0.1784,
+      "step": 1129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.862703279125567e-06,
+      "loss": 0.1633,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.86256998133831e-06,
+      "loss": 0.1857,
+      "step": 1131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.862436683551054e-06,
+      "loss": 0.1715,
+      "step": 1132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.862303385763797e-06,
+      "loss": 0.1449,
+      "step": 1133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.86217008797654e-06,
+      "loss": 0.1308,
+      "step": 1134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.862036790189283e-06,
+      "loss": 0.1578,
+      "step": 1135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.861903492402027e-06,
+      "loss": 0.1262,
+      "step": 1136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.86177019461477e-06,
+      "loss": 0.1631,
+      "step": 1137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.861636896827513e-06,
+      "loss": 0.1701,
+      "step": 1138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.861503599040256e-06,
+      "loss": 0.1764,
+      "step": 1139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.861370301253e-06,
+      "loss": 0.1564,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.861237003465743e-06,
+      "loss": 0.1483,
+      "step": 1141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.861103705678486e-06,
+      "loss": 0.1488,
+      "step": 1142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.860970407891229e-06,
+      "loss": 0.1641,
+      "step": 1143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.860837110103973e-06,
+      "loss": 0.2019,
+      "step": 1144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.860703812316716e-06,
+      "loss": 0.2056,
+      "step": 1145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.860570514529459e-06,
+      "loss": 0.1638,
+      "step": 1146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.860437216742203e-06,
+      "loss": 0.1602,
+      "step": 1147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.860303918954946e-06,
+      "loss": 0.1615,
+      "step": 1148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.860170621167689e-06,
+      "loss": 0.1504,
+      "step": 1149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.860037323380432e-06,
+      "loss": 0.1617,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.859904025593176e-06,
+      "loss": 0.205,
+      "step": 1151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.859770727805919e-06,
+      "loss": 0.1519,
+      "step": 1152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.859637430018662e-06,
+      "loss": 0.154,
+      "step": 1153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.859504132231406e-06,
+      "loss": 0.1129,
+      "step": 1154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.85937083444415e-06,
+      "loss": 0.1418,
+      "step": 1155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.859237536656892e-06,
+      "loss": 0.1684,
+      "step": 1156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.859104238869635e-06,
+      "loss": 0.1165,
+      "step": 1157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.85897094108238e-06,
+      "loss": 0.1074,
+      "step": 1158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.858837643295122e-06,
+      "loss": 0.1393,
+      "step": 1159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.858704345507865e-06,
+      "loss": 0.1383,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.85857104772061e-06,
+      "loss": 0.1408,
+      "step": 1161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.858437749933352e-06,
+      "loss": 0.1456,
+      "step": 1162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.858304452146095e-06,
+      "loss": 0.1634,
+      "step": 1163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.858171154358838e-06,
+      "loss": 0.1793,
+      "step": 1164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.858037856571582e-06,
+      "loss": 0.1347,
+      "step": 1165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.857904558784325e-06,
+      "loss": 0.1362,
+      "step": 1166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.857771260997068e-06,
+      "loss": 0.1466,
+      "step": 1167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.857637963209813e-06,
+      "loss": 0.1216,
+      "step": 1168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.857504665422555e-06,
+      "loss": 0.112,
+      "step": 1169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.857371367635298e-06,
+      "loss": 0.2115,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.857238069848041e-06,
+      "loss": 0.1558,
+      "step": 1171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.857104772060786e-06,
+      "loss": 0.144,
+      "step": 1172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.856971474273528e-06,
+      "loss": 0.1741,
+      "step": 1173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.856838176486271e-06,
+      "loss": 0.146,
+      "step": 1174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.856704878699016e-06,
+      "loss": 0.1828,
+      "step": 1175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.856571580911759e-06,
+      "loss": 0.1692,
+      "step": 1176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.856438283124501e-06,
+      "loss": 0.1688,
+      "step": 1177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.856304985337244e-06,
+      "loss": 0.1398,
+      "step": 1178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.856171687549989e-06,
+      "loss": 0.1627,
+      "step": 1179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.856038389762732e-06,
+      "loss": 0.1726,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.855905091975474e-06,
+      "loss": 0.1704,
+      "step": 1181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.855771794188217e-06,
+      "loss": 0.1605,
+      "step": 1182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.855638496400962e-06,
+      "loss": 0.1453,
+      "step": 1183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.855505198613705e-06,
+      "loss": 0.2029,
+      "step": 1184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.855371900826447e-06,
+      "loss": 0.1377,
+      "step": 1185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.85523860303919e-06,
+      "loss": 0.1554,
+      "step": 1186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.855105305251935e-06,
+      "loss": 0.1576,
+      "step": 1187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.854972007464677e-06,
+      "loss": 0.1807,
+      "step": 1188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.85483870967742e-06,
+      "loss": 0.1733,
+      "step": 1189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.854705411890163e-06,
+      "loss": 0.2,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.854572114102908e-06,
+      "loss": 0.1706,
+      "step": 1191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.85443881631565e-06,
+      "loss": 0.1726,
+      "step": 1192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.854305518528393e-06,
+      "loss": 0.1466,
+      "step": 1193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.854172220741136e-06,
+      "loss": 0.1799,
+      "step": 1194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.85403892295388e-06,
+      "loss": 0.1938,
+      "step": 1195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.853905625166623e-06,
+      "loss": 0.1515,
+      "step": 1196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.853772327379366e-06,
+      "loss": 0.1618,
+      "step": 1197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.853639029592109e-06,
+      "loss": 0.1426,
+      "step": 1198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.853505731804854e-06,
+      "loss": 0.1661,
+      "step": 1199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.853372434017596e-06,
+      "loss": 0.1523,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.85323913623034e-06,
+      "loss": 0.1323,
+      "step": 1201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.853105838443082e-06,
+      "loss": 0.1203,
+      "step": 1202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.852972540655827e-06,
+      "loss": 0.1625,
+      "step": 1203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.85283924286857e-06,
+      "loss": 0.123,
+      "step": 1204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.852705945081312e-06,
+      "loss": 0.1426,
+      "step": 1205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.852572647294055e-06,
+      "loss": 0.1143,
+      "step": 1206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.8524393495068e-06,
+      "loss": 0.13,
+      "step": 1207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.852306051719542e-06,
+      "loss": 0.1552,
+      "step": 1208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.852172753932285e-06,
+      "loss": 0.1566,
+      "step": 1209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.852039456145028e-06,
+      "loss": 0.0947,
+      "step": 1210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.851906158357772e-06,
+      "loss": 0.1307,
+      "step": 1211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.851772860570515e-06,
+      "loss": 0.1406,
+      "step": 1212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.851639562783258e-06,
+      "loss": 0.1437,
+      "step": 1213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.851506264996001e-06,
+      "loss": 0.1866,
+      "step": 1214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.851372967208745e-06,
+      "loss": 0.1453,
+      "step": 1215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.851239669421488e-06,
+      "loss": 0.1545,
+      "step": 1216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.851106371634231e-06,
+      "loss": 0.1434,
+      "step": 1217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.850973073846974e-06,
+      "loss": 0.176,
+      "step": 1218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.850839776059718e-06,
+      "loss": 0.1788,
+      "step": 1219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.850706478272461e-06,
+      "loss": 0.1622,
+      "step": 1220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.850573180485204e-06,
+      "loss": 0.1235,
+      "step": 1221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.850439882697947e-06,
+      "loss": 0.1851,
+      "step": 1222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.850306584910691e-06,
+      "loss": 0.1311,
+      "step": 1223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.850173287123434e-06,
+      "loss": 0.1262,
+      "step": 1224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.850039989336177e-06,
+      "loss": 0.1642,
+      "step": 1225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84990669154892e-06,
+      "loss": 0.148,
+      "step": 1226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.849773393761664e-06,
+      "loss": 0.124,
+      "step": 1227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.849640095974407e-06,
+      "loss": 0.1405,
+      "step": 1228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84950679818715e-06,
+      "loss": 0.1409,
+      "step": 1229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.849373500399895e-06,
+      "loss": 0.1779,
+      "step": 1230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.849240202612637e-06,
+      "loss": 0.1166,
+      "step": 1231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84910690482538e-06,
+      "loss": 0.1556,
+      "step": 1232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.848973607038123e-06,
+      "loss": 0.1369,
+      "step": 1233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.848840309250867e-06,
+      "loss": 0.1366,
+      "step": 1234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84870701146361e-06,
+      "loss": 0.1624,
+      "step": 1235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.848573713676353e-06,
+      "loss": 0.1705,
+      "step": 1236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.848440415889096e-06,
+      "loss": 0.1357,
+      "step": 1237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84830711810184e-06,
+      "loss": 0.1097,
+      "step": 1238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.848173820314583e-06,
+      "loss": 0.149,
+      "step": 1239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.848040522527326e-06,
+      "loss": 0.1186,
+      "step": 1240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84790722474007e-06,
+      "loss": 0.1092,
+      "step": 1241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.847773926952813e-06,
+      "loss": 0.1508,
+      "step": 1242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.847640629165556e-06,
+      "loss": 0.168,
+      "step": 1243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.847507331378299e-06,
+      "loss": 0.2179,
+      "step": 1244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.847374033591044e-06,
+      "loss": 0.1226,
+      "step": 1245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.847240735803786e-06,
+      "loss": 0.1152,
+      "step": 1246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84710743801653e-06,
+      "loss": 0.1134,
+      "step": 1247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.846974140229274e-06,
+      "loss": 0.1271,
+      "step": 1248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.846840842442017e-06,
+      "loss": 0.1748,
+      "step": 1249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84670754465476e-06,
+      "loss": 0.171,
+      "step": 1250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.846574246867502e-06,
+      "loss": 0.1708,
+      "step": 1251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.846440949080247e-06,
+      "loss": 0.1494,
+      "step": 1252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84630765129299e-06,
+      "loss": 0.1298,
+      "step": 1253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.846174353505732e-06,
+      "loss": 0.1634,
+      "step": 1254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.846041055718477e-06,
+      "loss": 0.1409,
+      "step": 1255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84590775793122e-06,
+      "loss": 0.1239,
+      "step": 1256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.845774460143962e-06,
+      "loss": 0.1435,
+      "step": 1257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.845641162356705e-06,
+      "loss": 0.1456,
+      "step": 1258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84550786456945e-06,
+      "loss": 0.1657,
+      "step": 1259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.845374566782193e-06,
+      "loss": 0.149,
+      "step": 1260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.845241268994935e-06,
+      "loss": 0.1228,
+      "step": 1261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84510797120768e-06,
+      "loss": 0.1432,
+      "step": 1262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.844974673420423e-06,
+      "loss": 0.1437,
+      "step": 1263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.844841375633166e-06,
+      "loss": 0.1698,
+      "step": 1264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.844708077845908e-06,
+      "loss": 0.1397,
+      "step": 1265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.844574780058653e-06,
+      "loss": 0.1638,
+      "step": 1266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.844441482271396e-06,
+      "loss": 0.1684,
+      "step": 1267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.844308184484139e-06,
+      "loss": 0.1382,
+      "step": 1268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.844174886696881e-06,
+      "loss": 0.1315,
+      "step": 1269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.844041588909626e-06,
+      "loss": 0.1407,
+      "step": 1270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.843908291122369e-06,
+      "loss": 0.2015,
+      "step": 1271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.843774993335112e-06,
+      "loss": 0.2166,
+      "step": 1272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.843641695547854e-06,
+      "loss": 0.1576,
+      "step": 1273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.843508397760599e-06,
+      "loss": 0.1552,
+      "step": 1274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.843375099973342e-06,
+      "loss": 0.0959,
+      "step": 1275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.843241802186085e-06,
+      "loss": 0.1151,
+      "step": 1276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.843108504398827e-06,
+      "loss": 0.155,
+      "step": 1277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.842975206611572e-06,
+      "loss": 0.1543,
+      "step": 1278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.842841908824315e-06,
+      "loss": 0.1537,
+      "step": 1279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.842708611037057e-06,
+      "loss": 0.1683,
+      "step": 1280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.8425753132498e-06,
+      "loss": 0.1513,
+      "step": 1281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.842442015462545e-06,
+      "loss": 0.1458,
+      "step": 1282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.842308717675288e-06,
+      "loss": 0.1487,
+      "step": 1283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84217541988803e-06,
+      "loss": 0.1414,
+      "step": 1284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.842042122100775e-06,
+      "loss": 0.1622,
+      "step": 1285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.841908824313518e-06,
+      "loss": 0.1318,
+      "step": 1286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84177552652626e-06,
+      "loss": 0.0979,
+      "step": 1287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.841642228739003e-06,
+      "loss": 0.1429,
+      "step": 1288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.841508930951748e-06,
+      "loss": 0.1356,
