diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,48021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.4170576582212491,
+  "eval_steps": 500,
+  "global_step": 8000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7361111111111112e-07,
+      "loss": 4.0375,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4722222222222224e-07,
+      "loss": 45045.7109,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.208333333333334e-07,
+      "loss": 72707.0703,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.944444444444445e-07,
+      "loss": 1.6342,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.680555555555556e-07,
+      "loss": 106471.4453,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0416666666666667e-06,
+      "loss": 92079.8516,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2152777777777778e-06,
+      "loss": 101468.4844,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.388888888888889e-06,
+      "loss": 113189.2188,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5625e-06,
+      "loss": 173649.7031,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7361111111111112e-06,
+      "loss": 1.7893,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9097222222222225e-06,
+      "loss": 85932.2344,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 130685.6953,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2569444444444443e-06,
+      "loss": 103056.1016,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4305555555555557e-06,
+      "loss": 4.0167,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.604166666666667e-06,
+      "loss": 1.8903,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.777777777777778e-06,
+      "loss": 1.7694,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.951388888888889e-06,
+      "loss": 72446.9688,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.125e-06,
+      "loss": 75883.6875,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2986111111111115e-06,
+      "loss": 73331.9375,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4722222222222224e-06,
+      "loss": 1.7427,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6458333333333337e-06,
+      "loss": 1.6011,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.819444444444445e-06,
+      "loss": 71752.0781,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.993055555555555e-06,
+      "loss": 41648.8359,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.556,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.340277777777778e-06,
+      "loss": 42269.3281,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.513888888888889e-06,
+      "loss": 37786.0547,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 40539.1758,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.861111111111111e-06,
+      "loss": 30536.9082,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.034722222222222e-06,
+      "loss": 32534.9062,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.208333333333334e-06,
+      "loss": 25173.6973,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.381944444444445e-06,
+      "loss": 1.6537,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 1.5276,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.729166666666667e-06,
+      "loss": 15525.2051,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.902777777777778e-06,
+      "loss": 1.5632,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.076388888888889e-06,
+      "loss": 1.5464,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.25e-06,
+      "loss": 3.8233,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.423611111111111e-06,
+      "loss": 10328.3701,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.597222222222223e-06,
+      "loss": 1.6268,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.770833333333333e-06,
+      "loss": 10153.1035,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 9753.1309,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.118055555555556e-06,
+      "loss": 7248.625,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2916666666666674e-06,
+      "loss": 1.5713,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.465277777777778e-06,
+      "loss": 4400.8989,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.63888888888889e-06,
+      "loss": 3937.2842,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8125e-06,
+      "loss": 2776.5886,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.98611111111111e-06,
+      "loss": 1.5833,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.159722222222222e-06,
+      "loss": 1.5782,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.4317,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.506944444444445e-06,
+      "loss": 1917.4503,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.680555555555556e-06,
+      "loss": 2917.7825,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.854166666666667e-06,
+      "loss": 1664.4835,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.027777777777777e-06,
+      "loss": 1653.9132,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.201388888888889e-06,
+      "loss": 1678.8979,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 1292.6934,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.548611111111111e-06,
+      "loss": 1.399,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.722222222222223e-06,
+      "loss": 1075.7371,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.895833333333333e-06,
+      "loss": 1356.3514,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0069444444444445e-05,
+      "loss": 902.5632,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0243055555555556e-05,
+      "loss": 614.5408,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0416666666666668e-05,
+      "loss": 1047.0327,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0590277777777778e-05,
+      "loss": 412.5146,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.076388888888889e-05,
+      "loss": 1.298,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.09375e-05,
+      "loss": 1.2551,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 677.3477,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1284722222222223e-05,
+      "loss": 352.3045,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1458333333333333e-05,
+      "loss": 1.1908,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1631944444444445e-05,
+      "loss": 1.2051,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1805555555555555e-05,
+      "loss": 415.7868,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1979166666666667e-05,
+      "loss": 1.1744,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2152777777777779e-05,
+      "loss": 2.4628,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.232638888888889e-05,
+      "loss": 353.4174,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-05,
+      "loss": 376.1311,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.267361111111111e-05,
+      "loss": 360.8073,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2847222222222222e-05,
+      "loss": 1.1725,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3020833333333334e-05,
+      "loss": 356.3974,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3194444444444446e-05,
+      "loss": 2.0606,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3368055555555556e-05,
+      "loss": 1.1243,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3541666666666666e-05,
+      "loss": 342.5744,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.371527777777778e-05,
+      "loss": 340.6937,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 330.9445,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4062500000000001e-05,
+      "loss": 320.1059,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4236111111111111e-05,
+      "loss": 1.6597,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4409722222222221e-05,
+      "loss": 548.0641,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4583333333333335e-05,
+      "loss": 325.6071,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4756944444444445e-05,
+      "loss": 287.5581,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4930555555555557e-05,
+      "loss": 278.9295,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5104166666666667e-05,
+      "loss": 280.5663,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.527777777777778e-05,
+      "loss": 273.3075,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.545138888888889e-05,
+      "loss": 1.2048,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5625e-05,
+      "loss": 1.1207,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.579861111111111e-05,
+      "loss": 252.4243,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.597222222222222e-05,
+      "loss": 1.1414,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6145833333333334e-05,
+      "loss": 240.6253,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6319444444444444e-05,
+      "loss": 1.3816,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6493055555555557e-05,
+      "loss": 1.1271,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 350.9878,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.684027777777778e-05,
+      "loss": 231.1811,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.701388888888889e-05,
+      "loss": 1.145,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.71875e-05,
+      "loss": 218.4799,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.736111111111111e-05,
+      "loss": 219.3234,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.753472222222222e-05,
+      "loss": 207.0664,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7708333333333335e-05,
+      "loss": 204.9095,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7881944444444445e-05,
+      "loss": 204.8115,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8055555555555555e-05,
+      "loss": 191.5279,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8229166666666668e-05,
+      "loss": 1.1116,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8402777777777778e-05,
+      "loss": 1.2571,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.857638888888889e-05,
+      "loss": 199.0737,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 192.5411,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8923611111111112e-05,
+      "loss": 196.683,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9097222222222222e-05,
+      "loss": 186.5255,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9270833333333335e-05,
+      "loss": 174.059,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 174.7249,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9618055555555555e-05,
+      "loss": 1.1894,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9791666666666665e-05,
+      "loss": 245.4651,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.996527777777778e-05,
+      "loss": 1.1187,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.013888888888889e-05,
+      "loss": 158.8681,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0312500000000002e-05,
+      "loss": 1.0781,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0486111111111113e-05,
+      "loss": 169.6486,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0659722222222223e-05,
+      "loss": 1.0687,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 164.3003,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1006944444444446e-05,
+      "loss": 1.0634,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1180555555555556e-05,
+      "loss": 1.0606,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1354166666666666e-05,
+      "loss": 159.8331,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.152777777777778e-05,
+      "loss": 223.3789,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.170138888888889e-05,
+      "loss": 155.2308,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1875e-05,
+      "loss": 154.2659,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2048611111111113e-05,
+      "loss": 1.2142,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 149.5258,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2395833333333337e-05,
+      "loss": 1.0332,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2569444444444447e-05,
+      "loss": 1.1564,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2743055555555557e-05,
+      "loss": 166.1467,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2916666666666667e-05,
+      "loss": 144.2635,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.309027777777778e-05,
+      "loss": 1.0655,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.326388888888889e-05,
+      "loss": 1.0194,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.34375e-05,
+      "loss": 1.0817,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.361111111111111e-05,
+      "loss": 171.8577,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.378472222222222e-05,
+      "loss": 1.0526,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3958333333333334e-05,
+      "loss": 156.8806,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4131944444444448e-05,
+      "loss": 152.3786,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4305555555555558e-05,
+      "loss": 135.47,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4479166666666668e-05,
+      "loss": 143.0895,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.465277777777778e-05,
+      "loss": 165.3984,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.482638888888889e-05,
+      "loss": 1.0814,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 1.1415,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.517361111111111e-05,
+      "loss": 219.8956,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.534722222222222e-05,
+      "loss": 152.3145,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.552083333333333e-05,
+      "loss": 132.4326,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5694444444444445e-05,
+      "loss": 134.9339,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5868055555555558e-05,
+      "loss": 1.0431,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.604166666666667e-05,
+      "loss": 1.0467,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6215277777777782e-05,
+      "loss": 1.1019,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6388888888888892e-05,
+      "loss": 1.0428,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6562500000000002e-05,
+      "loss": 1.0757,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6736111111111112e-05,
+      "loss": 146.6331,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6909722222222222e-05,
+      "loss": 143.658,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7083333333333332e-05,
+      "loss": 194.0388,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7256944444444442e-05,
+      "loss": 1.0195,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.743055555555556e-05,
+      "loss": 152.2088,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.760416666666667e-05,
+      "loss": 1.0545,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 133.995,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.795138888888889e-05,
+      "loss": 126.8912,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 124.8623,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8298611111111113e-05,
+      "loss": 1.0626,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8472222222222223e-05,
+      "loss": 121.1801,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8645833333333333e-05,
+      "loss": 0.9878,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8819444444444443e-05,
+      "loss": 1.0512,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.899305555555556e-05,
+      "loss": 123.4393,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 118.3312,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.934027777777778e-05,
+      "loss": 1.048,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.951388888888889e-05,
+      "loss": 115.7459,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.96875e-05,
+      "loss": 1.07,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9861111111111113e-05,
+      "loss": 0.9855,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0034722222222223e-05,
+      "loss": 113.8891,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0208333333333334e-05,
+      "loss": 112.7567,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0381944444444444e-05,