+      "step": 1289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84137563316449e-06,
+      "loss": 0.1025,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.841242335377234e-06,
+      "loss": 0.1603,
+      "step": 1291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.841109037589976e-06,
+      "loss": 0.1569,
+      "step": 1292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.840975739802721e-06,
+      "loss": 0.1443,
+      "step": 1293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.840842442015464e-06,
+      "loss": 0.1304,
+      "step": 1294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.840709144228207e-06,
+      "loss": 0.1415,
+      "step": 1295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84057584644095e-06,
+      "loss": 0.1522,
+      "step": 1296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.840442548653694e-06,
+      "loss": 0.1168,
+      "step": 1297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.840309250866437e-06,
+      "loss": 0.1443,
+      "step": 1298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84017595307918e-06,
+      "loss": 0.178,
+      "step": 1299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.840042655291922e-06,
+      "loss": 0.2247,
+      "step": 1300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.839909357504667e-06,
+      "loss": 0.1626,
+      "step": 1301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83977605971741e-06,
+      "loss": 0.1468,
+      "step": 1302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.839642761930152e-06,
+      "loss": 0.0967,
+      "step": 1303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.839509464142895e-06,
+      "loss": 0.1152,
+      "step": 1304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83937616635564e-06,
+      "loss": 0.1769,
+      "step": 1305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.839242868568383e-06,
+      "loss": 0.1338,
+      "step": 1306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.839109570781125e-06,
+      "loss": 0.1554,
+      "step": 1307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.838976272993868e-06,
+      "loss": 0.1667,
+      "step": 1308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.838842975206613e-06,
+      "loss": 0.122,
+      "step": 1309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.838709677419356e-06,
+      "loss": 0.16,
+      "step": 1310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.838576379632098e-06,
+      "loss": 0.1302,
+      "step": 1311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.838443081844841e-06,
+      "loss": 0.1379,
+      "step": 1312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.838309784057586e-06,
+      "loss": 0.1398,
+      "step": 1313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.838176486270329e-06,
+      "loss": 0.1267,
+      "step": 1314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.838043188483071e-06,
+      "loss": 0.1651,
+      "step": 1315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.837909890695814e-06,
+      "loss": 0.1082,
+      "step": 1316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.837776592908559e-06,
+      "loss": 0.1274,
+      "step": 1317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.837643295121302e-06,
+      "loss": 0.1604,
+      "step": 1318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.837509997334044e-06,
+      "loss": 0.1613,
+      "step": 1319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.837376699546787e-06,
+      "loss": 0.098,
+      "step": 1320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.837243401759532e-06,
+      "loss": 0.1148,
+      "step": 1321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.837110103972275e-06,
+      "loss": 0.1866,
+      "step": 1322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.836976806185017e-06,
+      "loss": 0.191,
+      "step": 1323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83684350839776e-06,
+      "loss": 0.1493,
+      "step": 1324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.836710210610505e-06,
+      "loss": 0.1271,
+      "step": 1325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.836576912823247e-06,
+      "loss": 0.135,
+      "step": 1326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83644361503599e-06,
+      "loss": 0.1167,
+      "step": 1327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.836310317248733e-06,
+      "loss": 0.1408,
+      "step": 1328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.836177019461478e-06,
+      "loss": 0.1655,
+      "step": 1329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83604372167422e-06,
+      "loss": 0.1438,
+      "step": 1330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.835910423886963e-06,
+      "loss": 0.218,
+      "step": 1331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.835777126099708e-06,
+      "loss": 0.1425,
+      "step": 1332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83564382831245e-06,
+      "loss": 0.1564,
+      "step": 1333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.835510530525193e-06,
+      "loss": 0.1394,
+      "step": 1334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.835377232737936e-06,
+      "loss": 0.1369,
+      "step": 1335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83524393495068e-06,
+      "loss": 0.1384,
+      "step": 1336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.835110637163424e-06,
+      "loss": 0.1463,
+      "step": 1337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.834977339376166e-06,
+      "loss": 0.1379,
+      "step": 1338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.834844041588911e-06,
+      "loss": 0.1415,
+      "step": 1339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.834710743801654e-06,
+      "loss": 0.1824,
+      "step": 1340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.834577446014397e-06,
+      "loss": 0.1994,
+      "step": 1341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83444414822714e-06,
+      "loss": 0.149,
+      "step": 1342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.834310850439884e-06,
+      "loss": 0.1501,
+      "step": 1343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.834177552652627e-06,
+      "loss": 0.1468,
+      "step": 1344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83404425486537e-06,
+      "loss": 0.1843,
+      "step": 1345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.833910957078114e-06,
+      "loss": 0.1689,
+      "step": 1346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.833777659290857e-06,
+      "loss": 0.097,
+      "step": 1347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.8336443615036e-06,
+      "loss": 0.1179,
+      "step": 1348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.833511063716342e-06,
+      "loss": 0.1773,
+      "step": 1349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.833377765929087e-06,
+      "loss": 0.111,
+      "step": 1350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83324446814183e-06,
+      "loss": 0.1147,
+      "step": 1351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.833111170354573e-06,
+      "loss": 0.1492,
+      "step": 1352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.832977872567317e-06,
+      "loss": 0.1335,
+      "step": 1353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83284457478006e-06,
+      "loss": 0.1356,
+      "step": 1354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.832711276992803e-06,
+      "loss": 0.1394,
+      "step": 1355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.832577979205546e-06,
+      "loss": 0.1976,
+      "step": 1356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83244468141829e-06,
+      "loss": 0.1188,
+      "step": 1357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.832311383631033e-06,
+      "loss": 0.1266,
+      "step": 1358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.832178085843776e-06,
+      "loss": 0.1649,
+      "step": 1359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83204478805652e-06,
+      "loss": 0.1351,
+      "step": 1360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.831911490269263e-06,
+      "loss": 0.1883,
+      "step": 1361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.831778192482006e-06,
+      "loss": 0.1735,
+      "step": 1362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.831644894694749e-06,
+      "loss": 0.1577,
+      "step": 1363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.831511596907493e-06,
+      "loss": 0.1288,
+      "step": 1364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.831378299120236e-06,
+      "loss": 0.1658,
+      "step": 1365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.831245001332979e-06,
+      "loss": 0.1476,
+      "step": 1366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.831111703545722e-06,
+      "loss": 0.1014,
+      "step": 1367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.830978405758466e-06,
+      "loss": 0.1267,
+      "step": 1368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.830845107971209e-06,
+      "loss": 0.1506,
+      "step": 1369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.830711810183952e-06,
+      "loss": 0.1443,
+      "step": 1370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.830578512396695e-06,
+      "loss": 0.1313,
+      "step": 1371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83044521460944e-06,
+      "loss": 0.1227,
+      "step": 1372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.830311916822182e-06,
+      "loss": 0.1404,
+      "step": 1373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.830178619034925e-06,
+      "loss": 0.1297,
+      "step": 1374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.830045321247668e-06,
+      "loss": 0.1788,
+      "step": 1375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.829912023460412e-06,
+      "loss": 0.164,
+      "step": 1376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.829778725673155e-06,
+      "loss": 0.117,
+      "step": 1377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.829645427885898e-06,
+      "loss": 0.1179,
+      "step": 1378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.82951213009864e-06,
+      "loss": 0.139,
+      "step": 1379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.829378832311385e-06,
+      "loss": 0.2051,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.829245534524128e-06,
+      "loss": 0.1343,
+      "step": 1381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.82911223673687e-06,
+      "loss": 0.1447,
+      "step": 1382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.828978938949614e-06,
+      "loss": 0.1441,
+      "step": 1383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.828845641162358e-06,
+      "loss": 0.1306,
+      "step": 1384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.828712343375101e-06,
+      "loss": 0.1501,
+      "step": 1385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.828579045587844e-06,
+      "loss": 0.1437,
+      "step": 1386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.828445747800587e-06,
+      "loss": 0.1449,
+      "step": 1387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.828312450013331e-06,
+      "loss": 0.1147,
+      "step": 1388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.828179152226074e-06,
+      "loss": 0.1477,
+      "step": 1389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.828045854438817e-06,
+      "loss": 0.1484,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.82791255665156e-06,
+      "loss": 0.1157,
+      "step": 1391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.827779258864304e-06,
+      "loss": 0.1364,
+      "step": 1392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.827645961077047e-06,
+      "loss": 0.1453,
+      "step": 1393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.82751266328979e-06,
+      "loss": 0.1527,
+      "step": 1394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.827379365502533e-06,
+      "loss": 0.1373,
+      "step": 1395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.827246067715277e-06,
+      "loss": 0.131,
+      "step": 1396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.82711276992802e-06,
+      "loss": 0.1874,
+      "step": 1397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.826979472140763e-06,
+      "loss": 0.1285,
+      "step": 1398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.826846174353507e-06,
+      "loss": 0.1975,
+      "step": 1399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.82671287656625e-06,
+      "loss": 0.1625,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.826579578778993e-06,
+      "loss": 0.2216,
+      "step": 1401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.826446280991736e-06,
+      "loss": 0.1442,
+      "step": 1402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.82631298320448e-06,
+      "loss": 0.121,
+      "step": 1403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.826179685417223e-06,
+      "loss": 0.1845,
+      "step": 1404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.826046387629966e-06,
+      "loss": 0.1393,
+      "step": 1405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.825913089842709e-06,
+      "loss": 0.1302,
+      "step": 1406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.825779792055453e-06,
+      "loss": 0.1867,
+      "step": 1407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.825646494268196e-06,
+      "loss": 0.1665,
+      "step": 1408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.825513196480939e-06,
+      "loss": 0.1191,
+      "step": 1409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.825379898693682e-06,
+      "loss": 0.1181,
+      "step": 1410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.825246600906426e-06,
+      "loss": 0.1961,
+      "step": 1411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.825113303119169e-06,
+      "loss": 0.1187,
+      "step": 1412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.824980005331912e-06,
+      "loss": 0.1631,
+      "step": 1413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.824846707544655e-06,
+      "loss": 0.1239,
+      "step": 1414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.824713409757399e-06,
+      "loss": 0.1156,
+      "step": 1415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.824580111970142e-06,
+      "loss": 0.1391,
+      "step": 1416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.824446814182885e-06,
+      "loss": 0.1536,
+      "step": 1417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.824313516395628e-06,
+      "loss": 0.1198,
+      "step": 1418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.824180218608372e-06,
+      "loss": 0.1751,
+      "step": 1419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.824046920821115e-06,
+      "loss": 0.1563,
+      "step": 1420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.823913623033858e-06,
+      "loss": 0.1035,
+      "step": 1421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.8237803252466e-06,
+      "loss": 0.1358,
+      "step": 1422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.823647027459345e-06,
+      "loss": 0.1284,
+      "step": 1423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.823513729672088e-06,
+      "loss": 0.1242,
+      "step": 1424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.82338043188483e-06,
+      "loss": 0.165,
+      "step": 1425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.823247134097575e-06,
+      "loss": 0.107,
+      "step": 1426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.823113836310318e-06,
+      "loss": 0.163,
+      "step": 1427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.82298053852306e-06,
+      "loss": 0.1406,
+      "step": 1428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.822847240735804e-06,
+      "loss": 0.1245,
+      "step": 1429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.822713942948548e-06,
+      "loss": 0.126,
+      "step": 1430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.822580645161291e-06,
+      "loss": 0.1406,
+      "step": 1431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.822447347374034e-06,
+      "loss": 0.1518,
+      "step": 1432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.822314049586778e-06,
+      "loss": 0.1498,
+      "step": 1433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.822180751799521e-06,
+      "loss": 0.1804,
+      "step": 1434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.822047454012264e-06,
+      "loss": 0.1049,
+      "step": 1435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.821914156225007e-06,
+      "loss": 0.1829,
+      "step": 1436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.821780858437751e-06,
+      "loss": 0.1576,
+      "step": 1437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.821647560650494e-06,
+      "loss": 0.1616,
+      "step": 1438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.821514262863237e-06,
+      "loss": 0.1432,
+      "step": 1439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.821380965075981e-06,
+      "loss": 0.1437,
+      "step": 1440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.821247667288724e-06,
+      "loss": 0.1554,
+      "step": 1441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.821114369501467e-06,
+      "loss": 0.1769,
+      "step": 1442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.82098107171421e-06,
+      "loss": 0.1379,
+      "step": 1443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.820847773926954e-06,
+      "loss": 0.1534,
+      "step": 1444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.820714476139697e-06,
+      "loss": 0.1527,
+      "step": 1445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.82058117835244e-06,
+      "loss": 0.1452,
+      "step": 1446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.820447880565185e-06,
+      "loss": 0.1046,
+      "step": 1447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.820314582777927e-06,
+      "loss": 0.1395,
+      "step": 1448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.82018128499067e-06,