+      "loss": 0.9904,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.055555555555556e-05,
+      "loss": 108.2369,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.072916666666667e-05,
+      "loss": 1.04,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.090277777777778e-05,
+      "loss": 1.0336,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.107638888888889e-05,
+      "loss": 1.0084,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.125e-05,
+      "loss": 0.9788,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.142361111111111e-05,
+      "loss": 0.9578,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.159722222222222e-05,
+      "loss": 829.1132,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.177083333333333e-05,
+      "loss": 498.2139,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.194444444444444e-05,
+      "loss": 0.8717,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.211805555555556e-05,
+      "loss": 0.8352,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.229166666666667e-05,
+      "loss": 108.5255,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.246527777777778e-05,
+      "loss": 0.8287,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.263888888888889e-05,
+      "loss": 110.0258,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2812500000000005e-05,
+      "loss": 0.9368,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2986111111111115e-05,
+      "loss": 0.9581,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3159722222222225e-05,
+      "loss": 113.4381,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 114.5087,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3506944444444445e-05,
+      "loss": 104.5135,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.368055555555556e-05,
+      "loss": 103.9,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.385416666666667e-05,
+      "loss": 120.7145,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.402777777777778e-05,
+      "loss": 0.9635,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.420138888888889e-05,
+      "loss": 0.6744,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4375e-05,
+      "loss": 111.9032,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.454861111111111e-05,
+      "loss": 0.6367,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.472222222222222e-05,
+      "loss": 102.471,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.489583333333333e-05,
+      "loss": 104.9312,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.506944444444444e-05,
+      "loss": 100.7445,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.524305555555556e-05,
+      "loss": 0.8939,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.541666666666667e-05,
+      "loss": 109.6047,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.559027777777778e-05,
+      "loss": 0.9418,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.576388888888889e-05,
+      "loss": 0.8337,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.59375e-05,
+      "loss": 0.5599,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.611111111111111e-05,
+      "loss": 0.8871,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.628472222222222e-05,
+      "loss": 115.9292,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6458333333333336e-05,
+      "loss": 102.013,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6631944444444446e-05,
+      "loss": 104.4466,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6805555555555556e-05,
+      "loss": 100.1761,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.697916666666667e-05,
+      "loss": 97.1255,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.715277777777778e-05,
+      "loss": 99.1342,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7326388888888893e-05,
+      "loss": 99.4381,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 98.9235,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7673611111111114e-05,
+      "loss": 88.34,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7847222222222224e-05,
+      "loss": 96.1837,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8020833333333334e-05,
+      "loss": 95.2069,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8194444444444444e-05,
+      "loss": 0.8459,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.836805555555556e-05,
+      "loss": 92.3914,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.854166666666667e-05,
+      "loss": 90.8146,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.871527777777778e-05,
+      "loss": 0.84,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 90.5142,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.90625e-05,
+      "loss": 0.8846,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.923611111111111e-05,
+      "loss": 0.8228,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.940972222222222e-05,
+      "loss": 85.5237,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.958333333333333e-05,
+      "loss": 0.3496,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.975694444444444e-05,
+      "loss": 0.8801,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.993055555555556e-05,
+      "loss": 89.7818,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.010416666666667e-05,
+      "loss": 0.8591,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.027777777777778e-05,
+      "loss": 0.2925,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.045138888888889e-05,
+      "loss": 0.872,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 90.4654,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0798611111111115e-05,
+      "loss": 0.2684,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0972222222222225e-05,
+      "loss": 87.6459,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1145833333333335e-05,
+      "loss": 86.2269,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1319444444444445e-05,
+      "loss": 0.8483,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.149305555555556e-05,
+      "loss": 0.8129,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.8693,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.184027777777778e-05,
+      "loss": 0.824,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.201388888888889e-05,
+      "loss": 92.64,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.21875e-05,
+      "loss": 0.8449,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.236111111111111e-05,
+      "loss": 88.0765,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.253472222222222e-05,
+      "loss": 0.8533,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.270833333333333e-05,
+      "loss": 79.9158,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.288194444444444e-05,
+      "loss": 84.5935,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.305555555555556e-05,
+      "loss": 87.8726,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.322916666666667e-05,
+      "loss": 84.6476,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.340277777777778e-05,
+      "loss": 89.1216,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.357638888888889e-05,
+      "loss": 0.7986,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.375e-05,
+      "loss": 85.2967,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.392361111111111e-05,
+      "loss": 0.8287,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4097222222222226e-05,
+      "loss": 88.4917,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4270833333333337e-05,
+      "loss": 90.7842,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.795,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4618055555555563e-05,
+      "loss": 82.5256,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4791666666666673e-05,
+      "loss": 84.6522,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4965277777777784e-05,
+      "loss": 0.139,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5138888888888894e-05,
+      "loss": 0.8653,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5312500000000004e-05,
+      "loss": 91.7546,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5486111111111114e-05,
+      "loss": 87.8058,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5659722222222224e-05,
+      "loss": 79.6235,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 88.9975,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6006944444444444e-05,
+      "loss": 91.5088,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.618055555555556e-05,
+      "loss": 88.2395,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.635416666666667e-05,
+      "loss": 0.1633,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.652777777777778e-05,
+      "loss": 0.834,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.670138888888889e-05,
+      "loss": 0.803,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6875e-05,
+      "loss": 85.1043,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.704861111111111e-05,
+      "loss": 84.333,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.722222222222222e-05,
+      "loss": 82.2577,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.739583333333333e-05,
+      "loss": 0.9083,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.756944444444444e-05,
+      "loss": 0.7708,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.774305555555556e-05,
+      "loss": 0.1001,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.791666666666667e-05,
+      "loss": 0.8642,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.809027777777778e-05,
+      "loss": 0.8301,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8263888888888895e-05,
+      "loss": 424.7609,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8437500000000005e-05,
+      "loss": 0.0993,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8611111111111115e-05,
+      "loss": 346.9087,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8784722222222225e-05,
+      "loss": 214.7958,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8958333333333335e-05,
+      "loss": 0.8303,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9131944444444445e-05,
+      "loss": 91.619,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.930555555555556e-05,
+      "loss": 0.8198,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.947916666666667e-05,
+      "loss": 85.7449,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965277777777778e-05,
+      "loss": 89.5292,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982638888888889e-05,
+      "loss": 90.5244,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 87.1136,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.017361111111112e-05,
+      "loss": 81.6383,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.034722222222222e-05,
+      "loss": 83.3347,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.052083333333334e-05,
+      "loss": 89.9009,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.069444444444444e-05,
+      "loss": 94.446,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.086805555555556e-05,
+      "loss": 87.0233,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.104166666666666e-05,
+      "loss": 0.1382,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.121527777777778e-05,
+      "loss": 0.886,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.138888888888889e-05,
+      "loss": 0.7995,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.15625e-05,
+      "loss": 84.1341,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1736111111111117e-05,
+      "loss": 0.1287,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.190972222222223e-05,
+      "loss": 0.1016,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.208333333333334e-05,
+      "loss": 0.8532,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.225694444444445e-05,
+      "loss": 0.1045,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2430555555555564e-05,
+      "loss": 87.6301,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.260416666666667e-05,
+      "loss": 0.872,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2777777777777784e-05,
+      "loss": 0.0895,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.295138888888889e-05,
+      "loss": 86.9678,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3125000000000004e-05,
+      "loss": 0.84,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.329861111111112e-05,
+      "loss": 0.8294,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3472222222222224e-05,
+      "loss": 82.6633,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.364583333333334e-05,
+      "loss": 87.4287,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3819444444444444e-05,
+      "loss": 80.2371,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.399305555555556e-05,
+      "loss": 83.4334,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4166666666666664e-05,
+      "loss": 0.8294,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.434027777777778e-05,
+      "loss": 0.771,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4513888888888884e-05,
+      "loss": 88.0706,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.46875e-05,
+      "loss": 86.6529,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.486111111111112e-05,
+      "loss": 0.8047,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.503472222222222e-05,
+      "loss": 0.7572,
+      "step": 317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.520833333333334e-05,
+      "loss": 78.7965,
+      "step": 318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.538194444444444e-05,
+      "loss": 80.8088,
+      "step": 319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.555555555555556e-05,
+      "loss": 83.6609,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.572916666666667e-05,
+      "loss": 80.9232,
+      "step": 321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.590277777777778e-05,
+      "loss": 79.7782,
+      "step": 322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.607638888888889e-05,
+      "loss": 0.0885,
+      "step": 323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 80.4718,
+      "step": 324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.642361111111112e-05,
+      "loss": 76.0821,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6597222222222225e-05,
+      "loss": 75.5221,
+      "step": 326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.677083333333334e-05,
+      "loss": 81.2214,
+      "step": 327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6944444444444445e-05,
+      "loss": 76.6528,
+      "step": 328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.711805555555556e-05,
+      "loss": 0.8661,
+      "step": 329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7291666666666666e-05,
+      "loss": 78.8847,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.746527777777778e-05,
+      "loss": 73.266,
+      "step": 331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7638888888888886e-05,
+      "loss": 0.2779,
+      "step": 332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.78125e-05,
+      "loss": 83.2001,
+      "step": 333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.798611111111112e-05,
+      "loss": 82.5089,
+      "step": 334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.815972222222222e-05,
+      "loss": 79.2043,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.833333333333334e-05,
+      "loss": 0.8314,
+      "step": 336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.850694444444444e-05,
+      "loss": 0.7873,
+      "step": 337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.868055555555556e-05,
+      "loss": 0.8012,
+      "step": 338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.885416666666666e-05,
+      "loss": 0.7688,
+      "step": 339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.902777777777778e-05,
+      "loss": 94.1174,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.920138888888889e-05,
+      "loss": 0.1029,
+      "step": 341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9375e-05,
+      "loss": 0.8225,
+      "step": 342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.954861111111112e-05,
+      "loss": 93.9488,
+      "step": 343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.972222222222223e-05,
+      "loss": 83.5193,
+      "step": 344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.989583333333334e-05,