+      "loss": 0.1899,
+      "step": 1449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.820047987203413e-06,
+      "loss": 0.1275,
+      "step": 1450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.819914689416157e-06,
+      "loss": 0.1513,
+      "step": 1451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.8197813916289e-06,
+      "loss": 0.1013,
+      "step": 1452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.819648093841643e-06,
+      "loss": 0.1444,
+      "step": 1453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.819514796054388e-06,
+      "loss": 0.1307,
+      "step": 1454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.81938149826713e-06,
+      "loss": 0.0958,
+      "step": 1455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.819248200479873e-06,
+      "loss": 0.128,
+      "step": 1456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.819114902692616e-06,
+      "loss": 0.191,
+      "step": 1457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.81898160490536e-06,
+      "loss": 0.1289,
+      "step": 1458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.818848307118103e-06,
+      "loss": 0.1145,
+      "step": 1459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.818715009330846e-06,
+      "loss": 0.1153,
+      "step": 1460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.818581711543589e-06,
+      "loss": 0.1446,
+      "step": 1461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.818448413756334e-06,
+      "loss": 0.1664,
+      "step": 1462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.818315115969076e-06,
+      "loss": 0.115,
+      "step": 1463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.81818181818182e-06,
+      "loss": 0.1544,
+      "step": 1464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.818048520394562e-06,
+      "loss": 0.1226,
+      "step": 1465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.817915222607307e-06,
+      "loss": 0.1252,
+      "step": 1466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.81778192482005e-06,
+      "loss": 0.1397,
+      "step": 1467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.817648627032792e-06,
+      "loss": 0.1209,
+      "step": 1468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.817515329245535e-06,
+      "loss": 0.1615,
+      "step": 1469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.81738203145828e-06,
+      "loss": 0.1518,
+      "step": 1470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.817248733671022e-06,
+      "loss": 0.1117,
+      "step": 1471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.817115435883765e-06,
+      "loss": 0.1666,
+      "step": 1472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.816982138096508e-06,
+      "loss": 0.146,
+      "step": 1473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.816848840309252e-06,
+      "loss": 0.1318,
+      "step": 1474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.816715542521995e-06,
+      "loss": 0.1566,
+      "step": 1475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.816582244734738e-06,
+      "loss": 0.1113,
+      "step": 1476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.816448946947481e-06,
+      "loss": 0.1291,
+      "step": 1477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.816315649160225e-06,
+      "loss": 0.1657,
+      "step": 1478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.816182351372968e-06,
+      "loss": 0.2098,
+      "step": 1479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.816049053585711e-06,
+      "loss": 0.1511,
+      "step": 1480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.815915755798454e-06,
+      "loss": 0.1592,
+      "step": 1481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.815782458011198e-06,
+      "loss": 0.1497,
+      "step": 1482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.815649160223941e-06,
+      "loss": 0.1518,
+      "step": 1483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.815515862436684e-06,
+      "loss": 0.1462,
+      "step": 1484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.815382564649427e-06,
+      "loss": 0.1473,
+      "step": 1485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.815249266862171e-06,
+      "loss": 0.1468,
+      "step": 1486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.815115969074914e-06,
+      "loss": 0.118,
+      "step": 1487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.814982671287657e-06,
+      "loss": 0.111,
+      "step": 1488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.8148493735004e-06,
+      "loss": 0.1452,
+      "step": 1489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.814716075713144e-06,
+      "loss": 0.1109,
+      "step": 1490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.814582777925887e-06,
+      "loss": 0.1761,
+      "step": 1491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.81444948013863e-06,
+      "loss": 0.1219,
+      "step": 1492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.814316182351373e-06,
+      "loss": 0.1749,
+      "step": 1493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.814182884564117e-06,
+      "loss": 0.1278,
+      "step": 1494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.81404958677686e-06,
+      "loss": 0.1392,
+      "step": 1495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.813916288989603e-06,
+      "loss": 0.216,
+      "step": 1496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.813782991202346e-06,
+      "loss": 0.1389,
+      "step": 1497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.81364969341509e-06,
+      "loss": 0.1542,
+      "step": 1498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.813516395627833e-06,
+      "loss": 0.1655,
+      "step": 1499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.813383097840576e-06,
+      "loss": 0.158,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.813249800053319e-06,
+      "loss": 0.1367,
+      "step": 1501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.813116502266063e-06,
+      "loss": 0.1287,
+      "step": 1502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.812983204478806e-06,
+      "loss": 0.1525,
+      "step": 1503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.812849906691549e-06,
+      "loss": 0.1405,
+      "step": 1504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.812716608904292e-06,
+      "loss": 0.1071,
+      "step": 1505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.812583311117036e-06,
+      "loss": 0.1519,
+      "step": 1506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.812450013329779e-06,
+      "loss": 0.1303,
+      "step": 1507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.812316715542522e-06,
+      "loss": 0.1194,
+      "step": 1508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.812183417755266e-06,
+      "loss": 0.1356,
+      "step": 1509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.81205011996801e-06,
+      "loss": 0.1597,
+      "step": 1510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.811916822180752e-06,
+      "loss": 0.129,
+      "step": 1511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.811783524393495e-06,
+      "loss": 0.1333,
+      "step": 1512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.81165022660624e-06,
+      "loss": 0.1241,
+      "step": 1513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.811516928818982e-06,
+      "loss": 0.1581,
+      "step": 1514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.811383631031725e-06,
+      "loss": 0.1279,
+      "step": 1515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.811250333244468e-06,
+      "loss": 0.1723,
+      "step": 1516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.811117035457212e-06,
+      "loss": 0.2155,
+      "step": 1517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.810983737669955e-06,
+      "loss": 0.1333,
+      "step": 1518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.810850439882698e-06,
+      "loss": 0.1721,
+      "step": 1519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.810717142095442e-06,
+      "loss": 0.1272,
+      "step": 1520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.810583844308185e-06,
+      "loss": 0.1474,
+      "step": 1521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.810450546520928e-06,
+      "loss": 0.1538,
+      "step": 1522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.810317248733671e-06,
+      "loss": 0.1513,
+      "step": 1523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.810183950946415e-06,
+      "loss": 0.1312,
+      "step": 1524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.810050653159158e-06,
+      "loss": 0.1216,
+      "step": 1525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.809917355371901e-06,
+      "loss": 0.143,
+      "step": 1526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.809784057584646e-06,
+      "loss": 0.1626,
+      "step": 1527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.809650759797388e-06,
+      "loss": 0.1183,
+      "step": 1528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.809517462010131e-06,
+      "loss": 0.1417,
+      "step": 1529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.809384164222874e-06,
+      "loss": 0.1144,
+      "step": 1530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.809250866435619e-06,
+      "loss": 0.1325,
+      "step": 1531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.809117568648361e-06,
+      "loss": 0.1181,
+      "step": 1532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.808984270861104e-06,
+      "loss": 0.0977,
+      "step": 1533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.808850973073849e-06,
+      "loss": 0.1595,
+      "step": 1534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.808717675286592e-06,
+      "loss": 0.1598,
+      "step": 1535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.808584377499334e-06,
+      "loss": 0.1638,
+      "step": 1536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.808451079712077e-06,
+      "loss": 0.0842,
+      "step": 1537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.808317781924822e-06,
+      "loss": 0.1765,
+      "step": 1538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.808184484137565e-06,
+      "loss": 0.1466,
+      "step": 1539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.808051186350307e-06,
+      "loss": 0.1186,
+      "step": 1540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.807917888563052e-06,
+      "loss": 0.1364,
+      "step": 1541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.807784590775795e-06,
+      "loss": 0.1395,
+      "step": 1542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.807651292988537e-06,
+      "loss": 0.172,
+      "step": 1543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.80751799520128e-06,
+      "loss": 0.1899,
+      "step": 1544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.807384697414025e-06,
+      "loss": 0.1408,
+      "step": 1545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.807251399626768e-06,
+      "loss": 0.148,
+      "step": 1546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.80711810183951e-06,
+      "loss": 0.2109,
+      "step": 1547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.806984804052253e-06,
+      "loss": 0.1319,
+      "step": 1548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.806851506264998e-06,
+      "loss": 0.1039,
+      "step": 1549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.80671820847774e-06,
+      "loss": 0.1406,
+      "step": 1550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.806584910690483e-06,
+      "loss": 0.1135,
+      "step": 1551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.806451612903226e-06,
+      "loss": 0.1363,
+      "step": 1552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.80631831511597e-06,
+      "loss": 0.096,
+      "step": 1553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.806185017328714e-06,
+      "loss": 0.1893,
+      "step": 1554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.806051719541456e-06,
+      "loss": 0.1662,
+      "step": 1555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.8059184217542e-06,
+      "loss": 0.1331,
+      "step": 1556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.805785123966944e-06,
+      "loss": 0.1287,
+      "step": 1557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.805651826179687e-06,
+      "loss": 0.1048,
+      "step": 1558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.80551852839243e-06,
+      "loss": 0.1321,
+      "step": 1559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.805385230605172e-06,
+      "loss": 0.1536,
+      "step": 1560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.805251932817917e-06,
+      "loss": 0.1155,
+      "step": 1561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.80511863503066e-06,
+      "loss": 0.1313,
+      "step": 1562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.804985337243402e-06,
+      "loss": 0.12,
+      "step": 1563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.804852039456145e-06,
+      "loss": 0.1154,
+      "step": 1564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.80471874166889e-06,
+      "loss": 0.1153,
+      "step": 1565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.804585443881632e-06,
+      "loss": 0.1384,
+      "step": 1566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.804452146094375e-06,
+      "loss": 0.1785,
+      "step": 1567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.80431884830712e-06,
+      "loss": 0.1313,
+      "step": 1568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.804185550519863e-06,
+      "loss": 0.1414,
+      "step": 1569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.804052252732605e-06,
+      "loss": 0.2052,
+      "step": 1570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.803918954945348e-06,
+      "loss": 0.1409,
+      "step": 1571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.803785657158093e-06,
+      "loss": 0.1208,
+      "step": 1572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.803652359370836e-06,
+      "loss": 0.1406,
+      "step": 1573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.803519061583578e-06,
+      "loss": 0.1629,
+      "step": 1574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.803385763796321e-06,
+      "loss": 0.1563,
+      "step": 1575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.803252466009066e-06,
+      "loss": 0.1592,
+      "step": 1576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.803119168221809e-06,
+      "loss": 0.0897,
+      "step": 1577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.802985870434551e-06,
+      "loss": 0.1139,
+      "step": 1578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.802852572647294e-06,
+      "loss": 0.1311,
+      "step": 1579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.802719274860039e-06,
+      "loss": 0.1239,
+      "step": 1580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.802585977072782e-06,
+      "loss": 0.1483,
+      "step": 1581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.802452679285524e-06,
+      "loss": 0.1473,
+      "step": 1582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.802319381498267e-06,
+      "loss": 0.1188,
+      "step": 1583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.802186083711012e-06,
+      "loss": 0.1188,
+      "step": 1584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.802052785923755e-06,
+      "loss": 0.152,
+      "step": 1585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.801919488136497e-06,
+      "loss": 0.1287,
+      "step": 1586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.80178619034924e-06,
+      "loss": 0.1736,
+      "step": 1587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.801652892561985e-06,
+      "loss": 0.1341,
+      "step": 1588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.801519594774727e-06,
+      "loss": 0.1133,
+      "step": 1589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.80138629698747e-06,
+      "loss": 0.1328,
+      "step": 1590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.801252999200213e-06,
+      "loss": 0.1319,
+      "step": 1591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.801119701412958e-06,
+      "loss": 0.1637,
+      "step": 1592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.8009864036257e-06,
+      "loss": 0.1449,
+      "step": 1593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.800853105838443e-06,
+      "loss": 0.1446,
+      "step": 1594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.800719808051186e-06,
+      "loss": 0.0939,
+      "step": 1595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.80058651026393e-06,
+      "loss": 0.1674,
+      "step": 1596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.800453212476673e-06,
+      "loss": 0.1422,
+      "step": 1597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.800319914689416e-06,
+      "loss": 0.1544,
+      "step": 1598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.800186616902159e-06,
+      "loss": 0.1548,
+      "step": 1599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.800053319114904e-06,
+      "loss": 0.1698,
+      "step": 1600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.799920021327646e-06,
+      "loss": 0.1414,
+      "step": 1601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.79978672354039e-06,
+      "loss": 0.1149,
+      "step": 1602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.799653425753132e-06,
+      "loss": 0.1485,
+      "step": 1603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.799520127965877e-06,
+      "loss": 0.1259,
+      "step": 1604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.79938683017862e-06,
+      "loss": 0.1121,
+      "step": 1605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.799253532391362e-06,
+      "loss": 0.1419,
+      "step": 1606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.799120234604107e-06,
+      "loss": 0.147,
+      "step": 1607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.79898693681685e-06,
+      "loss": 0.0957,
+      "step": 1608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.798853639029592e-06,
+      "loss": 0.1392,
+      "step": 1609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.798720341242335e-06,
+      "loss": 0.1481,
+      "step": 1610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.79858704345508e-06,
+      "loss": 0.1757,
+      "step": 1611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.798453745667822e-06,
+      "loss": 0.1667,
+      "step": 1612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.798320447880565e-06,
+      "loss": 0.143,
+      "step": 1613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.79818715009331e-06,
+      "loss": 0.1292,
+      "step": 1614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.798053852306053e-06,
+      "loss": 0.1445,
+      "step": 1615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.797920554518795e-06,
+      "loss": 0.1322,
+      "step": 1616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.797787256731538e-06,
+      "loss": 0.0963,
+      "step": 1617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.797653958944283e-06,
+      "loss": 0.1444,
+      "step": 1618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.797520661157026e-06,
+      "loss": 0.1227,
+      "step": 1619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.797387363369768e-06,
+      "loss": 0.12,
+      "step": 1620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.797254065582513e-06,
+      "loss": 0.1356,
+      "step": 1621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.797120767795256e-06,
+      "loss": 0.1424,
+      "step": 1622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.796987470007999e-06,
+      "loss": 0.159,
+      "step": 1623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.796854172220741e-06,
+      "loss": 0.1148,
+      "step": 1624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.796720874433486e-06,
+      "loss": 0.1069,
+      "step": 1625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.796587576646229e-06,
+      "loss": 0.1265,
+      "step": 1626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.796454278858972e-06,
+      "loss": 0.1181,
+      "step": 1627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.796320981071716e-06,
+      "loss": 0.1216,
+      "step": 1628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.796187683284459e-06,
+      "loss": 0.1486,
+      "step": 1629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.796054385497202e-06,
+      "loss": 0.1136,
+      "step": 1630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.795921087709945e-06,
+      "loss": 0.1384,
+      "step": 1631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.795787789922689e-06,
+      "loss": 0.1474,
+      "step": 1632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.795654492135432e-06,
+      "loss": 0.1579,
+      "step": 1633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.795521194348175e-06,
+      "loss": 0.1364,
+      "step": 1634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.79538789656092e-06,
+      "loss": 0.1174,
+      "step": 1635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.795254598773662e-06,
+      "loss": 0.1488,
+      "step": 1636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.795121300986405e-06,
+      "loss": 0.1453,
+      "step": 1637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.794988003199148e-06,
+      "loss": 0.159,
+      "step": 1638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.794854705411892e-06,
+      "loss": 0.1027,
+      "step": 1639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.794721407624635e-06,
+      "loss": 0.1115,
+      "step": 1640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.794588109837378e-06,
+      "loss": 0.1476,
+      "step": 1641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.79445481205012e-06,
+      "loss": 0.1531,
+      "step": 1642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.794321514262865e-06,
+      "loss": 0.1294,
+      "step": 1643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.794188216475608e-06,
+      "loss": 0.1569,
+      "step": 1644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.79405491868835e-06,
+      "loss": 0.1246,
+      "step": 1645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.793921620901094e-06,
+      "loss": 0.1588,
+      "step": 1646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.793788323113838e-06,
+      "loss": 0.146,
+      "step": 1647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.793655025326581e-06,
+      "loss": 0.1474,
+      "step": 1648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.793521727539324e-06,
+      "loss": 0.1509,
+      "step": 1649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.793388429752067e-06,
+      "loss": 0.0878,
+      "step": 1650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.793255131964811e-06,
+      "loss": 0.1768,
+      "step": 1651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.793121834177554e-06,
+      "loss": 0.0965,
+      "step": 1652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.792988536390297e-06,
+      "loss": 0.1104,
+      "step": 1653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.79285523860304e-06,
+      "loss": 0.1612,
+      "step": 1654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.792721940815784e-06,
+      "loss": 0.1361,
+      "step": 1655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.792588643028527e-06,
+      "loss": 0.1401,
+      "step": 1656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.79245534524127e-06,
+      "loss": 0.1355,
+      "step": 1657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.792322047454012e-06,
+      "loss": 0.1558,
+      "step": 1658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.792188749666757e-06,
+      "loss": 0.1973,
+      "step": 1659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.7920554518795e-06,
+      "loss": 0.1393,
+      "step": 1660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.791922154092243e-06,
+      "loss": 0.1168,
+      "step": 1661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.791788856304985e-06,
+      "loss": 0.1036,
+      "step": 1662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.79165555851773e-06,
+      "loss": 0.164,
+      "step": 1663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.791522260730473e-06,
+      "loss": 0.1338,
+      "step": 1664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.791388962943216e-06,
+      "loss": 0.1511,
+      "step": 1665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.791255665155958e-06,
+      "loss": 0.1175,
+      "step": 1666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.791122367368703e-06,
+      "loss": 0.1763,
+      "step": 1667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.790989069581446e-06,
+      "loss": 0.1217,
+      "step": 1668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.790855771794189e-06,
+      "loss": 0.1055,
+      "step": 1669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.790722474006931e-06,
+      "loss": 0.1193,
+      "step": 1670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.790589176219676e-06,
+      "loss": 0.1046,
+      "step": 1671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.790455878432419e-06,
+      "loss": 0.1131,
+      "step": 1672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.790322580645162e-06,
+      "loss": 0.1414,
+      "step": 1673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.790189282857904e-06,
+      "loss": 0.1326,
+      "step": 1674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.790055985070649e-06,
+      "loss": 0.1659,
+      "step": 1675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.789922687283392e-06,
+      "loss": 0.0931,
+      "step": 1676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.789789389496135e-06,
+      "loss": 0.1283,
+      "step": 1677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.789656091708879e-06,
+      "loss": 0.1311,
+      "step": 1678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.789522793921622e-06,
+      "loss": 0.1428,
+      "step": 1679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.789389496134365e-06,
+      "loss": 0.1506,
+      "step": 1680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.789256198347107e-06,
+      "loss": 0.16,
+      "step": 1681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.789122900559852e-06,
+      "loss": 0.1282,
+      "step": 1682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.788989602772595e-06,
+      "loss": 0.0957,
+      "step": 1683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.788856304985338e-06,
+      "loss": 0.1457,
+      "step": 1684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.78872300719808e-06,
+      "loss": 0.1581,
+      "step": 1685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.788589709410825e-06,
+      "loss": 0.1442,
+      "step": 1686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.788456411623568e-06,
+      "loss": 0.1172,
+      "step": 1687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.78832311383631e-06,
+      "loss": 0.1446,
+      "step": 1688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.788189816049053e-06,
+      "loss": 0.0777,
+      "step": 1689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.788056518261798e-06,
+      "loss": 0.17,
+      "step": 1690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.78792322047454e-06,
+      "loss": 0.1509,
+      "step": 1691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.787789922687284e-06,
+      "loss": 0.108,
+      "step": 1692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.787656624900026e-06,
+      "loss": 0.1631,
+      "step": 1693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.787523327112771e-06,
+      "loss": 0.1112,
+      "step": 1694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.787390029325514e-06,
+      "loss": 0.136,
+      "step": 1695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.787256731538257e-06,
+      "loss": 0.1447,
+      "step": 1696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.787123433751e-06,
+      "loss": 0.1584,
+      "step": 1697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.786990135963744e-06,
+      "loss": 0.1452,
+      "step": 1698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.786856838176487e-06,
+      "loss": 0.1256,
+      "step": 1699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.78672354038923e-06,
+      "loss": 0.1198,
+      "step": 1700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.786590242601974e-06,
+      "loss": 0.1098,
+      "step": 1701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.786456944814717e-06,
+      "loss": 0.1539,
+      "step": 1702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.78632364702746e-06,
+      "loss": 0.138,
+      "step": 1703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.786190349240202e-06,
+      "loss": 0.138,
+      "step": 1704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.786057051452947e-06,
+      "loss": 0.1094,
+      "step": 1705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.78592375366569e-06,
+      "loss": 0.143,
+      "step": 1706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.785790455878433e-06,
+      "loss": 0.1189,
+      "step": 1707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.785657158091177e-06,
+      "loss": 0.1411,
+      "step": 1708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.78552386030392e-06,
+      "loss": 0.129,
+      "step": 1709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.785390562516663e-06,
+      "loss": 0.1183,
+      "step": 1710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.785257264729406e-06,
+      "loss": 0.1298,
+      "step": 1711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.78512396694215e-06,
+      "loss": 0.1565,
+      "step": 1712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.784990669154893e-06,
+      "loss": 0.1159,
+      "step": 1713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.784857371367636e-06,
+      "loss": 0.1205,
+      "step": 1714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.78472407358038e-06,
+      "loss": 0.1759,
+      "step": 1715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.784590775793123e-06,
+      "loss": 0.158,
+      "step": 1716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.784457478005866e-06,
+      "loss": 0.1755,
+      "step": 1717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.784324180218609e-06,
+      "loss": 0.1219,
+      "step": 1718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.784190882431353e-06,
+      "loss": 0.1663,
+      "step": 1719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.784057584644096e-06,
+      "loss": 0.1365,
+      "step": 1720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.783924286856839e-06,
+      "loss": 0.1323,
+      "step": 1721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.783790989069583e-06,
+      "loss": 0.1096,
+      "step": 1722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.783657691282326e-06,
+      "loss": 0.1769,
+      "step": 1723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.783524393495069e-06,
+      "loss": 0.1117,
+      "step": 1724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.783391095707812e-06,
+      "loss": 0.1006,
+      "step": 1725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.783257797920556e-06,
+      "loss": 0.1517,
+      "step": 1726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.7831245001333e-06,
+      "loss": 0.1064,
+      "step": 1727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.782991202346042e-06,
+      "loss": 0.1322,
+      "step": 1728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.782857904558785e-06,
+      "loss": 0.0953,
+      "step": 1729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.78272460677153e-06,
+      "loss": 0.1349,
+      "step": 1730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.782591308984272e-06,
+      "loss": 0.1127,
+      "step": 1731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.782458011197015e-06,
+      "loss": 0.1106,
+      "step": 1732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.782324713409758e-06,
+      "loss": 0.1498,
+      "step": 1733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.782191415622502e-06,
+      "loss": 0.1063,
+      "step": 1734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.782058117835245e-06,
+      "loss": 0.1406,
+      "step": 1735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.781924820047988e-06,
+      "loss": 0.1521,
+      "step": 1736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.781791522260732e-06,
+      "loss": 0.1195,
+      "step": 1737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.781658224473475e-06,
+      "loss": 0.1207,
+      "step": 1738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.781524926686218e-06,
+      "loss": 0.1391,
+      "step": 1739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.781391628898961e-06,
+      "loss": 0.1398,
+      "step": 1740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.781258331111705e-06,
+      "loss": 0.1228,
+      "step": 1741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.781125033324448e-06,
+      "loss": 0.1363,
+      "step": 1742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.780991735537191e-06,
+      "loss": 0.1378,
+      "step": 1743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.780858437749934e-06,
+      "loss": 0.1119,
+      "step": 1744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.780725139962678e-06,
+      "loss": 0.155,
+      "step": 1745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.780591842175421e-06,
+      "loss": 0.1424,
+      "step": 1746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.780458544388164e-06,
+      "loss": 0.1536,
+      "step": 1747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.780325246600907e-06,
+      "loss": 0.0943,
+      "step": 1748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.780191948813651e-06,
+      "loss": 0.1506,
+      "step": 1749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.780058651026394e-06,
+      "loss": 0.1132,
+      "step": 1750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.779925353239137e-06,
+      "loss": 0.1422,
+      "step": 1751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77979205545188e-06,
+      "loss": 0.1338,
+      "step": 1752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.779658757664624e-06,
+      "loss": 0.1814,
+      "step": 1753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.779525459877367e-06,
+      "loss": 0.12,
+      "step": 1754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77939216209011e-06,
+      "loss": 0.1689,
+      "step": 1755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.779258864302853e-06,
+      "loss": 0.1059,
+      "step": 1756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.779125566515597e-06,
+      "loss": 0.1362,
+      "step": 1757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77899226872834e-06,
+      "loss": 0.1212,
+      "step": 1758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.778858970941083e-06,
+      "loss": 0.1392,
+      "step": 1759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.778725673153826e-06,
+      "loss": 0.1904,
+      "step": 1760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77859237536657e-06,
+      "loss": 0.1618,
+      "step": 1761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.778459077579313e-06,
+      "loss": 0.121,
+      "step": 1762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.778325779792056e-06,
+      "loss": 0.1183,
+      "step": 1763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.778192482004799e-06,
+      "loss": 0.138,
+      "step": 1764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.778059184217543e-06,
+      "loss": 0.1561,
+      "step": 1765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.777925886430286e-06,
+      "loss": 0.1445,
+      "step": 1766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.777792588643029e-06,