+      "loss": 79.893,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.006944444444445e-05,
+      "loss": 80.1831,
+      "step": 346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0243055555555564e-05,
+      "loss": 0.8101,
+      "step": 347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.041666666666667e-05,
+      "loss": 0.7586,
+      "step": 348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0590277777777784e-05,
+      "loss": 0.8469,
+      "step": 349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.076388888888889e-05,
+      "loss": 371.1703,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0937500000000004e-05,
+      "loss": 0.7736,
+      "step": 351
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.111111111111112e-05,
+      "loss": 509.3132,
+      "step": 352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.128472222222222e-05,
+      "loss": 0.1108,
+      "step": 353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.145833333333334e-05,
+      "loss": 295.3356,
+      "step": 354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.163194444444445e-05,
+      "loss": 173.535,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.180555555555556e-05,
+      "loss": 0.8089,
+      "step": 356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.197916666666667e-05,
+      "loss": 81.5101,
+      "step": 357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.215277777777778e-05,
+      "loss": 0.7823,
+      "step": 358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.232638888888889e-05,
+      "loss": 78.0809,
+      "step": 359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.25e-05,
+      "loss": 0.7905,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.267361111111112e-05,
+      "loss": 0.8139,
+      "step": 361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.284722222222222e-05,
+      "loss": 94.5368,
+      "step": 362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.302083333333335e-05,
+      "loss": 88.1279,
+      "step": 363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.319444444444444e-05,
+      "loss": 0.241,
+      "step": 364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.336805555555557e-05,
+      "loss": 85.0807,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.354166666666666e-05,
+      "loss": 74.9467,
+      "step": 366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.371527777777779e-05,
+      "loss": 75.2542,
+      "step": 367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.388888888888888e-05,
+      "loss": 79.8005,
+      "step": 368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.40625e-05,
+      "loss": 77.2998,
+      "step": 369
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.423611111111112e-05,
+      "loss": 74.6204,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.440972222222223e-05,
+      "loss": 69.684,
+      "step": 371
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.458333333333334e-05,
+      "loss": 0.3781,
+      "step": 372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.475694444444445e-05,
+      "loss": 77.9023,
+      "step": 373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.493055555555556e-05,
+      "loss": 0.8785,
+      "step": 374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.510416666666667e-05,
+      "loss": 78.7001,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.527777777777778e-05,
+      "loss": 73.0716,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.545138888888889e-05,
+      "loss": 0.7901,
+      "step": 377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.562500000000001e-05,
+      "loss": 76.7487,
+      "step": 378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.579861111111112e-05,
+      "loss": 76.2168,
+      "step": 379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.597222222222223e-05,
+      "loss": 74.1973,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.614583333333334e-05,
+      "loss": 0.8317,
+      "step": 381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.631944444444445e-05,
+      "loss": 81.9756,
+      "step": 382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.649305555555556e-05,
+      "loss": 79.0244,
+      "step": 383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 76.5657,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.684027777777778e-05,
+      "loss": 0.8296,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.701388888888889e-05,
+      "loss": 75.7908,
+      "step": 386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.71875e-05,
+      "loss": 0.8233,
+      "step": 387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.736111111111112e-05,
+      "loss": 77.1434,
+      "step": 388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.753472222222222e-05,
+      "loss": 75.7695,
+      "step": 389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.770833333333334e-05,
+      "loss": 0.1546,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.788194444444444e-05,
+      "loss": 78.4663,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.805555555555556e-05,
+      "loss": 0.8059,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.822916666666666e-05,
+      "loss": 0.7777,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.840277777777778e-05,
+      "loss": 0.8152,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.857638888888888e-05,
+      "loss": 75.2416,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.875e-05,
+      "loss": 0.8232,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.892361111111111e-05,
+      "loss": 73.0567,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.909722222222222e-05,
+      "loss": 0.8506,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.927083333333333e-05,
+      "loss": 0.7806,
+      "step": 399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.944444444444444e-05,
+      "loss": 81.0372,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.961805555555557e-05,
+      "loss": 82.8865,
+      "step": 401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.979166666666666e-05,
+      "loss": 0.8327,
+      "step": 402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.996527777777779e-05,
+      "loss": 0.8042,
+      "step": 403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.013888888888888e-05,
+      "loss": 0.7716,
+      "step": 404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.031250000000001e-05,
+      "loss": 75.5143,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.048611111111112e-05,
+      "loss": 0.82,
+      "step": 406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.065972222222223e-05,
+      "loss": 72.2855,
+      "step": 407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.083333333333334e-05,
+      "loss": 0.8092,
+      "step": 408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.100694444444445e-05,
+      "loss": 0.1449,
+      "step": 409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.118055555555556e-05,
+      "loss": 75.0282,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.135416666666667e-05,
+      "loss": 74.1737,
+      "step": 411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.152777777777778e-05,
+      "loss": 0.7866,
+      "step": 412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.170138888888889e-05,
+      "loss": 72.5652,
+      "step": 413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.1875e-05,
+      "loss": 0.1396,
+      "step": 414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.204861111111112e-05,
+      "loss": 0.7864,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.222222222222222e-05,
+      "loss": 66.7563,
+      "step": 416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.239583333333334e-05,
+      "loss": 0.7749,
+      "step": 417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.256944444444444e-05,
+      "loss": 0.8713,
+      "step": 418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.274305555555556e-05,
+      "loss": 0.8049,
+      "step": 419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.291666666666667e-05,
+      "loss": 74.2959,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.309027777777778e-05,
+      "loss": 77.6594,
+      "step": 421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.326388888888889e-05,
+      "loss": 69.7266,
+      "step": 422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.34375e-05,
+      "loss": 0.1528,
+      "step": 423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.361111111111111e-05,
+      "loss": 0.7891,
+      "step": 424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.378472222222222e-05,
+      "loss": 0.8241,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.395833333333335e-05,
+      "loss": 78.657,
+      "step": 426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.413194444444444e-05,
+      "loss": 76.347,
+      "step": 427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.430555555555557e-05,
+      "loss": 79.8964,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.447916666666666e-05,
+      "loss": 72.1168,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.465277777777779e-05,
+      "loss": 0.7988,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.482638888888888e-05,
+      "loss": 0.1304,
+      "step": 431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.804,
+      "step": 432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.517361111111112e-05,
+      "loss": 156.044,
+      "step": 433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.534722222222223e-05,
+      "loss": 0.7668,
+      "step": 434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.552083333333334e-05,
+      "loss": 183.1571,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.569444444444445e-05,
+      "loss": 0.7746,
+      "step": 436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.586805555555556e-05,
+      "loss": 0.8168,
+      "step": 437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.604166666666667e-05,
+      "loss": 104.3347,
+      "step": 438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.621527777777778e-05,
+      "loss": 82.4435,
+      "step": 439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.638888888888889e-05,
+      "loss": 0.8175,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.65625e-05,
+      "loss": 66.6976,
+      "step": 441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.673611111111112e-05,
+      "loss": 73.3679,
+      "step": 442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.690972222222223e-05,
+      "loss": 67.3016,
+      "step": 443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.708333333333334e-05,
+      "loss": 70.9439,
+      "step": 444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.725694444444445e-05,
+      "loss": 70.9399,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.743055555555556e-05,
+      "loss": 0.2126,
+      "step": 446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.760416666666667e-05,
+      "loss": 68.3897,
+      "step": 447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.777777777777778e-05,
+      "loss": 0.7945,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.795138888888889e-05,
+      "loss": 71.8494,
+      "step": 449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.8125e-05,
+      "loss": 68.7709,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.829861111111113e-05,
+      "loss": 0.112,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.847222222222222e-05,
+      "loss": 0.7834,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.864583333333335e-05,
+      "loss": 0.7642,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.881944444444444e-05,
+      "loss": 70.8258,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.899305555555557e-05,
+      "loss": 77.0166,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.916666666666666e-05,
+      "loss": 72.4607,
+      "step": 456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.934027777777779e-05,
+      "loss": 0.7572,
+      "step": 457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.951388888888888e-05,
+      "loss": 0.8196,
+      "step": 458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.96875e-05,
+      "loss": 75.3972,
+      "step": 459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.986111111111112e-05,
+      "loss": 79.973,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.003472222222223e-05,
+      "loss": 0.1105,
+      "step": 461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.020833333333334e-05,
+      "loss": 71.4909,
+      "step": 462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.038194444444445e-05,
+      "loss": 69.2889,
+      "step": 463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.055555555555556e-05,
+      "loss": 70.2024,
+      "step": 464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.072916666666667e-05,
+      "loss": 0.7527,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.090277777777778e-05,
+      "loss": 68.2457,
+      "step": 466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.107638888888889e-05,
+      "loss": 72.5095,
+      "step": 467
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 66.8765,
+      "step": 468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.142361111111112e-05,
+      "loss": 70.7129,
+      "step": 469
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.159722222222223e-05,
+      "loss": 72.0796,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.177083333333334e-05,
+      "loss": 67.5444,
+      "step": 471
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.194444444444445e-05,
+      "loss": 0.0945,
+      "step": 472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.211805555555556e-05,
+      "loss": 70.5687,
+      "step": 473
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.229166666666667e-05,
+      "loss": 0.8167,
+      "step": 474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.246527777777778e-05,
+      "loss": 69.2774,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.263888888888889e-05,
+      "loss": 0.7834,
+      "step": 476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.28125e-05,
+      "loss": 65.4707,
+      "step": 477
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.298611111111112e-05,
+      "loss": 0.8195,
+      "step": 478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.315972222222222e-05,
+      "loss": 0.7788,
+      "step": 479
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 66.5561,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.350694444444444e-05,
+      "loss": 0.7671,
+      "step": 481
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.368055555555556e-05,
+      "loss": 69.339,
+      "step": 482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.385416666666666e-05,
+      "loss": 68.9138,
+      "step": 483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.402777777777778e-05,
+      "loss": 66.1949,
+      "step": 484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.42013888888889e-05,
+      "loss": 64.7596,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.4375e-05,
+      "loss": 66.8169,
+      "step": 486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.454861111111111e-05,
+      "loss": 0.7098,
+      "step": 487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.472222222222222e-05,
+      "loss": 0.821,
+      "step": 488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.489583333333333e-05,
+      "loss": 0.7797,
+      "step": 489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.506944444444444e-05,
+      "loss": 0.1583,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.524305555555557e-05,
+      "loss": 68.017,
+      "step": 491
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.541666666666666e-05,
+      "loss": 0.8691,
+      "step": 492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.559027777777779e-05,
+      "loss": 0.0732,
+      "step": 493
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.576388888888888e-05,
+      "loss": 67.1419,
+      "step": 494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.593750000000001e-05,
+      "loss": 0.7903,
+      "step": 495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.611111111111112e-05,
+      "loss": 68.935,
+      "step": 496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.628472222222223e-05,
+      "loss": 67.3537,
+      "step": 497
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.645833333333334e-05,
+      "loss": 0.8027,
+      "step": 498
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.663194444444445e-05,
+      "loss": 0.8175,
+      "step": 499
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.680555555555556e-05,
+      "loss": 0.7366,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.697916666666667e-05,
+      "loss": 0.8575,
+      "step": 501
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.715277777777778e-05,
+      "loss": 133.2286,
+      "step": 502
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.732638888888889e-05,
+      "loss": 128.9279,
+      "step": 503
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.75e-05,
+      "loss": 89.1498,
+      "step": 504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.767361111111112e-05,
+      "loss": 77.6881,
+      "step": 505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.784722222222222e-05,
+      "loss": 70.5465,
+      "step": 506
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.802083333333334e-05,
+      "loss": 0.1065,
+      "step": 507
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.819444444444445e-05,
+      "loss": 0.8461,
+      "step": 508
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.836805555555556e-05,
+      "loss": 79.1885,
+      "step": 509
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.854166666666667e-05,
+      "loss": 0.8076,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.871527777777778e-05,
+      "loss": 0.0984,
+      "step": 511
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.888888888888889e-05,
+      "loss": 68.1952,
+      "step": 512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.90625e-05,
+      "loss": 67.4691,
+      "step": 513
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.923611111111113e-05,
+      "loss": 61.725,
+      "step": 514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.940972222222222e-05,
+      "loss": 63.6247,
+      "step": 515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.958333333333335e-05,
+      "loss": 68.9859,
+      "step": 516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.975694444444444e-05,
+      "loss": 65.5243,
+      "step": 517
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.993055555555557e-05,
+      "loss": 67.3343,
+      "step": 518
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.010416666666666e-05,
+      "loss": 0.8579,
+      "step": 519
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.027777777777779e-05,
+      "loss": 67.8559,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.045138888888888e-05,
+      "loss": 0.7973,
+      "step": 521
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.062500000000001e-05,
+      "loss": 0.8019,
+      "step": 522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.079861111111112e-05,
+      "loss": 0.8075,
+      "step": 523
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.097222222222223e-05,
+      "loss": 72.1142,
+      "step": 524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.114583333333334e-05,
+      "loss": 0.8081,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.131944444444445e-05,
+      "loss": 0.8235,
+      "step": 526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.149305555555556e-05,
+      "loss": 72.3422,
+      "step": 527
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.166666666666667e-05,
+      "loss": 0.1228,
+      "step": 528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.184027777777778e-05,
+      "loss": 0.7975,
+      "step": 529
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.201388888888889e-05,
+      "loss": 0.8574,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.21875e-05,
+      "loss": 0.8302,
+      "step": 531
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.236111111111112e-05,
+      "loss": 64.7728,
+      "step": 532
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.253472222222223e-05,
+      "loss": 61.6413,
+      "step": 533
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.270833333333334e-05,
+      "loss": 0.8061,
+      "step": 534
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.288194444444445e-05,
+      "loss": 63.6469,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.305555555555556e-05,
+      "loss": 0.8567,
+      "step": 536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.322916666666667e-05,
+      "loss": 68.242,
+      "step": 537
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.340277777777778e-05,
+      "loss": 64.1344,
+      "step": 538
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.357638888888889e-05,
+      "loss": 64.093,
+      "step": 539
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.375e-05,
+      "loss": 64.8912,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.392361111111113e-05,
+      "loss": 0.1235,
+      "step": 541
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.409722222222222e-05,
+      "loss": 61.9715,
+      "step": 542
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.427083333333335e-05,
+      "loss": 0.8124,
+      "step": 543
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.444444444444444e-05,
+      "loss": 0.0836,
+      "step": 544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.461805555555557e-05,
+      "loss": 0.83,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.479166666666666e-05,
+      "loss": 69.4473,
+      "step": 546
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.496527777777779e-05,
+      "loss": 63.4802,
+      "step": 547
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.513888888888888e-05,
+      "loss": 64.7705,
+      "step": 548
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.53125e-05,
+      "loss": 64.0141,
+      "step": 549
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.548611111111112e-05,
+      "loss": 69.2653,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.565972222222223e-05,
+      "loss": 0.1553,
+      "step": 551
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.583333333333334e-05,
+      "loss": 0.8134,
+      "step": 552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.600694444444445e-05,
+      "loss": 0.8198,
+      "step": 553
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.618055555555556e-05,
+      "loss": 64.0497,
+      "step": 554
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.635416666666667e-05,
+      "loss": 63.6452,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.652777777777779e-05,
+      "loss": 66.051,
+      "step": 556
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.670138888888889e-05,
+      "loss": 59.454,
+      "step": 557
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.687500000000001e-05,
+      "loss": 0.1001,
+      "step": 558
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.704861111111112e-05,
+      "loss": 0.7893,
+      "step": 559
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.722222222222223e-05,
+      "loss": 64.8456,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.739583333333334e-05,
+      "loss": 65.3228,
+      "step": 561
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.756944444444445e-05,
+      "loss": 63.5593,
+      "step": 562
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.774305555555556e-05,
+      "loss": 63.9309,
+      "step": 563
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.791666666666667e-05,
+      "loss": 63.7704,
+      "step": 564
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.809027777777778e-05,
+      "loss": 63.718,
+      "step": 565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.826388888888889e-05,
+      "loss": 63.3512,
+      "step": 566
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.84375e-05,
+      "loss": 0.2091,
+      "step": 567
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.861111111111112e-05,
+      "loss": 0.8203,
+      "step": 568
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.878472222222222e-05,
+      "loss": 0.7355,
+      "step": 569
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.895833333333334e-05,
+      "loss": 65.1789,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.913194444444444e-05,
+      "loss": 65.6934,
+      "step": 571
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.930555555555556e-05,
+      "loss": 65.0039,
+      "step": 572
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.947916666666666e-05,
+      "loss": 0.8107,
+      "step": 573
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.965277777777778e-05,
+      "loss": 0.799,
+      "step": 574
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98263888888889e-05,
+      "loss": 85.2869,
+      "step": 575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 0.7946,
+      "step": 576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999928725577e-05,
+      "loss": 0.7834,
+      "step": 577
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999714902304e-05,
+      "loss": 0.7392,
+      "step": 578
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999358530193e-05,
+      "loss": 0.7796,
+      "step": 579
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999998859609247e-05,
+      "loss": 116.9219,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999998218139488e-05,
+      "loss": 0.8326,
+      "step": 581
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999997434120929e-05,
+      "loss": 83.7997,
+      "step": 582
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999996507553595e-05,
+      "loss": 0.7958,
+      "step": 583
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999995438437511e-05,
+      "loss": 63.2921,
+      "step": 584
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999994226772706e-05,
+      "loss": 64.0244,
+      "step": 585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999992872559219e-05,
+      "loss": 67.3554,
+      "step": 586
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999991375797085e-05,
+      "loss": 0.7155,
+      "step": 587
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999989736486348e-05,
+      "loss": 59.2205,
+      "step": 588
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999987954627055e-05,
+      "loss": 0.741,
+      "step": 589
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999986030219255e-05,
+      "loss": 61.252,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999983963263004e-05,
+      "loss": 61.5061,
+      "step": 591
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999981753758363e-05,
+      "loss": 63.7858,
+      "step": 592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999979401705392e-05,
+      "loss": 58.9464,
+      "step": 593
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999976907104158e-05,
+      "loss": 63.7448,
+      "step": 594
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999974269954735e-05,
+      "loss": 0.8272,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999971490257196e-05,
+      "loss": 0.7868,
+      "step": 596
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999968568011619e-05,
+      "loss": 63.6121,
+      "step": 597
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999965503218091e-05,
+      "loss": 62.5862,
+      "step": 598
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999962295876697e-05,
+      "loss": 0.82,
+      "step": 599
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999958945987528e-05,
+      "loss": 0.8046,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99995545355068e-05,
+      "loss": 0.7931,
+      "step": 601
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999951818566253e-05,
+      "loss": 0.76,
+      "step": 602
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999948041034352e-05,
+      "loss": 60.8847,
+      "step": 603
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999944120955083e-05,
+      "loss": 67.497,
+      "step": 604
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999940058328557e-05,
+      "loss": 0.7926,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99993585315489e-05,
+      "loss": 60.1817,
+      "step": 606
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999931505434203e-05,
+      "loss": 0.7639,
+      "step": 607
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99992701516662e-05,
+      "loss": 66.3143,
+      "step": 608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999922382352269e-05,
+      "loss": 0.113,
+      "step": 609
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999917606991282e-05,
+      "loss": 0.0927,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999912689083794e-05,
+      "loss": 64.2098,
+      "step": 611
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999907628629946e-05,
+      "loss": 62.9245,
+      "step": 612
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999902425629884e-05,
+      "loss": 61.8366,
+      "step": 613
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999897080083752e-05,
+      "loss": 0.1066,
+      "step": 614
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999891591991707e-05,
+      "loss": 58.9196,
+      "step": 615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999885961353903e-05,
+      "loss": 0.7781,
+      "step": 616
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999880188170501e-05,
+      "loss": 0.8726,
+      "step": 617
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999874272441665e-05,
+      "loss": 0.0713,
+      "step": 618
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999868214167565e-05,
+      "loss": 68.5329,
+      "step": 619
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999862013348374e-05,
+      "loss": 63.6177,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999855669984266e-05,
+      "loss": 60.3072,
+      "step": 621
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999849184075425e-05,
+      "loss": 59.8687,
+      "step": 622
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999842555622035e-05,
+      "loss": 59.9891,
+      "step": 623
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999835784624283e-05,
+      "loss": 61.85,
+      "step": 624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999828871082364e-05,
+      "loss": 64.6047,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999821814996475e-05,
+      "loss": 0.0732,
+      "step": 626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999814616366816e-05,
+      "loss": 0.8366,
+      "step": 627
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999807275193593e-05,
+      "loss": 57.9161,
+      "step": 628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999799791477018e-05,
+      "loss": 0.8429,
+      "step": 629
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999792165217298e-05,
+      "loss": 60.0592,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999784396414657e-05,
+      "loss": 0.1257,
+      "step": 631
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999776485069313e-05,
+      "loss": 0.1027,
+      "step": 632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999768431181492e-05,
+      "loss": 0.8655,
+      "step": 633
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999760234751425e-05,
+      "loss": 59.5116,
+      "step": 634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999751895779342e-05,
+      "loss": 0.7929,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999743414265486e-05,
+      "loss": 0.8015,
+      "step": 636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999734790210094e-05,
+      "loss": 0.7392,
+      "step": 637
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999726023613414e-05,
+      "loss": 0.8281,
+      "step": 638
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999717114475696e-05,
+      "loss": 0.72,
+      "step": 639
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999708062797195e-05,
+      "loss": 0.8614,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999698868578166e-05,
+      "loss": 64.8355,
+      "step": 641
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999689531818874e-05,
+      "loss": 60.3918,
+      "step": 642
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999680052519585e-05,
+      "loss": 0.8405,
+      "step": 643
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999670430680568e-05,