+      "loss": 0.1366,
+      "step": 1767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.777659290855772e-06,
+      "loss": 0.1508,
+      "step": 1768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.777525993068516e-06,
+      "loss": 0.1167,
+      "step": 1769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.777392695281259e-06,
+      "loss": 0.1271,
+      "step": 1770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.777259397494002e-06,
+      "loss": 0.139,
+      "step": 1771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.777126099706745e-06,
+      "loss": 0.1475,
+      "step": 1772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77699280191949e-06,
+      "loss": 0.1098,
+      "step": 1773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.776859504132232e-06,
+      "loss": 0.1701,
+      "step": 1774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.776726206344975e-06,
+      "loss": 0.1366,
+      "step": 1775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.776592908557718e-06,
+      "loss": 0.1266,
+      "step": 1776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.776459610770462e-06,
+      "loss": 0.1086,
+      "step": 1777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.776326312983205e-06,
+      "loss": 0.1463,
+      "step": 1778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.776193015195948e-06,
+      "loss": 0.1563,
+      "step": 1779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77605971740869e-06,
+      "loss": 0.1593,
+      "step": 1780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.775926419621435e-06,
+      "loss": 0.1155,
+      "step": 1781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.775793121834178e-06,
+      "loss": 0.1376,
+      "step": 1782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77565982404692e-06,
+      "loss": 0.135,
+      "step": 1783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.775526526259664e-06,
+      "loss": 0.1665,
+      "step": 1784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.775393228472408e-06,
+      "loss": 0.111,
+      "step": 1785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.775259930685151e-06,
+      "loss": 0.1414,
+      "step": 1786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.775126632897894e-06,
+      "loss": 0.1146,
+      "step": 1787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.774993335110638e-06,
+      "loss": 0.1146,
+      "step": 1788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.774860037323381e-06,
+      "loss": 0.1142,
+      "step": 1789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.774726739536124e-06,
+      "loss": 0.1382,
+      "step": 1790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.774593441748867e-06,
+      "loss": 0.1291,
+      "step": 1791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.774460143961611e-06,
+      "loss": 0.1205,
+      "step": 1792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.774326846174354e-06,
+      "loss": 0.1495,
+      "step": 1793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.774193548387097e-06,
+      "loss": 0.1579,
+      "step": 1794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.774060250599841e-06,
+      "loss": 0.1323,
+      "step": 1795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.773926952812584e-06,
+      "loss": 0.1395,
+      "step": 1796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.773793655025327e-06,
+      "loss": 0.1179,
+      "step": 1797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77366035723807e-06,
+      "loss": 0.1135,
+      "step": 1798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.773527059450814e-06,
+      "loss": 0.2038,
+      "step": 1799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.773393761663557e-06,
+      "loss": 0.1902,
+      "step": 1800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.7732604638763e-06,
+      "loss": 0.119,
+      "step": 1801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.773127166089044e-06,
+      "loss": 0.1534,
+      "step": 1802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.772993868301787e-06,
+      "loss": 0.1078,
+      "step": 1803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77286057051453e-06,
+      "loss": 0.0912,
+      "step": 1804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.772727272727273e-06,
+      "loss": 0.1102,
+      "step": 1805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.772593974940017e-06,
+      "loss": 0.1384,
+      "step": 1806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77246067715276e-06,
+      "loss": 0.1403,
+      "step": 1807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.772327379365503e-06,
+      "loss": 0.1266,
+      "step": 1808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.772194081578248e-06,
+      "loss": 0.1436,
+      "step": 1809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77206078379099e-06,
+      "loss": 0.167,
+      "step": 1810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.771927486003733e-06,
+      "loss": 0.1505,
+      "step": 1811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.771794188216476e-06,
+      "loss": 0.1627,
+      "step": 1812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77166089042922e-06,
+      "loss": 0.1118,
+      "step": 1813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.771527592641963e-06,
+      "loss": 0.1544,
+      "step": 1814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.771394294854706e-06,
+      "loss": 0.1682,
+      "step": 1815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77126099706745e-06,
+      "loss": 0.1495,
+      "step": 1816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.771127699280194e-06,
+      "loss": 0.1757,
+      "step": 1817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.770994401492936e-06,
+      "loss": 0.1483,
+      "step": 1818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77086110370568e-06,
+      "loss": 0.107,
+      "step": 1819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.770727805918424e-06,
+      "loss": 0.1152,
+      "step": 1820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.770594508131167e-06,
+      "loss": 0.1475,
+      "step": 1821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77046121034391e-06,
+      "loss": 0.1418,
+      "step": 1822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.770327912556652e-06,
+      "loss": 0.146,
+      "step": 1823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.770194614769397e-06,
+      "loss": 0.1206,
+      "step": 1824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.77006131698214e-06,
+      "loss": 0.1792,
+      "step": 1825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.769928019194882e-06,
+      "loss": 0.1259,
+      "step": 1826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.769794721407625e-06,
+      "loss": 0.1304,
+      "step": 1827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.76966142362037e-06,
+      "loss": 0.1385,
+      "step": 1828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.769528125833112e-06,
+      "loss": 0.1205,
+      "step": 1829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.769394828045855e-06,
+      "loss": 0.0905,
+      "step": 1830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.769261530258598e-06,
+      "loss": 0.137,
+      "step": 1831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.769128232471343e-06,
+      "loss": 0.1113,
+      "step": 1832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.768994934684085e-06,
+      "loss": 0.0988,
+      "step": 1833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.768861636896828e-06,
+      "loss": 0.1314,
+      "step": 1834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.768728339109571e-06,
+      "loss": 0.1535,
+      "step": 1835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.768595041322316e-06,
+      "loss": 0.1246,
+      "step": 1836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.768461743535058e-06,
+      "loss": 0.1274,
+      "step": 1837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.768328445747801e-06,
+      "loss": 0.1472,
+      "step": 1838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.768195147960544e-06,
+      "loss": 0.1248,
+      "step": 1839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.768061850173289e-06,
+      "loss": 0.1242,
+      "step": 1840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.767928552386031e-06,
+      "loss": 0.1134,
+      "step": 1841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.767795254598774e-06,
+      "loss": 0.1543,
+      "step": 1842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.767661956811517e-06,
+      "loss": 0.174,
+      "step": 1843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.767528659024262e-06,
+      "loss": 0.1264,
+      "step": 1844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.767395361237004e-06,
+      "loss": 0.1095,
+      "step": 1845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.767262063449747e-06,
+      "loss": 0.1232,
+      "step": 1846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.767128765662492e-06,
+      "loss": 0.0957,
+      "step": 1847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.766995467875234e-06,
+      "loss": 0.1329,
+      "step": 1848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.766862170087977e-06,
+      "loss": 0.0959,
+      "step": 1849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.76672887230072e-06,
+      "loss": 0.1291,
+      "step": 1850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.766595574513465e-06,
+      "loss": 0.12,
+      "step": 1851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.766462276726207e-06,
+      "loss": 0.1454,
+      "step": 1852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.76632897893895e-06,
+      "loss": 0.1606,
+      "step": 1853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.766195681151693e-06,
+      "loss": 0.1409,
+      "step": 1854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.766062383364438e-06,
+      "loss": 0.1063,
+      "step": 1855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.76592908557718e-06,
+      "loss": 0.1121,
+      "step": 1856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.765795787789923e-06,
+      "loss": 0.1548,
+      "step": 1857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.765662490002666e-06,
+      "loss": 0.1909,
+      "step": 1858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.76552919221541e-06,
+      "loss": 0.115,
+      "step": 1859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.765395894428153e-06,
+      "loss": 0.1573,
+      "step": 1860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.765262596640896e-06,
+      "loss": 0.1227,
+      "step": 1861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.765129298853639e-06,
+      "loss": 0.1872,
+      "step": 1862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.764996001066384e-06,
+      "loss": 0.1069,
+      "step": 1863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.764862703279126e-06,
+      "loss": 0.1331,
+      "step": 1864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.76472940549187e-06,
+      "loss": 0.1046,
+      "step": 1865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.764596107704612e-06,
+      "loss": 0.1954,
+      "step": 1866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.764462809917357e-06,
+      "loss": 0.103,
+      "step": 1867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.7643295121301e-06,
+      "loss": 0.1296,
+      "step": 1868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.764196214342842e-06,
+      "loss": 0.1528,
+      "step": 1869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.764062916555585e-06,
+      "loss": 0.1346,
+      "step": 1870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.76392961876833e-06,
+      "loss": 0.0964,
+      "step": 1871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.763796320981072e-06,
+      "loss": 0.1312,
+      "step": 1872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.763663023193815e-06,
+      "loss": 0.1098,
+      "step": 1873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.763529725406558e-06,
+      "loss": 0.1417,
+      "step": 1874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.763396427619302e-06,
+      "loss": 0.1186,
+      "step": 1875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.763263129832045e-06,
+      "loss": 0.1102,
+      "step": 1876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.763129832044788e-06,
+      "loss": 0.1238,
+      "step": 1877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.762996534257531e-06,
+      "loss": 0.1405,
+      "step": 1878
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.762863236470275e-06,
+      "loss": 0.1778,
+      "step": 1879
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.762729938683018e-06,
+      "loss": 0.1354,
+      "step": 1880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.762596640895761e-06,
+      "loss": 0.1317,
+      "step": 1881
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.762463343108506e-06,
+      "loss": 0.1562,
+      "step": 1882
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.762330045321248e-06,
+      "loss": 0.1528,
+      "step": 1883
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.762196747533991e-06,
+      "loss": 0.1447,
+      "step": 1884
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.762063449746734e-06,
+      "loss": 0.1243,
+      "step": 1885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.761930151959479e-06,
+      "loss": 0.1213,
+      "step": 1886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.761796854172221e-06,
+      "loss": 0.1456,
+      "step": 1887
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.761663556384964e-06,
+      "loss": 0.1662,
+      "step": 1888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.761530258597709e-06,
+      "loss": 0.1223,
+      "step": 1889
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.761396960810452e-06,
+      "loss": 0.1864,
+      "step": 1890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.761263663023194e-06,
+      "loss": 0.1207,
+      "step": 1891
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.761130365235937e-06,
+      "loss": 0.1477,
+      "step": 1892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.760997067448682e-06,
+      "loss": 0.0921,
+      "step": 1893
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.760863769661424e-06,
+      "loss": 0.1602,
+      "step": 1894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.760730471874167e-06,
+      "loss": 0.1454,
+      "step": 1895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.760597174086912e-06,
+      "loss": 0.0986,
+      "step": 1896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.760463876299655e-06,
+      "loss": 0.1073,
+      "step": 1897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.760330578512397e-06,
+      "loss": 0.1261,
+      "step": 1898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.76019728072514e-06,
+      "loss": 0.1233,
+      "step": 1899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.760063982937885e-06,
+      "loss": 0.123,
+      "step": 1900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.759930685150628e-06,
+      "loss": 0.1512,
+      "step": 1901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75979738736337e-06,
+      "loss": 0.0907,
+      "step": 1902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.759664089576115e-06,
+      "loss": 0.1465,
+      "step": 1903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.759530791788858e-06,
+      "loss": 0.0993,
+      "step": 1904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.7593974940016e-06,
+      "loss": 0.1419,
+      "step": 1905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.759264196214343e-06,
+      "loss": 0.0921,
+      "step": 1906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.759130898427088e-06,
+      "loss": 0.1457,
+      "step": 1907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75899760063983e-06,
+      "loss": 0.1238,
+      "step": 1908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.758864302852574e-06,
+      "loss": 0.1355,
+      "step": 1909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.758731005065318e-06,
+      "loss": 0.1516,
+      "step": 1910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.758597707278061e-06,
+      "loss": 0.1598,
+      "step": 1911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.758464409490804e-06,
+      "loss": 0.1406,
+      "step": 1912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.758331111703547e-06,
+      "loss": 0.1293,
+      "step": 1913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.758197813916291e-06,
+      "loss": 0.096,
+      "step": 1914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.758064516129034e-06,
+      "loss": 0.1782,
+      "step": 1915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.757931218341777e-06,
+      "loss": 0.1232,
+      "step": 1916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75779792055452e-06,
+      "loss": 0.1581,
+      "step": 1917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.757664622767264e-06,
+      "loss": 0.1308,
+      "step": 1918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.757531324980007e-06,
+      "loss": 0.1211,
+      "step": 1919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75739802719275e-06,
+      "loss": 0.1317,
+      "step": 1920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.757264729405492e-06,
+      "loss": 0.1216,
+      "step": 1921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.757131431618237e-06,
+      "loss": 0.1807,
+      "step": 1922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75699813383098e-06,
+      "loss": 0.1097,
+      "step": 1923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.756864836043723e-06,
+      "loss": 0.147,
+      "step": 1924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.756731538256465e-06,
+      "loss": 0.1341,
+      "step": 1925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75659824046921e-06,
+      "loss": 0.1203,
+      "step": 1926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.756464942681953e-06,
+      "loss": 0.1331,
+      "step": 1927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.756331644894696e-06,
+      "loss": 0.1524,
+      "step": 1928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.756198347107438e-06,
+      "loss": 0.1224,
+      "step": 1929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.756065049320183e-06,
+      "loss": 0.1262,
+      "step": 1930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.755931751532926e-06,
+      "loss": 0.1611,
+      "step": 1931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.755798453745669e-06,
+      "loss": 0.098,
+      "step": 1932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.755665155958411e-06,
+      "loss": 0.1169,
+      "step": 1933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.755531858171156e-06,
+      "loss": 0.1093,
+      "step": 1934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.755398560383899e-06,
+      "loss": 0.2282,
+      "step": 1935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.755265262596642e-06,
+      "loss": 0.1635,
+      "step": 1936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.755131964809384e-06,
+      "loss": 0.1107,
+      "step": 1937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.754998667022129e-06,
+      "loss": 0.1308,
+      "step": 1938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.754865369234872e-06,
+      "loss": 0.1069,
+      "step": 1939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.754732071447615e-06,
+      "loss": 0.1555,
+      "step": 1940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.754598773660357e-06,
+      "loss": 0.1802,
+      "step": 1941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.754465475873102e-06,
+      "loss": 0.1327,
+      "step": 1942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.754332178085845e-06,
+      "loss": 0.1564,
+      "step": 1943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.754198880298587e-06,
+      "loss": 0.1397,
+      "step": 1944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75406558251133e-06,
+      "loss": 0.1214,
+      "step": 1945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.753932284724075e-06,
+      "loss": 0.1168,
+      "step": 1946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.753798986936818e-06,
+      "loss": 0.1174,
+      "step": 1947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75366568914956e-06,
+      "loss": 0.102,
+      "step": 1948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.753532391362303e-06,
+      "loss": 0.1305,
+      "step": 1949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.753399093575048e-06,
+      "loss": 0.1607,
+      "step": 1950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75326579578779e-06,
+      "loss": 0.1181,
+      "step": 1951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.753132498000533e-06,
+      "loss": 0.1243,
+      "step": 1952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.752999200213276e-06,
+      "loss": 0.1055,
+      "step": 1953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75286590242602e-06,
+      "loss": 0.0914,
+      "step": 1954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.752732604638764e-06,
+      "loss": 0.1298,
+      "step": 1955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.752599306851506e-06,
+      "loss": 0.1078,
+      "step": 1956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75246600906425e-06,
+      "loss": 0.1526,
+      "step": 1957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.752332711276994e-06,
+      "loss": 0.126,
+      "step": 1958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.752199413489737e-06,
+      "loss": 0.1312,
+      "step": 1959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75206611570248e-06,
+      "loss": 0.1199,
+      "step": 1960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.751932817915224e-06,
+      "loss": 0.1334,
+      "step": 1961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.751799520127967e-06,
+      "loss": 0.1161,
+      "step": 1962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75166622234071e-06,
+      "loss": 0.1195,
+      "step": 1963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.751532924553452e-06,
+      "loss": 0.1148,
+      "step": 1964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.751399626766197e-06,
+      "loss": 0.1487,
+      "step": 1965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75126632897894e-06,
+      "loss": 0.0934,
+      "step": 1966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.751133031191682e-06,
+      "loss": 0.1215,
+      "step": 1967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.750999733404425e-06,
+      "loss": 0.1432,
+      "step": 1968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.75086643561717e-06,
+      "loss": 0.1575,
+      "step": 1969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.750733137829913e-06,
+      "loss": 0.1273,
+      "step": 1970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.750599840042655e-06,
+      "loss": 0.1115,
+      "step": 1971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.750466542255398e-06,
+      "loss": 0.1316,
+      "step": 1972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.750333244468143e-06,
+      "loss": 0.1476,
+      "step": 1973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.750199946680886e-06,
+      "loss": 0.1389,
+      "step": 1974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.750066648893628e-06,
+      "loss": 0.1126,
+      "step": 1975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.749933351106371e-06,
+      "loss": 0.1639,
+      "step": 1976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.749800053319116e-06,
+      "loss": 0.1397,
+      "step": 1977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.749666755531859e-06,
+      "loss": 0.1299,
+      "step": 1978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.749533457744601e-06,
+      "loss": 0.1223,
+      "step": 1979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.749400159957346e-06,
+      "loss": 0.1389,
+      "step": 1980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.749266862170089e-06,
+      "loss": 0.1702,
+      "step": 1981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.749133564382832e-06,
+      "loss": 0.118,
+      "step": 1982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.749000266595574e-06,
+      "loss": 0.152,
+      "step": 1983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.748866968808319e-06,
+      "loss": 0.145,
+      "step": 1984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.748733671021062e-06,
+      "loss": 0.1117,
+      "step": 1985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.748600373233805e-06,
+      "loss": 0.1094,
+      "step": 1986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.748467075446549e-06,
+      "loss": 0.1378,
+      "step": 1987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.748333777659292e-06,
+      "loss": 0.1125,
+      "step": 1988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.748200479872035e-06,
+      "loss": 0.1638,
+      "step": 1989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.748067182084777e-06,
+      "loss": 0.182,
+      "step": 1990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.747933884297522e-06,
+      "loss": 0.1632,
+      "step": 1991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.747800586510265e-06,
+      "loss": 0.1147,
+      "step": 1992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.747667288723008e-06,
+      "loss": 0.1459,
+      "step": 1993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.747533990935752e-06,
+      "loss": 0.1286,
+      "step": 1994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.747400693148495e-06,
+      "loss": 0.1447,
+      "step": 1995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.747267395361238e-06,
+      "loss": 0.1019,
+      "step": 1996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.74713409757398e-06,
+      "loss": 0.1079,
+      "step": 1997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.747000799786725e-06,
+      "loss": 0.1023,
+      "step": 1998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.746867501999468e-06,
+      "loss": 0.1252,
+      "step": 1999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.74673420421221e-06,
+      "loss": 0.1265,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.746600906424955e-06,
+      "loss": 0.152,
+      "step": 2001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.746467608637698e-06,
+      "loss": 0.1737,
+      "step": 2002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.746334310850441e-06,
+      "loss": 0.1302,
+      "step": 2003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.746201013063184e-06,
+      "loss": 0.1398,
+      "step": 2004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.746067715275928e-06,
+      "loss": 0.142,
+      "step": 2005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.745934417488671e-06,
+      "loss": 0.1437,
+      "step": 2006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.745801119701414e-06,
+      "loss": 0.1283,
+      "step": 2007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.745667821914157e-06,
+      "loss": 0.1607,
+      "step": 2008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.745534524126901e-06,
+      "loss": 0.1137,
+      "step": 2009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.745401226339644e-06,
+      "loss": 0.1553,
+      "step": 2010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.745267928552387e-06,
+      "loss": 0.1159,
+      "step": 2011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.74513463076513e-06,
+      "loss": 0.092,
+      "step": 2012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.745001332977874e-06,
+      "loss": 0.1683,
+      "step": 2013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.744868035190617e-06,
+      "loss": 0.1111,
+      "step": 2014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.74473473740336e-06,
+      "loss": 0.1324,
+      "step": 2015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.744601439616104e-06,
+      "loss": 0.109,
+      "step": 2016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.744468141828847e-06,
+      "loss": 0.1253,
+      "step": 2017
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.74433484404159e-06,
+      "loss": 0.111,
+      "step": 2018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.744201546254333e-06,
+      "loss": 0.1493,
+      "step": 2019
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.744068248467077e-06,
+      "loss": 0.1208,
+      "step": 2020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.74393495067982e-06,
+      "loss": 0.1318,
+      "step": 2021
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.743801652892563e-06,
+      "loss": 0.1168,
+      "step": 2022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.743668355105306e-06,
+      "loss": 0.0886,
+      "step": 2023
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.74353505731805e-06,
+      "loss": 0.1234,
+      "step": 2024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.743401759530793e-06,
+      "loss": 0.1899,
+      "step": 2025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.743268461743536e-06,
+      "loss": 0.1265,
+      "step": 2026
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.743135163956279e-06,
+      "loss": 0.1014,
+      "step": 2027
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.743001866169023e-06,
+      "loss": 0.1042,
+      "step": 2028
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.742868568381766e-06,
+      "loss": 0.1498,
+      "step": 2029
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.742735270594509e-06,
+      "loss": 0.1261,
+      "step": 2030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.742601972807252e-06,
+      "loss": 0.1411,
+      "step": 2031
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.742468675019996e-06,
+      "loss": 0.151,
+      "step": 2032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.742335377232739e-06,
+      "loss": 0.1392,
+      "step": 2033
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.742202079445482e-06,
+      "loss": 0.1163,
+      "step": 2034
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.742068781658225e-06,
+      "loss": 0.1659,
+      "step": 2035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.74193548387097e-06,
+      "loss": 0.2141,
+      "step": 2036
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.741802186083712e-06,
+      "loss": 0.1776,
+      "step": 2037
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.741668888296455e-06,
+      "loss": 0.1016,
+      "step": 2038
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.741535590509198e-06,
+      "loss": 0.0897,
+      "step": 2039
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.741402292721942e-06,
+      "loss": 0.1122,
+      "step": 2040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.741268994934685e-06,
+      "loss": 0.1201,
+      "step": 2041
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.741135697147428e-06,
+      "loss": 0.1289,
+      "step": 2042
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.74100239936017e-06,
+      "loss": 0.1059,
+      "step": 2043
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.740869101572915e-06,
+      "loss": 0.0845,
+      "step": 2044
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.740735803785658e-06,
+      "loss": 0.1439,
+      "step": 2045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.7406025059984e-06,
+      "loss": 0.136,
+      "step": 2046
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.740469208211144e-06,
+      "loss": 0.1126,
+      "step": 2047
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.740335910423888e-06,
+      "loss": 0.1516,
+      "step": 2048
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.740202612636631e-06,
+      "loss": 0.1301,
+      "step": 2049
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.740069314849374e-06,
+      "loss": 0.1605,
+      "step": 2050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.739936017062117e-06,
+      "loss": 0.1062,
+      "step": 2051
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.739802719274861e-06,
+      "loss": 0.1162,
+      "step": 2052
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.739669421487604e-06,
+      "loss": 0.1009,
+      "step": 2053
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.739536123700347e-06,
+      "loss": 0.1483,
+      "step": 2054
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.73940282591309e-06,
+      "loss": 0.0846,
+      "step": 2055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.739269528125834e-06,
+      "loss": 0.1309,
+      "step": 2056
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.739136230338577e-06,
+      "loss": 0.0959,
+      "step": 2057
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.73900293255132e-06,
+      "loss": 0.1103,
+      "step": 2058
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.738869634764062e-06,
+      "loss": 0.1448,
+      "step": 2059
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.738736336976807e-06,
+      "loss": 0.0846,
+      "step": 2060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.73860303918955e-06,
+      "loss": 0.1239,
+      "step": 2061
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.738469741402293e-06,
+      "loss": 0.1405,
+      "step": 2062
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.738336443615035e-06,
+      "loss": 0.1084,
+      "step": 2063
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.73820314582778e-06,
+      "loss": 0.1286,
+      "step": 2064
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.738069848040523e-06,
+      "loss": 0.1372,
+      "step": 2065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.737936550253266e-06,
+      "loss": 0.0977,
+      "step": 2066
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.73780325246601e-06,
+      "loss": 0.1504,
+      "step": 2067
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.737669954678753e-06,
+      "loss": 0.0934,
+      "step": 2068
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.737536656891496e-06,
+      "loss": 0.1323,
+      "step": 2069
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.737403359104239e-06,
+      "loss": 0.1554,
+      "step": 2070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.737270061316983e-06,
+      "loss": 0.1157,
+      "step": 2071
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.737136763529726e-06,
+      "loss": 0.1569,
+      "step": 2072
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.737003465742469e-06,
+      "loss": 0.1217,
+      "step": 2073
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.736870167955213e-06,
+      "loss": 0.1317,
+      "step": 2074
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.736736870167956e-06,
+      "loss": 0.126,
+      "step": 2075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.736603572380699e-06,
+      "loss": 0.1192,
+      "step": 2076
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.736470274593442e-06,
+      "loss": 0.1483,
+      "step": 2077
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.736336976806186e-06,
+      "loss": 0.1427,
+      "step": 2078
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.736203679018929e-06,
+      "loss": 0.1232,
+      "step": 2079
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.736070381231672e-06,
+      "loss": 0.1311,
+      "step": 2080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.735937083444416e-06,
+      "loss": 0.1222,
+      "step": 2081
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.73580378565716e-06,
+      "loss": 0.1212,
+      "step": 2082
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.735670487869902e-06,
+      "loss": 0.1038,
+      "step": 2083
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.735537190082645e-06,