+      "loss": 56.9676,
+      "step": 644
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999660666302098e-05,
+      "loss": 0.7513,
+      "step": 645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999650759384453e-05,
+      "loss": 58.6096,
+      "step": 646
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999640709927916e-05,
+      "loss": 60.4495,
+      "step": 647
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999630517932771e-05,
+      "loss": 57.3314,
+      "step": 648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999620183399313e-05,
+      "loss": 0.8185,
+      "step": 649
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999609706327834e-05,
+      "loss": 56.0137,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999599086718632e-05,
+      "loss": 59.8839,
+      "step": 651
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999588324572011e-05,
+      "loss": 0.0958,
+      "step": 652
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999577419888278e-05,
+      "loss": 57.4501,
+      "step": 653
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999566372667744e-05,
+      "loss": 62.8978,
+      "step": 654
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999555182910722e-05,
+      "loss": 57.1134,
+      "step": 655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999543850617535e-05,
+      "loss": 56.2888,
+      "step": 656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999532375788502e-05,
+      "loss": 0.8472,
+      "step": 657
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999520758423952e-05,
+      "loss": 55.3846,
+      "step": 658
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999508998524215e-05,
+      "loss": 53.9135,
+      "step": 659
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999497096089627e-05,
+      "loss": 0.8085,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999485051120529e-05,
+      "loss": 56.7814,
+      "step": 661
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999472863617263e-05,
+      "loss": 0.7603,
+      "step": 662
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999460533580175e-05,
+      "loss": 58.7203,
+      "step": 663
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999448061009619e-05,
+      "loss": 56.1763,
+      "step": 664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99943544590595e-05,
+      "loss": 0.7926,
+      "step": 665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999422688269525e-05,
+      "loss": 60.7915,
+      "step": 666
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999409788100711e-05,
+      "loss": 55.4846,
+      "step": 667
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999396745399875e-05,
+      "loss": 58.0443,
+      "step": 668
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999383560167388e-05,
+      "loss": 62.7707,
+      "step": 669
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999370232403627e-05,
+      "loss": 57.0773,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999356762108971e-05,
+      "loss": 56.8935,
+      "step": 671
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999343149283805e-05,
+      "loss": 60.0414,
+      "step": 672
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999329393928516e-05,
+      "loss": 0.7865,
+      "step": 673
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999315496043495e-05,
+      "loss": 57.096,
+      "step": 674
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999301455629142e-05,
+      "loss": 53.9492,
+      "step": 675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999287272685855e-05,
+      "loss": 0.8145,
+      "step": 676
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999272947214038e-05,
+      "loss": 57.9151,
+      "step": 677
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9992584792141e-05,
+      "loss": 0.7753,
+      "step": 678
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999243868686455e-05,
+      "loss": 0.8012,
+      "step": 679
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999229115631515e-05,
+      "loss": 57.1502,
+      "step": 680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999214220049707e-05,
+      "loss": 54.7234,
+      "step": 681
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999199181941449e-05,
+      "loss": 0.8705,
+      "step": 682
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999184001307174e-05,
+      "loss": 0.7912,
+      "step": 683
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999168678147315e-05,
+      "loss": 0.8344,
+      "step": 684
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999153212462306e-05,
+      "loss": 0.791,
+      "step": 685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99913760425259e-05,
+      "loss": 68.3885,
+      "step": 686
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999121853518613e-05,
+      "loss": 64.3645,
+      "step": 687
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999105960260822e-05,
+      "loss": 55.2384,
+      "step": 688
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99908992447967e-05,
+      "loss": 57.0833,
+      "step": 689
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999073746175615e-05,
+      "loss": 57.8469,
+      "step": 690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999057425349119e-05,
+      "loss": 0.7579,
+      "step": 691
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999040962000645e-05,
+      "loss": 55.0385,
+      "step": 692
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999024356130664e-05,
+      "loss": 57.6778,
+      "step": 693
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99900760773965e-05,
+      "loss": 55.9187,
+      "step": 694
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99899071682808e-05,
+      "loss": 0.8434,
+      "step": 695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998973683396435e-05,
+      "loss": 0.8128,
+      "step": 696
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998956507445201e-05,
+      "loss": 63.1576,
+      "step": 697
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998939188974866e-05,
+      "loss": 63.5221,
+      "step": 698
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998921727985926e-05,
+      "loss": 0.8048,
+      "step": 699
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99890412447888e-05,
+      "loss": 58.71,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998886378454226e-05,
+      "loss": 54.2475,
+      "step": 701
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998868489912472e-05,
+      "loss": 0.8018,
+      "step": 702
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99885045885413e-05,
+      "loss": 0.1044,
+      "step": 703
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998832285279709e-05,
+      "loss": 0.754,
+      "step": 704
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998813969189734e-05,
+      "loss": 65.0298,
+      "step": 705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99879551058472e-05,
+      "loss": 59.9428,
+      "step": 706
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998776909465197e-05,
+      "loss": 0.091,
+      "step": 707
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998758165831695e-05,
+      "loss": 64.4085,
+      "step": 708
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998739279684749e-05,
+      "loss": 59.1255,
+      "step": 709
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998720251024896e-05,
+      "loss": 55.8608,
+      "step": 710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998701079852679e-05,
+      "loss": 53.9412,
+      "step": 711
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998681766168646e-05,
+      "loss": 0.8508,
+      "step": 712
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998662309973346e-05,
+      "loss": 62.6175,
+      "step": 713
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998642711267334e-05,
+      "loss": 0.0832,
+      "step": 714
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998622970051168e-05,
+      "loss": 75.4671,
+      "step": 715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998603086325413e-05,
+      "loss": 0.8628,
+      "step": 716
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998583060090634e-05,
+      "loss": 65.0469,
+      "step": 717
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998562891347402e-05,
+      "loss": 0.8507,
+      "step": 718
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998542580096293e-05,
+      "loss": 0.8322,
+      "step": 719
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998522126337885e-05,
+      "loss": 53.1941,
+      "step": 720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998501530072762e-05,
+      "loss": 60.1981,
+      "step": 721
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99848079130151e-05,
+      "loss": 56.9545,
+      "step": 722
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998459910024722e-05,
+      "loss": 51.4491,
+      "step": 723
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998438886242994e-05,
+      "loss": 0.1027,
+      "step": 724
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998417719956922e-05,
+      "loss": 54.1232,
+      "step": 725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998396411167111e-05,
+      "loss": 54.1459,
+      "step": 726
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998374959874169e-05,
+      "loss": 56.4646,
+      "step": 727
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998353366078708e-05,
+      "loss": 0.793,
+      "step": 728
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998331629781343e-05,
+      "loss": 53.778,
+      "step": 729
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998309750982692e-05,
+      "loss": 53.8147,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998287729683383e-05,
+      "loss": 0.1165,
+      "step": 731
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99826556588404e-05,
+      "loss": 0.8328,
+      "step": 732
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998243259585297e-05,
+      "loss": 0.7732,
+      "step": 733
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998220810787789e-05,
+      "loss": 53.2335,
+      "step": 734
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998198219492155e-05,
+      "loss": 51.4144,
+      "step": 735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998175485699041e-05,
+      "loss": 51.8417,
+      "step": 736
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998152609409094e-05,
+      "loss": 53.8655,
+      "step": 737
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998129590622967e-05,
+      "loss": 0.8308,
+      "step": 738
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998106429341315e-05,
+      "loss": 53.4881,
+      "step": 739
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998083125564799e-05,
+      "loss": 55.592,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998059679294085e-05,
+      "loss": 0.0853,
+      "step": 741
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998036090529839e-05,
+      "loss": 51.5364,
+      "step": 742
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998012359272735e-05,
+      "loss": 0.737,
+      "step": 743
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997988485523448e-05,
+      "loss": 0.7889,
+      "step": 744
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997964469282659e-05,
+      "loss": 50.6328,
+      "step": 745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997940310551055e-05,
+      "loss": 50.8107,
+      "step": 746
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997916009329322e-05,
+      "loss": 52.3616,
+      "step": 747
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997891565618156e-05,
+      "loss": 0.7967,
+      "step": 748
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997866979418249e-05,
+      "loss": 0.78,
+      "step": 749
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997842250730307e-05,
+      "loss": 0.8141,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997817379555032e-05,
+      "loss": 59.994,
+      "step": 751
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997792365893133e-05,
+      "loss": 0.7896,
+      "step": 752
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997767209745326e-05,
+      "loss": 0.7914,
+      "step": 753
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997741911112324e-05,
+      "loss": 57.8104,
+      "step": 754
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997716469994851e-05,
+      "loss": 51.9617,
+      "step": 755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997690886393632e-05,
+      "loss": 51.1048,
+      "step": 756
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997665160309397e-05,
+      "loss": 0.7463,
+      "step": 757
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997639291742877e-05,
+      "loss": 0.1112,
+      "step": 758
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997613280694813e-05,
+      "loss": 0.074,
+      "step": 759
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997587127165944e-05,
+      "loss": 55.4649,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997560831157016e-05,
+      "loss": 0.1213,
+      "step": 761
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997534392668778e-05,
+      "loss": 0.7891,
+      "step": 762
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997507811701986e-05,
+      "loss": 57.9094,
+      "step": 763
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997481088257397e-05,
+      "loss": 0.7788,
+      "step": 764
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99745422233577e-05,
+      "loss": 0.7761,
+      "step": 765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997427213937876e-05,
+      "loss": 0.7622,
+      "step": 766
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997400063064482e-05,
+      "loss": 57.0441,
+      "step": 767
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997372769716361e-05,
+      "loss": 53.4447,
+      "step": 768
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997345333894294e-05,
+      "loss": 0.7635,
+      "step": 769
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997317755599061e-05,
+      "loss": 0.8497,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997290034831448e-05,
+      "loss": 46.2751,
+      "step": 771
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997262171592248e-05,
+      "loss": 47.0273,
+      "step": 772
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997234165882254e-05,
+      "loss": 0.7748,
+      "step": 773
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997206017702264e-05,
+      "loss": 0.8161,
+      "step": 774
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99717772705308e-05,
+      "loss": 50.8348,
+      "step": 775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997149293935509e-05,
+      "loss": 51.7963,
+      "step": 776
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997120718350361e-05,
+      "loss": 0.081,
+      "step": 777
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997092000298454e-05,
+      "loss": 54.6951,
+      "step": 778
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997063139780602e-05,
+      "loss": 47.175,
+      "step": 779
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99703413679763e-05,
+      "loss": 0.792,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997004991350366e-05,
+      "loss": 48.3753,
+      "step": 781
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99697570343964e-05,
+      "loss": 57.6826,
+      "step": 782
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996946273066286e-05,
+      "loss": 0.0903,
+      "step": 783
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996916700231145e-05,
+      "loss": 49.5398,
+      "step": 784
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996886984935059e-05,
+      "loss": 48.0019,
+      "step": 785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996857127178873e-05,
+      "loss": 48.2475,
+      "step": 786
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996827126963443e-05,
+      "loss": 46.8695,
+      "step": 787
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99679698428962e-05,
+      "loss": 0.7494,