+      "loss": 0.1458,
+      "step": 2084
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.73540389229539e-06,
+      "loss": 0.1092,
+      "step": 2085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.735270594508132e-06,
+      "loss": 0.1262,
+      "step": 2086
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.735137296720875e-06,
+      "loss": 0.1153,
+      "step": 2087
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.73500399893362e-06,
+      "loss": 0.1174,
+      "step": 2088
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.734870701146362e-06,
+      "loss": 0.1031,
+      "step": 2089
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.734737403359105e-06,
+      "loss": 0.0967,
+      "step": 2090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.734604105571848e-06,
+      "loss": 0.1313,
+      "step": 2091
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.734470807784592e-06,
+      "loss": 0.1373,
+      "step": 2092
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.734337509997335e-06,
+      "loss": 0.1115,
+      "step": 2093
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.734204212210078e-06,
+      "loss": 0.1507,
+      "step": 2094
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.734070914422823e-06,
+      "loss": 0.1505,
+      "step": 2095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.733937616635565e-06,
+      "loss": 0.1243,
+      "step": 2096
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.733804318848308e-06,
+      "loss": 0.1649,
+      "step": 2097
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.733671021061051e-06,
+      "loss": 0.124,
+      "step": 2098
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.733537723273796e-06,
+      "loss": 0.125,
+      "step": 2099
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.733404425486538e-06,
+      "loss": 0.138,
+      "step": 2100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.733271127699281e-06,
+      "loss": 0.1014,
+      "step": 2101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.733137829912024e-06,
+      "loss": 0.1536,
+      "step": 2102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.733004532124769e-06,
+      "loss": 0.1865,
+      "step": 2103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.732871234337511e-06,
+      "loss": 0.1425,
+      "step": 2104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.732737936550254e-06,
+      "loss": 0.1651,
+      "step": 2105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.732604638762997e-06,
+      "loss": 0.1046,
+      "step": 2106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.732471340975742e-06,
+      "loss": 0.1701,
+      "step": 2107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.732338043188484e-06,
+      "loss": 0.1356,
+      "step": 2108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.732204745401227e-06,
+      "loss": 0.1451,
+      "step": 2109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.73207144761397e-06,
+      "loss": 0.1216,
+      "step": 2110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.731938149826714e-06,
+      "loss": 0.1562,
+      "step": 2111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.731804852039457e-06,
+      "loss": 0.1224,
+      "step": 2112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.7316715542522e-06,
+      "loss": 0.0921,
+      "step": 2113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.731538256464943e-06,
+      "loss": 0.1142,
+      "step": 2114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.731404958677687e-06,
+      "loss": 0.1319,
+      "step": 2115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.73127166089043e-06,
+      "loss": 0.1533,
+      "step": 2116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.731138363103173e-06,
+      "loss": 0.1773,
+      "step": 2117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.731005065315916e-06,
+      "loss": 0.1268,
+      "step": 2118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.73087176752866e-06,
+      "loss": 0.1233,
+      "step": 2119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.730738469741403e-06,
+      "loss": 0.1082,
+      "step": 2120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.730605171954146e-06,
+      "loss": 0.1402,
+      "step": 2121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.730471874166889e-06,
+      "loss": 0.1084,
+      "step": 2122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.730338576379633e-06,
+      "loss": 0.1352,
+      "step": 2123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.730205278592376e-06,
+      "loss": 0.1464,
+      "step": 2124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.730071980805119e-06,
+      "loss": 0.1202,
+      "step": 2125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.729938683017862e-06,
+      "loss": 0.1292,
+      "step": 2126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.729805385230606e-06,
+      "loss": 0.1113,
+      "step": 2127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.72967208744335e-06,
+      "loss": 0.1326,
+      "step": 2128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.729538789656092e-06,
+      "loss": 0.0887,
+      "step": 2129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.729405491868837e-06,
+      "loss": 0.145,
+      "step": 2130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72927219408158e-06,
+      "loss": 0.1211,
+      "step": 2131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.729138896294322e-06,
+      "loss": 0.1509,
+      "step": 2132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.729005598507065e-06,
+      "loss": 0.133,
+      "step": 2133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72887230071981e-06,
+      "loss": 0.0972,
+      "step": 2134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.728739002932552e-06,
+      "loss": 0.1394,
+      "step": 2135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.728605705145295e-06,
+      "loss": 0.1622,
+      "step": 2136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.728472407358038e-06,
+      "loss": 0.1159,
+      "step": 2137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.728339109570782e-06,
+      "loss": 0.1404,
+      "step": 2138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.728205811783525e-06,
+      "loss": 0.0883,
+      "step": 2139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.728072513996268e-06,
+      "loss": 0.1164,
+      "step": 2140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.727939216209011e-06,
+      "loss": 0.12,
+      "step": 2141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.727805918421755e-06,
+      "loss": 0.1356,
+      "step": 2142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.727672620634498e-06,
+      "loss": 0.1226,
+      "step": 2143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.727539322847241e-06,
+      "loss": 0.2081,
+      "step": 2144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.727406025059984e-06,
+      "loss": 0.1468,
+      "step": 2145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.727272727272728e-06,
+      "loss": 0.1634,
+      "step": 2146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.727139429485471e-06,
+      "loss": 0.135,
+      "step": 2147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.727006131698214e-06,
+      "loss": 0.1445,
+      "step": 2148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.726872833910957e-06,
+      "loss": 0.1411,
+      "step": 2149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.726739536123701e-06,
+      "loss": 0.1244,
+      "step": 2150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.726606238336444e-06,
+      "loss": 0.1516,
+      "step": 2151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.726472940549187e-06,
+      "loss": 0.137,
+      "step": 2152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72633964276193e-06,
+      "loss": 0.0863,
+      "step": 2153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.726206344974674e-06,
+      "loss": 0.0895,
+      "step": 2154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.726073047187417e-06,
+      "loss": 0.135,
+      "step": 2155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72593974940016e-06,
+      "loss": 0.1499,
+      "step": 2156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.725806451612903e-06,
+      "loss": 0.125,
+      "step": 2157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.725673153825647e-06,
+      "loss": 0.1397,
+      "step": 2158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72553985603839e-06,
+      "loss": 0.1215,
+      "step": 2159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.725406558251133e-06,
+      "loss": 0.117,
+      "step": 2160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.725273260463877e-06,
+      "loss": 0.1159,
+      "step": 2161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72513996267662e-06,
+      "loss": 0.1108,
+      "step": 2162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.725006664889363e-06,
+      "loss": 0.1352,
+      "step": 2163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.724873367102106e-06,
+      "loss": 0.1298,
+      "step": 2164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72474006931485e-06,
+      "loss": 0.0934,
+      "step": 2165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.724606771527593e-06,
+      "loss": 0.1384,
+      "step": 2166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.724473473740336e-06,
+      "loss": 0.1015,
+      "step": 2167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72434017595308e-06,
+      "loss": 0.122,
+      "step": 2168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.724206878165823e-06,
+      "loss": 0.1132,
+      "step": 2169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.724073580378566e-06,
+      "loss": 0.1103,
+      "step": 2170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.723940282591309e-06,
+      "loss": 0.1221,
+      "step": 2171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.723806984804054e-06,
+      "loss": 0.129,
+      "step": 2172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.723673687016796e-06,
+      "loss": 0.095,
+      "step": 2173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72354038922954e-06,
+      "loss": 0.1196,
+      "step": 2174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.723407091442284e-06,
+      "loss": 0.1518,
+      "step": 2175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.723273793655027e-06,
+      "loss": 0.1067,
+      "step": 2176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72314049586777e-06,
+      "loss": 0.1528,
+      "step": 2177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.723007198080512e-06,
+      "loss": 0.119,
+      "step": 2178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.722873900293257e-06,
+      "loss": 0.1026,
+      "step": 2179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.722740602506e-06,
+      "loss": 0.1365,
+      "step": 2180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.722607304718742e-06,
+      "loss": 0.146,
+      "step": 2181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.722474006931487e-06,
+      "loss": 0.1439,
+      "step": 2182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72234070914423e-06,
+      "loss": 0.0935,
+      "step": 2183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.722207411356972e-06,
+      "loss": 0.113,
+      "step": 2184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.722074113569715e-06,
+      "loss": 0.1226,
+      "step": 2185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72194081578246e-06,
+      "loss": 0.1286,
+      "step": 2186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.721807517995203e-06,
+      "loss": 0.1043,
+      "step": 2187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.721674220207945e-06,
+      "loss": 0.1557,
+      "step": 2188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.72154092242069e-06,
+      "loss": 0.1515,
+      "step": 2189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.721407624633433e-06,
+      "loss": 0.1204,
+      "step": 2190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.721274326846176e-06,
+      "loss": 0.0865,
+      "step": 2191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.721141029058918e-06,
+      "loss": 0.1234,
+      "step": 2192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.721007731271663e-06,
+      "loss": 0.0997,
+      "step": 2193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.720874433484406e-06,
+      "loss": 0.1075,
+      "step": 2194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.720741135697149e-06,
+      "loss": 0.1163,
+      "step": 2195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.720607837909891e-06,
+      "loss": 0.1645,
+      "step": 2196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.720474540122636e-06,
+      "loss": 0.1265,
+      "step": 2197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.720341242335379e-06,
+      "loss": 0.13,
+      "step": 2198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.720207944548122e-06,
+      "loss": 0.1119,
+      "step": 2199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.720074646760864e-06,
+      "loss": 0.1338,
+      "step": 2200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.719941348973609e-06,
+      "loss": 0.1236,
+      "step": 2201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.719808051186352e-06,
+      "loss": 0.1225,
+      "step": 2202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.719674753399094e-06,
+      "loss": 0.128,
+      "step": 2203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.719541455611837e-06,
+      "loss": 0.0845,
+      "step": 2204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.719408157824582e-06,
+      "loss": 0.1166,
+      "step": 2205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.719274860037325e-06,
+      "loss": 0.1787,
+      "step": 2206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.719141562250067e-06,
+      "loss": 0.1141,
+      "step": 2207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71900826446281e-06,
+      "loss": 0.1653,
+      "step": 2208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.718874966675555e-06,
+      "loss": 0.1087,
+      "step": 2209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.718741668888298e-06,
+      "loss": 0.1391,
+      "step": 2210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71860837110104e-06,
+      "loss": 0.161,
+      "step": 2211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.718475073313783e-06,
+      "loss": 0.1224,
+      "step": 2212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.718341775526528e-06,
+      "loss": 0.1015,
+      "step": 2213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71820847773927e-06,
+      "loss": 0.1009,
+      "step": 2214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.718075179952013e-06,
+      "loss": 0.0886,
+      "step": 2215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.717941882164756e-06,
+      "loss": 0.1379,
+      "step": 2216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.7178085843775e-06,
+      "loss": 0.1317,
+      "step": 2217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.717675286590244e-06,
+      "loss": 0.1152,
+      "step": 2218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.717541988802986e-06,
+      "loss": 0.1021,
+      "step": 2219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71740869101573e-06,
+      "loss": 0.1932,
+      "step": 2220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.717275393228474e-06,
+      "loss": 0.1122,
+      "step": 2221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.717142095441217e-06,
+      "loss": 0.1156,
+      "step": 2222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71700879765396e-06,
+      "loss": 0.1641,
+      "step": 2223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.716875499866702e-06,
+      "loss": 0.0979,
+      "step": 2224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.716742202079447e-06,
+      "loss": 0.1247,
+      "step": 2225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71660890429219e-06,
+      "loss": 0.1125,
+      "step": 2226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.716475606504932e-06,
+      "loss": 0.1129,
+      "step": 2227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.716342308717675e-06,
+      "loss": 0.1222,
+      "step": 2228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71620901093042e-06,
+      "loss": 0.1329,
+      "step": 2229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.716075713143162e-06,
+      "loss": 0.1195,
+      "step": 2230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.715942415355905e-06,
+      "loss": 0.102,
+      "step": 2231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.715809117568648e-06,
+      "loss": 0.1405,
+      "step": 2232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.715675819781393e-06,
+      "loss": 0.178,
+      "step": 2233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.715542521994135e-06,
+      "loss": 0.1061,
+      "step": 2234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.715409224206878e-06,
+      "loss": 0.1514,
+      "step": 2235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.715275926419621e-06,
+      "loss": 0.11,
+      "step": 2236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.715142628632366e-06,
+      "loss": 0.1555,
+      "step": 2237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.715009330845108e-06,
+      "loss": 0.1583,
+      "step": 2238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.714876033057851e-06,
+      "loss": 0.1616,
+      "step": 2239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.714742735270594e-06,
+      "loss": 0.1238,
+      "step": 2240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.714609437483339e-06,
+      "loss": 0.1404,
+      "step": 2241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.714476139696081e-06,
+      "loss": 0.1305,
+      "step": 2242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.714342841908824e-06,
+      "loss": 0.1051,
+      "step": 2243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.714209544121569e-06,
+      "loss": 0.0934,
+      "step": 2244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.714076246334312e-06,
+      "loss": 0.1325,
+      "step": 2245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.713942948547054e-06,
+      "loss": 0.1249,
+      "step": 2246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.713809650759797e-06,
+      "loss": 0.1253,
+      "step": 2247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.713676352972542e-06,
+      "loss": 0.1091,
+      "step": 2248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.713543055185284e-06,
+      "loss": 0.1365,
+      "step": 2249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.713409757398027e-06,
+      "loss": 0.1682,
+      "step": 2250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71327645961077e-06,
+      "loss": 0.1163,
+      "step": 2251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.713143161823515e-06,
+      "loss": 0.1492,
+      "step": 2252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.713009864036257e-06,
+      "loss": 0.1192,
+      "step": 2253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.712876566249e-06,
+      "loss": 0.127,
+      "step": 2254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.712743268461745e-06,
+      "loss": 0.1173,
+      "step": 2255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.712609970674488e-06,
+      "loss": 0.1577,
+      "step": 2256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71247667288723e-06,
+      "loss": 0.1141,
+      "step": 2257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.712343375099973e-06,
+      "loss": 0.122,
+      "step": 2258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.712210077312718e-06,
+      "loss": 0.1325,
+      "step": 2259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71207677952546e-06,
+      "loss": 0.265,
+      "step": 2260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.711943481738203e-06,
+      "loss": 0.1778,
+      "step": 2261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.711810183950948e-06,
+      "loss": 0.1511,
+      "step": 2262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71167688616369e-06,
+      "loss": 0.1194,
+      "step": 2263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.711543588376434e-06,
+      "loss": 0.1033,
+      "step": 2264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.711410290589176e-06,
+      "loss": 0.1107,
+      "step": 2265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.711276992801921e-06,
+      "loss": 0.1545,
+      "step": 2266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.711143695014664e-06,
+      "loss": 0.1084,
+      "step": 2267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.711010397227407e-06,
+      "loss": 0.1276,
+      "step": 2268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.710877099440151e-06,
+      "loss": 0.1136,
+      "step": 2269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.710743801652894e-06,
+      "loss": 0.1053,
+      "step": 2270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.710610503865637e-06,
+      "loss": 0.119,
+      "step": 2271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71047720607838e-06,
+      "loss": 0.1589,
+      "step": 2272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.710343908291124e-06,
+      "loss": 0.1136,
+      "step": 2273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.710210610503867e-06,
+      "loss": 0.1477,
+      "step": 2274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71007731271661e-06,
+      "loss": 0.0959,
+      "step": 2275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.709944014929354e-06,
+      "loss": 0.1073,
+      "step": 2276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.709810717142097e-06,
+      "loss": 0.1137,
+      "step": 2277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.70967741935484e-06,
+      "loss": 0.117,
+      "step": 2278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.709544121567583e-06,
+      "loss": 0.1249,
+      "step": 2279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.709410823780327e-06,
+      "loss": 0.1814,
+      "step": 2280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.70927752599307e-06,
+      "loss": 0.1244,
+      "step": 2281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.709144228205813e-06,
+      "loss": 0.1344,
+      "step": 2282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.709010930418556e-06,
+      "loss": 0.1335,
+      "step": 2283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.7088776326313e-06,
+      "loss": 0.1338,
+      "step": 2284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.708744334844043e-06,
+      "loss": 0.0972,
+      "step": 2285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.708611037056786e-06,
+      "loss": 0.1444,
+      "step": 2286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.708477739269529e-06,
+      "loss": 0.1246,
+      "step": 2287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.708344441482273e-06,
+      "loss": 0.1062,
+      "step": 2288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.708211143695016e-06,
+      "loss": 0.1209,
+      "step": 2289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.708077845907759e-06,
+      "loss": 0.1677,
+      "step": 2290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.707944548120502e-06,
+      "loss": 0.127,
+      "step": 2291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.707811250333246e-06,
+      "loss": 0.1708,
+      "step": 2292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.707677952545989e-06,
+      "loss": 0.1468,
+      "step": 2293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.707544654758732e-06,
+      "loss": 0.1246,
+      "step": 2294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.707411356971474e-06,
+      "loss": 0.0963,
+      "step": 2295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.707278059184219e-06,
+      "loss": 0.1028,
+      "step": 2296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.707144761396962e-06,
+      "loss": 0.1691,
+      "step": 2297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.707011463609705e-06,
+      "loss": 0.1136,
+      "step": 2298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.70687816582245e-06,
+      "loss": 0.1161,
+      "step": 2299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.706744868035192e-06,
+      "loss": 0.1248,
+      "step": 2300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.706611570247935e-06,
+      "loss": 0.149,
+      "step": 2301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.706478272460678e-06,
+      "loss": 0.1277,
+      "step": 2302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.706344974673422e-06,
+      "loss": 0.1427,
+      "step": 2303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.706211676886165e-06,
+      "loss": 0.1808,
+      "step": 2304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.706078379098908e-06,
+      "loss": 0.1468,
+      "step": 2305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.70594508131165e-06,
+      "loss": 0.1415,
+      "step": 2306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.705811783524395e-06,
+      "loss": 0.1225,
+      "step": 2307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.705678485737138e-06,
+      "loss": 0.1066,
+      "step": 2308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.70554518794988e-06,
+      "loss": 0.0824,
+      "step": 2309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.705411890162624e-06,
+      "loss": 0.1344,
+      "step": 2310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.705278592375368e-06,
+      "loss": 0.1665,
+      "step": 2311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.705145294588111e-06,
+      "loss": 0.1205,
+      "step": 2312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.705011996800854e-06,
+      "loss": 0.1141,
+      "step": 2313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.704878699013597e-06,
+      "loss": 0.1145,
+      "step": 2314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.704745401226341e-06,
+      "loss": 0.1317,
+      "step": 2315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.704612103439084e-06,
+      "loss": 0.1173,
+      "step": 2316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.704478805651827e-06,
+      "loss": 0.1274,
+      "step": 2317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.70434550786457e-06,
+      "loss": 0.091,
+      "step": 2318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.704212210077314e-06,
+      "loss": 0.1195,
+      "step": 2319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.704078912290057e-06,
+      "loss": 0.1015,
+      "step": 2320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.7039456145028e-06,
+      "loss": 0.1615,
+      "step": 2321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.703812316715542e-06,
+      "loss": 0.1357,
+      "step": 2322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.703679018928287e-06,
+      "loss": 0.121,
+      "step": 2323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.70354572114103e-06,
+      "loss": 0.1574,
+      "step": 2324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.703412423353773e-06,
+      "loss": 0.1341,
+      "step": 2325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.703279125566515e-06,
+      "loss": 0.1087,
+      "step": 2326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.70314582777926e-06,
+      "loss": 0.1087,
+      "step": 2327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.703012529992003e-06,
+      "loss": 0.1257,
+      "step": 2328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.702879232204746e-06,
+      "loss": 0.1246,
+      "step": 2329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.702745934417488e-06,
+      "loss": 0.124,
+      "step": 2330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.702612636630233e-06,
+      "loss": 0.1435,
+      "step": 2331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.702479338842976e-06,
+      "loss": 0.1345,
+      "step": 2332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.702346041055719e-06,
+      "loss": 0.1726,
+      "step": 2333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.702212743268461e-06,
+      "loss": 0.0985,
+      "step": 2334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.702079445481206e-06,
+      "loss": 0.1115,
+      "step": 2335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.701946147693949e-06,
+      "loss": 0.1226,
+      "step": 2336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.701812849906692e-06,
+      "loss": 0.1175,
+      "step": 2337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.701679552119434e-06,
+      "loss": 0.1271,
+      "step": 2338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.701546254332179e-06,
+      "loss": 0.1261,
+      "step": 2339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.701412956544922e-06,
+      "loss": 0.1134,
+      "step": 2340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.701279658757664e-06,
+      "loss": 0.1401,
+      "step": 2341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.701146360970409e-06,
+      "loss": 0.1013,
+      "step": 2342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.701013063183152e-06,
+      "loss": 0.1345,
+      "step": 2343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.700879765395895e-06,
+      "loss": 0.1216,
+      "step": 2344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.700746467608637e-06,
+      "loss": 0.1348,
+      "step": 2345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.700613169821382e-06,
+      "loss": 0.1103,
+      "step": 2346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.700479872034125e-06,
+      "loss": 0.1227,
+      "step": 2347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.700346574246868e-06,
+      "loss": 0.132,
+      "step": 2348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.700213276459612e-06,
+      "loss": 0.1242,
+      "step": 2349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.700079978672355e-06,
+      "loss": 0.0911,
+      "step": 2350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.699946680885098e-06,
+      "loss": 0.1149,
+      "step": 2351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.69981338309784e-06,
+      "loss": 0.0975,
+      "step": 2352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.699680085310585e-06,
+      "loss": 0.1285,
+      "step": 2353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.699546787523328e-06,
+      "loss": 0.1599,
+      "step": 2354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.69941348973607e-06,
+      "loss": 0.1588,
+      "step": 2355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.699280191948815e-06,
+      "loss": 0.1119,
+      "step": 2356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.699146894161558e-06,
+      "loss": 0.1386,
+      "step": 2357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.699013596374301e-06,
+      "loss": 0.1824,
+      "step": 2358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.698880298587044e-06,
+      "loss": 0.1029,
+      "step": 2359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.698747000799788e-06,
+      "loss": 0.1512,
+      "step": 2360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.698613703012531e-06,
+      "loss": 0.1156,
+      "step": 2361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.698480405225274e-06,
+      "loss": 0.1169,
+      "step": 2362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.698347107438018e-06,
+      "loss": 0.1595,
+      "step": 2363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.698213809650761e-06,
+      "loss": 0.0976,
+      "step": 2364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.698080511863504e-06,
+      "loss": 0.1426,
+      "step": 2365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.697947214076247e-06,
+      "loss": 0.0931,
+      "step": 2366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.697813916288991e-06,
+      "loss": 0.1368,
+      "step": 2367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.697680618501734e-06,
+      "loss": 0.1148,
+      "step": 2368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.697547320714477e-06,
+      "loss": 0.1021,
+      "step": 2369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.697414022927221e-06,
+      "loss": 0.1649,
+      "step": 2370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.697280725139964e-06,
+      "loss": 0.1333,
+      "step": 2371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.697147427352707e-06,
+      "loss": 0.1124,
+      "step": 2372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.69701412956545e-06,
+      "loss": 0.1162,
+      "step": 2373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.696880831778194e-06,
+      "loss": 0.1002,
+      "step": 2374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.696747533990937e-06,
+      "loss": 0.1342,
+      "step": 2375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.69661423620368e-06,
+      "loss": 0.1111,
+      "step": 2376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.696480938416423e-06,
+      "loss": 0.1215,
+      "step": 2377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.696347640629167e-06,
+      "loss": 0.1777,
+      "step": 2378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.69621434284191e-06,
+      "loss": 0.1532,
+      "step": 2379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.696081045054653e-06,
+      "loss": 0.1615,
+      "step": 2380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.695947747267396e-06,
+      "loss": 0.0897,
+      "step": 2381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.69581444948014e-06,
+      "loss": 0.1297,
+      "step": 2382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.695681151692883e-06,
+      "loss": 0.1441,
+      "step": 2383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.695547853905626e-06,
+      "loss": 0.1142,
+      "step": 2384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.695414556118369e-06,
+      "loss": 0.0901,
+      "step": 2385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.695281258331113e-06,
+      "loss": 0.1396,
+      "step": 2386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.695147960543856e-06,
+      "loss": 0.1316,
+      "step": 2387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.695014662756599e-06,
+      "loss": 0.1302,
+      "step": 2388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.694881364969342e-06,
+      "loss": 0.1772,
+      "step": 2389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.694748067182086e-06,
+      "loss": 0.1004,
+      "step": 2390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.69461476939483e-06,
+      "loss": 0.1022,
+      "step": 2391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.694481471607572e-06,
+      "loss": 0.12,
+      "step": 2392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.694348173820315e-06,
+      "loss": 0.1222,
+      "step": 2393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.69421487603306e-06,
+      "loss": 0.1092,
+      "step": 2394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.694081578245802e-06,
+      "loss": 0.117,
+      "step": 2395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.693948280458545e-06,
+      "loss": 0.0963,
+      "step": 2396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.693814982671288e-06,
+      "loss": 0.1107,
+      "step": 2397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.693681684884032e-06,
+      "loss": 0.1583,
+      "step": 2398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.693548387096775e-06,
+      "loss": 0.1098,
+      "step": 2399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.693415089309518e-06,
+      "loss": 0.1299,
+      "step": 2400
+    }
+  ],
+  "max_steps": 75120,
+  "num_train_epochs": 3,
+  "total_flos": 3.072579066431668e+18,
+  "trial_name": null,
+  "trial_params": null
+}