+      "step": 788
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996766699158267e-05,
+      "loss": 52.5435,
+      "step": 789
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996736271570244e-05,
+      "loss": 50.0702,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99670570152642e-05,
+      "loss": 53.6928,
+      "step": 791
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996674989027666e-05,
+      "loss": 50.6638,
+      "step": 792
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99664413407486e-05,
+      "loss": 0.8033,
+      "step": 793
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996613136668879e-05,
+      "loss": 0.7902,
+      "step": 794
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996581996810608e-05,
+      "loss": 49.9987,
+      "step": 795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996550714500933e-05,
+      "loss": 0.786,
+      "step": 796
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996519289740747e-05,
+      "loss": 51.4351,
+      "step": 797
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996487722530948e-05,
+      "loss": 50.7985,
+      "step": 798
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996456012872432e-05,
+      "loss": 49.6539,
+      "step": 799
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996424160766106e-05,
+      "loss": 0.7605,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996392166212878e-05,
+      "loss": 0.7703,
+      "step": 801
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996360029213659e-05,
+      "loss": 0.7824,
+      "step": 802
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996327749769365e-05,
+      "loss": 54.9216,
+      "step": 803
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996295327880916e-05,
+      "loss": 0.8476,
+      "step": 804
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99626276354924e-05,
+      "loss": 0.8199,
+      "step": 805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99623005677526e-05,
+      "loss": 0.8112,
+      "step": 806
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996197207559912e-05,
+      "loss": 54.5773,
+      "step": 807
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996164215904132e-05,
+      "loss": 0.7663,
+      "step": 808
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99613108180886e-05,
+      "loss": 49.1856,
+      "step": 809
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99609780527504e-05,
+      "loss": 49.7773,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996064386303622e-05,
+      "loss": 52.8173,
+      "step": 811
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996030824895559e-05,
+      "loss": 48.7358,
+      "step": 812
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995997121051805e-05,
+      "loss": 49.7502,
+      "step": 813
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995963274773326e-05,
+      "loss": 46.4802,
+      "step": 814
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995929286061083e-05,
+      "loss": 49.8415,
+      "step": 815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995895154916044e-05,
+      "loss": 46.4949,
+      "step": 816
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995860881339185e-05,
+      "loss": 0.816,
+      "step": 817
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995826465331484e-05,
+      "loss": 0.7251,
+      "step": 818
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995791906893919e-05,
+      "loss": 45.9459,
+      "step": 819
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995757206027477e-05,
+      "loss": 0.1031,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995722362733145e-05,
+      "loss": 0.7766,
+      "step": 821
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99568737701192e-05,
+      "loss": 0.7656,
+      "step": 822
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995652248864797e-05,
+      "loss": 47.5859,
+      "step": 823
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995616978292779e-05,
+      "loss": 0.7768,
+      "step": 824
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99558156529687e-05,
+      "loss": 48.4619,
+      "step": 825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995546009878081e-05,
+      "loss": 49.4025,
+      "step": 826
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995510312037426e-05,
+      "loss": 46.5463,
+      "step": 827
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99547447177592e-05,
+      "loss": 45.1723,
+      "step": 828
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995438489094588e-05,
+      "loss": 0.7946,
+      "step": 829
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995402363994453e-05,
+      "loss": 46.0243,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995366096476549e-05,
+      "loss": 46.3453,
+      "step": 831
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995329686541906e-05,
+      "loss": 50.7901,
+      "step": 832
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995293134191561e-05,
+      "loss": 44.4223,
+      "step": 833
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995256439426561e-05,
+      "loss": 45.9642,
+      "step": 834
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995219602247949e-05,
+      "loss": 47.3043,
+      "step": 835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995182622656777e-05,
+      "loss": 0.8143,
+      "step": 836
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995145500654097e-05,
+      "loss": 0.7974,
+      "step": 837
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995108236240969e-05,
+      "loss": 48.4091,
+      "step": 838
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995070829418455e-05,
+      "loss": 44.6531,
+      "step": 839
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99503328018762e-05,
+      "loss": 49.552,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994995588549538e-05,
+      "loss": 45.3943,
+      "step": 841
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99495775450528e-05,
+      "loss": 41.2551,
+      "step": 842
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994919778055927e-05,
+      "loss": 47.5709,
+      "step": 843
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994881659202559e-05,
+      "loss": 47.5055,
+      "step": 844
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994843397946268e-05,
+      "loss": 46.8384,
+      "step": 845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994804994288139e-05,
+      "loss": 0.0852,
+      "step": 846
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994766448229269e-05,
+      "loss": 42.6195,
+      "step": 847
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994727759770757e-05,
+      "loss": 45.0149,
+      "step": 848
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994688928913708e-05,
+      "loss": 45.6458,
+      "step": 849
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994649955659225e-05,
+      "loss": 42.6615,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994610840008423e-05,
+      "loss": 0.7783,
+      "step": 851
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994571581962415e-05,
+      "loss": 0.7875,
+      "step": 852
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99453218152232e-05,
+      "loss": 0.7585,
+      "step": 853
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994492638689263e-05,
+      "loss": 52.0989,
+      "step": 854
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994452953464372e-05,
+      "loss": 52.4868,
+      "step": 855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994413125848775e-05,
+      "loss": 0.8659,
+      "step": 856
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994373155843609e-05,
+      "loss": 0.0942,
+      "step": 857
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994333043450014e-05,
+      "loss": 48.6599,
+      "step": 858
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994292788669135e-05,
+      "loss": 0.8341,
+      "step": 859
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994252391502118e-05,
+      "loss": 0.0557,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994211851950112e-05,
+      "loss": 0.7355,
+      "step": 861
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994171170014279e-05,
+      "loss": 45.6252,
+      "step": 862
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994130345695773e-05,
+      "loss": 45.6707,
+      "step": 863
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994089378995762e-05,
+      "loss": 0.1068,
+      "step": 864
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994048269915412e-05,
+      "loss": 46.2124,
+      "step": 865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994007018455895e-05,
+      "loss": 43.3074,
+      "step": 866
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993965624618388e-05,
+      "loss": 46.4281,
+      "step": 867
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99392408840407e-05,
+      "loss": 45.3727,
+      "step": 868
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993882409814125e-05,
+      "loss": 45.1268,
+      "step": 869
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993840588849742e-05,
+      "loss": 0.0846,
+      "step": 870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993798625512114e-05,
+      "loss": 0.8398,
+      "step": 871
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993756519802436e-05,
+      "loss": 43.3208,
+      "step": 872
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993714271721911e-05,
+      "loss": 0.8136,
+      "step": 873
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993671881271738e-05,
+      "loss": 48.7253,
+      "step": 874
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993629348453131e-05,
+      "loss": 43.577,
+      "step": 875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993586673267302e-05,
+      "loss": 43.4703,
+      "step": 876
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993543855715463e-05,
+      "loss": 47.1965,
+      "step": 877
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993500895798842e-05,
+      "loss": 41.4844,
+      "step": 878
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993457793518657e-05,
+      "loss": 42.3932,
+      "step": 879
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99341454887614e-05,
+      "loss": 44.7082,
+      "step": 880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993371161872524e-05,
+      "loss": 0.8621,
+      "step": 881
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993327632509046e-05,
+      "loss": 44.743,
+      "step": 882
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993283960786945e-05,
+      "loss": 0.8109,
+      "step": 883
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993240146707469e-05,
+      "loss": 0.102,
+      "step": 884
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993196190271865e-05,
+      "loss": 0.1025,
+      "step": 885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993152091481388e-05,
+      "loss": 0.0824,
+      "step": 886
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993107850337293e-05,
+      "loss": 44.4231,
+      "step": 887
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993063466840842e-05,
+      "loss": 46.022,
+      "step": 888
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993018940993302e-05,
+      "loss": 0.8491,
+      "step": 889
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99297427279594e-05,
+      "loss": 44.0592,
+      "step": 890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992929462250032e-05,
+      "loss": 43.1082,
+      "step": 891
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992884509356854e-05,
+      "loss": 0.8246,
+      "step": 892
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992839414117687e-05,
+      "loss": 0.7989,
+      "step": 893
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992794176533818e-05,
+      "loss": 0.0682,
+      "step": 894
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992748796606538e-05,
+      "loss": 42.0558,
+      "step": 895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992703274337136e-05,
+      "loss": 0.7922,
+      "step": 896
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992657609726914e-05,
+      "loss": 43.1652,
+      "step": 897
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992611802777172e-05,
+      "loss": 0.074,
+      "step": 898
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992565853489217e-05,
+      "loss": 0.744,
+      "step": 899
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99251976186436e-05,
+      "loss": 43.3906,
+      "step": 900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992473527903913e-05,
+      "loss": 0.8167,
+      "step": 901
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992427151609193e-05,
+      "loss": 44.4976,
+      "step": 902
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992380632981527e-05,
+      "loss": 0.092,
+      "step": 903
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992333972022237e-05,
+      "loss": 0.7497,
+      "step": 904
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992287168732655e-05,
+      "loss": 0.7462,
+      "step": 905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992240223114113e-05,
+      "loss": 0.7407,
+      "step": 906
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992193135167953e-05,
+      "loss": 42.3856,
+      "step": 907
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992145904895516e-05,
+      "loss": 41.3211,
+      "step": 908
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992098532298147e-05,
+      "loss": 0.7528,
+      "step": 909
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9920510173772e-05,
+      "loss": 40.8779,
+      "step": 910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992003360134026e-05,
+      "loss": 0.7708,
+      "step": 911
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991955560569984e-05,
+      "loss": 44.5223,
+      "step": 912
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99190761868644e-05,
+      "loss": 0.0994,
+      "step": 913
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991859534484757e-05,
+      "loss": 0.7706,
+      "step": 914
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991811307966309e-05,
+      "loss": 43.3219,
+      "step": 915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99176293913247e-05,
+      "loss": 0.8146,
+      "step": 916
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991714427984618e-05,
+      "loss": 0.7722,
+      "step": 917
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991665774524135e-05,
+      "loss": 0.787,
+      "step": 918
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991616978752411e-05,
+      "loss": 42.4615,
+      "step": 919
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991568040670836e-05,
+      "loss": 0.7516,
+      "step": 920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991518960280806e-05,
+      "loss": 0.8257,
+      "step": 921
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991469737583718e-05,
+      "loss": 44.4727,
+      "step": 922
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991420372580977e-05,
+      "loss": 47.6547,
+      "step": 923
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99137086527399e-05,
+      "loss": 43.7198,
+      "step": 924
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991321215664169e-05,
+      "loss": 42.7662,
+      "step": 925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991271423752928e-05,
+      "loss": 0.7782,
+      "step": 926
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991221489541688e-05,
+      "loss": 40.0944,
+      "step": 927
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991171413031872e-05,
+      "loss": 40.2179,
+      "step": 928
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991121194224908e-05,
+      "loss": 44.5593,
+      "step": 929
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991070833122227e-05,
+      "loss": 40.8641,
+      "step": 930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991020329725267e-05,
+      "loss": 39.7671,
+      "step": 931
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990969684035465e-05,
+      "loss": 40.8226,
+      "step": 932
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990918896054267e-05,
+      "loss": 42.0502,
+      "step": 933
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990867965783117e-05,
+      "loss": 41.1927,
+      "step": 934
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990816893223473e-05,
+      "loss": 41.7947,
+      "step": 935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990765678376788e-05,
+      "loss": 44.8777,
+      "step": 936
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99071432124452e-05,
+      "loss": 0.7963,
+      "step": 937
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990662821828138e-05,
+      "loss": 38.8794,
+      "step": 938
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990611180129107e-05,
+      "loss": 40.9298,
+      "step": 939
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9905593961489e-05,
+      "loss": 0.7727,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990507469888992e-05,
+      "loss": 0.7937,
+      "step": 941
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990455401350865e-05,
+      "loss": 0.0987,
+      "step": 942
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990403190536004e-05,
+      "loss": 0.078,
+      "step": 943
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990350837445897e-05,
+      "loss": 41.5738,
+      "step": 944
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990298342082036e-05,
+      "loss": 43.5836,
+      "step": 945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990245704445917e-05,
+      "loss": 0.0591,
+      "step": 946
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990192924539042e-05,
+      "loss": 41.0657,
+      "step": 947
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990140002362915e-05,
+      "loss": 37.1736,
+      "step": 948
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990086937919045e-05,
+      "loss": 0.101,
+      "step": 949
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990033731208946e-05,
+      "loss": 0.0896,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989980382234134e-05,
+      "loss": 40.6026,
+      "step": 951
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989926890996129e-05,
+      "loss": 45.012,
+      "step": 952
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989873257496457e-05,
+      "loss": 41.4056,
+      "step": 953
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989819481736646e-05,
+      "loss": 42.2816,
+      "step": 954
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989765563718232e-05,
+      "loss": 42.708,
+      "step": 955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98971150344275e-05,
+      "loss": 39.6326,
+      "step": 956
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989657300911742e-05,
+      "loss": 0.8003,
+      "step": 957
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98960295612675e-05,
+      "loss": 0.8138,
+      "step": 958
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989548469089329e-05,
+      "loss": 0.8525,
+      "step": 959
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989493839801027e-05,
+      "loss": 43.9529,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989439068263406e-05,
+      "loss": 45.664,
+      "step": 961
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989384154478026e-05,
+      "loss": 45.0837,
+      "step": 962
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989329098446451e-05,
+      "loss": 40.0746,
+      "step": 963
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989273900170251e-05,
+      "loss": 0.0791,
+      "step": 964
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989218559651001e-05,
+      "loss": 42.1705,
+      "step": 965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98916307689028e-05,
+      "loss": 45.2847,
+      "step": 966
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989107451889666e-05,
+      "loss": 0.8713,
+      "step": 967
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989051684650748e-05,
+      "loss": 42.5262,
+      "step": 968
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988995775175114e-05,
+      "loss": 0.864,
+      "step": 969
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98893972346436e-05,
+      "loss": 42.4058,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988883529520081e-05,
+      "loss": 37.7888,
+      "step": 971
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988827193343882e-05,
+      "loss": 36.9758,
+      "step": 972
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988770714937367e-05,
+      "loss": 38.444,
+      "step": 973
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98871409430215e-05,
+      "loss": 44.2606,
+      "step": 974
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98865733143984e-05,
+      "loss": 0.7842,
+      "step": 975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988600426352058e-05,
+      "loss": 42.7503,
+      "step": 976
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988543379040427e-05,
+      "loss": 0.8062,
+      "step": 977
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98848618950657e-05,
+      "loss": 0.8083,
+      "step": 978
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988428857752122e-05,
+      "loss": 42.93,
+      "step": 979
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988371383778716e-05,
+      "loss": 39.3097,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988313767587988e-05,
+      "loss": 42.2745,
+      "step": 981
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988256009181585e-05,
+      "loss": 40.7358,
+      "step": 982
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98819810856115e-05,
+      "loss": 40.2981,
+      "step": 983
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988140065728335e-05,
+      "loss": 39.2419,
+      "step": 984
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988081880684794e-05,
+      "loss": 38.8702,
+      "step": 985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988023553432189e-05,
+      "loss": 0.0822,
+      "step": 986
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987965083972178e-05,
+      "loss": 0.7719,
+      "step": 987
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987906472306432e-05,
+      "loss": 0.7992,
+      "step": 988
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987847718436621e-05,
+      "loss": 0.0747,
+      "step": 989
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987788822364419e-05,
+      "loss": 45.5594,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987729784091505e-05,
+      "loss": 0.755,
+      "step": 991
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987670603619563e-05,
+      "loss": 37.5209,
+      "step": 992
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987611280950282e-05,
+      "loss": 0.7539,
+      "step": 993
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98755181608535e-05,
+      "loss": 38.3338,
+      "step": 994
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987492209026463e-05,
+      "loss": 0.7962,
+      "step": 995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987432459775322e-05,
+      "loss": 41.4963,
+      "step": 996
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987372568333628e-05,
+      "loss": 38.2427,
+      "step": 997
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987312534703093e-05,
+      "loss": 0.7518,
+      "step": 998
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987252358885422e-05,
+      "loss": 40.5426,
+      "step": 999
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987192040882337e-05,
+      "loss": 40.0752,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987131580695554e-05,
+      "loss": 0.7522,
+      "step": 1001
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987070978326799e-05,
+      "loss": 0.0953,
+      "step": 1002
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987010233777795e-05,
+      "loss": 0.8303,
+      "step": 1003
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986949347050278e-05,
+      "loss": 39.4784,
+      "step": 1004
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986888318145984e-05,
+      "loss": 0.7276,
+      "step": 1005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986827147066652e-05,
+      "loss": 0.0974,
+      "step": 1006
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986765833814025e-05,
+      "loss": 0.7416,
+      "step": 1007
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986704378389853e-05,
+      "loss": 0.082,
+      "step": 1008
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986642780795885e-05,
+      "loss": 0.7249,
+      "step": 1009
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98658104103388e-05,
+      "loss": 0.7408,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986519159105599e-05,
+      "loss": 37.4819,
+      "step": 1011
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986457135012803e-05,
+      "loss": 0.7729,
+      "step": 1012
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986394968757262e-05,
+      "loss": 40.0902,
+      "step": 1013
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986332660340748e-05,
+      "loss": 0.7326,
+      "step": 1014
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986270209765038e-05,
+      "loss": 37.2437,
+      "step": 1015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98620761703191e-05,
+      "loss": 0.7429,
+      "step": 1016
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986144882143153e-05,
+      "loss": 38.6339,
+      "step": 1017
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986082005100552e-05,
+      "loss": 40.3711,
+      "step": 1018
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986018985905901e-05,
+      "loss": 38.9105,
+      "step": 1019
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985955824560997e-05,
+      "loss": 36.4053,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985892521067639e-05,
+      "loss": 0.7846,
+      "step": 1021
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985829075427633e-05,
+      "loss": 0.8233,
+      "step": 1022
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985765487642787e-05,
+      "loss": 0.7949,
+      "step": 1023
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985701757714916e-05,
+      "loss": 43.1573,
+      "step": 1024
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985637885645835e-05,
+      "loss": 0.7168,
+      "step": 1025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985573871437365e-05,
+      "loss": 0.0924,
+      "step": 1026
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985509715091332e-05,
+      "loss": 43.4347,
+      "step": 1027
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985445416609564e-05,
+      "loss": 43.3967,
+      "step": 1028
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985380975993895e-05,
+      "loss": 0.8336,
+      "step": 1029
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985316393246161e-05,
+      "loss": 39.7125,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985251668368205e-05,
+      "loss": 38.0025,
+      "step": 1031
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985186801361871e-05,
+      "loss": 0.774,
+      "step": 1032
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985121792229008e-05,
+      "loss": 33.3877,
+      "step": 1033
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985056640971472e-05,
+      "loss": 38.5285,
+      "step": 1034
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984991347591118e-05,
+      "loss": 38.9747,
+      "step": 1035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984925912089806e-05,
+      "loss": 0.8136,
+      "step": 1036
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984860334469405e-05,
+      "loss": 34.4424,
+      "step": 1037
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984794614731785e-05,
+      "loss": 39.5213,
+      "step": 1038
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984728752878815e-05,
+      "loss": 38.6009,
+      "step": 1039
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984662748912378e-05,
+      "loss": 37.163,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984596602834353e-05,
+      "loss": 0.8315,
+      "step": 1041
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984530314646625e-05,
+      "loss": 36.9457,
+      "step": 1042
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984463884351085e-05,
+      "loss": 35.7729,
+      "step": 1043
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984397311949627e-05,
+      "loss": 36.118,
+      "step": 1044
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984330597444149e-05,
+      "loss": 0.1038,
+      "step": 1045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984263740836555e-05,
+      "loss": 0.7919,
+      "step": 1046
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984196742128747e-05,
+      "loss": 0.7221,
+      "step": 1047
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984129601322636e-05,
+      "loss": 0.8392,
+      "step": 1048
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984062318420139e-05,
+      "loss": 42.5577,
+      "step": 1049
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983994893423172e-05,
+      "loss": 0.8255,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983927326333658e-05,
+      "loss": 41.597,
+      "step": 1051
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983859617153521e-05,
+      "loss": 0.8176,
+      "step": 1052
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983791765884695e-05,
+      "loss": 40.0665,
+      "step": 1053
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983723772529114e-05,
+      "loss": 0.8175,
+      "step": 1054
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983655637088716e-05,
+      "loss": 34.9421,
+      "step": 1055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98358735956544e-05,
+      "loss": 36.9441,
+      "step": 1056
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983518939961238e-05,
+      "loss": 36.3253,
+      "step": 1057
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983450378278057e-05,
+      "loss": 0.092,
+      "step": 1058
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98338167451785e-05,
+      "loss": 0.8019,
+      "step": 1059
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983312828682583e-05,
+      "loss": 40.2148,
+      "step": 1060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983243840774212e-05,
+      "loss": 0.8439,
+      "step": 1061
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983174710794707e-05,
+      "loss": 41.8438,
+      "step": 1062
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983105438746037e-05,
+      "loss": 39.4416,
+      "step": 1063
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98303602463018e-05,
+      "loss": 0.8352,
+      "step": 1064
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982966468449109e-05,
+      "loss": 39.8825,
+      "step": 1065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982896770204814e-05,
+      "loss": 40.2164,
+      "step": 1066
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982826929899278e-05,
+      "loss": 0.8172,
+      "step": 1067
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982756947534492e-05,
+      "loss": 36.3916,
+      "step": 1068
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982686823112454e-05,
+      "loss": 0.7426,
+      "step": 1069
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982616556635162e-05,
+      "loss": 40.3618,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982546148104617e-05,
+      "loss": 0.8464,
+      "step": 1071
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982475597522828e-05,
+      "loss": 41.9259,
+      "step": 1072
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982404904891808e-05,
+      "loss": 0.8012,
+      "step": 1073
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982334070213572e-05,
+      "loss": 37.5189,
+      "step": 1074
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982263093490136e-05,
+      "loss": 0.791,
+      "step": 1075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982191974723527e-05,
+      "loss": 37.5477,
+      "step": 1076
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982120713915772e-05,
+      "loss": 38.5834,
+      "step": 1077
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982049311068902e-05,
+      "loss": 0.8031,
+      "step": 1078
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981977766184951e-05,
+      "loss": 0.7704,
+      "step": 1079
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981906079265963e-05,
+      "loss": 0.8149,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981834250313978e-05,
+      "loss": 0.7919,
+      "step": 1081
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981762279331046e-05,
+      "loss": 40.3849,
+      "step": 1082
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981690166319217e-05,
+      "loss": 0.802,
+      "step": 1083
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98161791128055e-05,
+      "loss": 0.8029,
+      "step": 1084
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981545514217101e-05,
+      "loss": 43.2182,
+      "step": 1085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981472975130937e-05,
+      "loss": 0.7915,
+      "step": 1086
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981400294024123e-05,
+      "loss": 0.8249,
+      "step": 1087
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981327470898736e-05,
+      "loss": 0.1059,
+      "step": 1088
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981254505756847e-05,
+      "loss": 0.7992,
+      "step": 1089
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981181398600541e-05,
+      "loss": 41.8181,
+      "step": 1090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9811081494319e-05,
+      "loss": 38.192,
+      "step": 1091
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98103475825301e-05,
+      "loss": 0.7747,
+      "step": 1092
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980961225065965e-05,
+      "loss": 0.7973,
+      "step": 1093
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980887549872864e-05,
+      "loss": 40.507,
+      "step": 1094
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980813732675804e-05,
+      "loss": 0.7352,
+      "step": 1095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980739773476892e-05,
+      "loss": 0.802,
+      "step": 1096
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980665672278236e-05,
+      "loss": 35.7243,
+      "step": 1097
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980591429081947e-05,
+      "loss": 0.0649,
+      "step": 1098
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980517043890143e-05,
+      "loss": 34.9443,
+      "step": 1099
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980442516704944e-05,
+      "loss": 0.7955,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980367847528475e-05,
+      "loss": 41.5374,
+      "step": 1101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980293036362866e-05,
+      "loss": 40.9436,
+      "step": 1102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980218083210247e-05,
+      "loss": 0.7755,
+      "step": 1103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980142988072759e-05,
+      "loss": 0.7934,
+      "step": 1104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98006775095254e-05,
+      "loss": 34.6644,
+      "step": 1105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979992371851734e-05,
+      "loss": 0.7808,
+      "step": 1106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979916850772494e-05,
+      "loss": 0.7252,
+      "step": 1107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97984118771697e-05,
+      "loss": 33.8543,
+      "step": 1108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979765382687321e-05,
+      "loss": 0.7686,
+      "step": 1109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979689435685705e-05,
+      "loss": 34.8858,
+      "step": 1110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97961334671429e-05,
+      "loss": 34.3199,
+      "step": 1111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979537115775246e-05,
+      "loss": 0.7615,
+      "step": 1112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979460742870743e-05,
+      "loss": 0.7873,
+      "step": 1113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979384228002962e-05,
+      "loss": 34.1854,
+      "step": 1114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979307571174082e-05,
+      "loss": 0.1072,
+      "step": 1115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97923077238629e-05,
+      "loss": 33.1289,
+      "step": 1116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979153831641774e-05,
+      "loss": 35.012,
+      "step": 1117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979076748942727e-05,
+      "loss": 39.296,
+      "step": 1118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97899952429135e-05,
+      "loss": 33.9648,
+      "step": 1119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978922157689842e-05,
+      "loss": 0.7911,
+      "step": 1120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978844649140408e-05,
+      "loss": 0.8125,
+      "step": 1121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978766998645261e-05,
+      "loss": 35.0827,
+      "step": 1122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978689206206611e-05,
+      "loss": 0.7686,
+      "step": 1123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978611271826679e-05,
+      "loss": 0.0978,
+      "step": 1124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978533195507685e-05,
+      "loss": 0.7496,
+      "step": 1125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978454977251857e-05,
+      "loss": 35.2316,
+      "step": 1126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978376617061422e-05,
+      "loss": 34.968,
+      "step": 1127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978298114938616e-05,
+      "loss": 34.2438,
+      "step": 1128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978219470885676e-05,
+      "loss": 0.7313,
+      "step": 1129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978140684904844e-05,
+      "loss": 36.81,
+      "step": 1130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978061756998369e-05,
+      "loss": 0.7977,
+      "step": 1131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977982687168498e-05,
+      "loss": 37.2295,
+      "step": 1132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977903475417486e-05,
+      "loss": 37.0804,
+      "step": 1133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977824121747592e-05,
+      "loss": 0.8081,
+      "step": 1134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977744626161078e-05,
+      "loss": 0.7925,
+      "step": 1135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977664988660211e-05,
+      "loss": 0.0729,
+      "step": 1136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97758520924726e-05,
+      "loss": 0.0787,
+      "step": 1137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977505287924501e-05,
+      "loss": 40.0584,
+      "step": 1138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977425224694212e-05,
+      "loss": 0.8025,
+      "step": 1139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977345019558675e-05,
+      "loss": 37.3027,
+      "step": 1140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977264672520177e-05,
+      "loss": 0.7838,
+      "step": 1141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97718418358101e-05,
+      "loss": 0.8099,
+      "step": 1142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977103552743465e-05,
+      "loss": 0.769,
+      "step": 1143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977022780009846e-05,
+      "loss": 38.291,
+      "step": 1144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976941865382453e-05,
+      "loss": 0.7842,
+      "step": 1145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976860808863591e-05,
+      "loss": 37.6981,
+      "step": 1146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976779610455573e-05,
+      "loss": 0.7776,
+      "step": 1147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976698270160716e-05,
+      "loss": 38.3695,
+      "step": 1148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976616787981334e-05,
+      "loss": 38.6725,
+      "step": 1149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976535163919756e-05,
+      "loss": 38.7416,
+      "step": 1150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976453397978304e-05,
+      "loss": 39.4618,
+      "step": 1151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976371490159311e-05,
+      "loss": 35.7515,
+      "step": 1152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976289440465112e-05,
+      "loss": 0.7432,
+      "step": 1153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976207248898046e-05,
+      "loss": 35.9662,
+      "step": 1154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976124915460459e-05,
+      "loss": 38.2222,
+      "step": 1155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976042440154694e-05,
+      "loss": 36.4155,
+      "step": 1156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975959822983105e-05,
+      "loss": 36.5118,
+      "step": 1157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975877063948047e-05,
+      "loss": 0.1,
+      "step": 1158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97579416305188e-05,
+      "loss": 36.6379,
+      "step": 1159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975711120296964e-05,
+      "loss": 0.0729,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975627935685672e-05,
+      "loss": 34.5785,
+      "step": 1161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975544609220371e-05,
+      "loss": 0.0947,
+      "step": 1162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97546114090344e-05,
+      "loss": 0.7971,
+      "step": 1163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975377530737256e-05,
+      "loss": 35.3699,
+      "step": 1164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975293778724204e-05,
+      "loss": 34.3238,
+      "step": 1165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975209884866671e-05,
+      "loss": 0.7877,
+      "step": 1166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975125849167049e-05,
+      "loss": 35.3772,
+      "step": 1167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975041671627736e-05,
+      "loss": 0.8288,
+      "step": 1168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974957352251129e-05,
+      "loss": 37.1981,
+      "step": 1169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974872891039632e-05,
+      "loss": 36.5492,
+      "step": 1170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974788287995655e-05,
+      "loss": 35.7128,
+      "step": 1171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974703543121609e-05,
+      "loss": 36.5611,
+      "step": 1172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97461865641991e-05,
+      "loss": 33.1134,
+      "step": 1173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974533627892978e-05,
+      "loss": 35.7541,
+      "step": 1174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974448457543238e-05,
+      "loss": 0.8277,
+      "step": 1175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974363145373115e-05,
+      "loss": 33.0667,
+      "step": 1176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974277691385045e-05,
+      "loss": 35.036,
+      "step": 1177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974192095581465e-05,
+      "loss": 0.8214,
+      "step": 1178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97410635796481e-05,
+      "loss": 35.7679,
+      "step": 1179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974020478537529e-05,
+      "loss": 35.3367,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97393445730207e-05,
+      "loss": 35.8802,
+      "step": 1181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973848294260884e-05,
+      "loss": 0.755,
+      "step": 1182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973761989416426e-05,
+      "loss": 0.783,
+      "step": 1183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973675542771162e-05,
+      "loss": 34.8359,
+      "step": 1184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973588954327551e-05,
+      "loss": 0.7186,
+      "step": 1185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973502224088063e-05,
+      "loss": 0.7472,
+      "step": 1186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973415352055172e-05,
+      "loss": 0.0743,
+      "step": 1187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973328338231355e-05,
+      "loss": 0.0707,
+      "step": 1188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97324118261909e-05,
+      "loss": 0.0819,
+      "step": 1189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973153885220865e-05,
+      "loss": 35.8849,
+      "step": 1190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973066446039168e-05,
+      "loss": 34.0087,
+      "step": 1191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97297886507649e-05,
+      "loss": 0.8031,
+      "step": 1192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972891142335328e-05,
+      "loss": 0.7592,
+      "step": 1193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972803277818185e-05,
+      "loss": 41.0797,
+      "step": 1194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972715271527566e-05,
+      "loss": 33.079,
+      "step": 1195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97262712346598e-05,
+      "loss": 0.7536,
+      "step": 1196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972538833635938e-05,
+      "loss": 32.7657,
+      "step": 1197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972450402039959e-05,
+      "loss": 31.663,
+      "step": 1198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972361828680562e-05,
+      "loss": 35.2476,
+      "step": 1199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972273113560275e-05,
+      "loss": 0.8253,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972184256681626e-05,
+      "loss": 0.7294,
+      "step": 1201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972095258047147e-05,
+      "loss": 37.6048,
+      "step": 1202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972006117659378e-05,
+      "loss": 0.0777,
+      "step": 1203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97191683552086e-05,
+      "loss": 0.816,
+      "step": 1204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971827411634136e-05,
+      "loss": 0.7424,
+      "step": 1205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971737846001756e-05,
+      "loss": 35.8895,
+      "step": 1206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971648138626275e-05,
+      "loss": 0.7541,
+      "step": 1207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97155828951025e-05,
+      "loss": 0.7593,
+      "step": 1208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971468298656243e-05,
+      "loss": 32.9072,
+      "step": 1209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971378166066818e-05,
+      "loss": 33.2183,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971287891744548e-05,
+      "loss": 35.8355,
+      "step": 1211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971197475692002e-05,
+      "loss": 34.4221,
+      "step": 1212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971106917911762e-05,
+      "loss": 32.7337,
+      "step": 1213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971016218406406e-05,
+      "loss": 34.9995,
+      "step": 1214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970925377178523e-05,
+      "loss": 31.2476,
+      "step": 1215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970834394230701e-05,
+      "loss": 33.1895,
+      "step": 1216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970743269565536e-05,
+      "loss": 35.7155,
+      "step": 1217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970652003185624e-05,
+      "loss": 0.8119,
+      "step": 1218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970560595093565e-05,
+      "loss": 31.2195,
+      "step": 1219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97046904529197e-05,
+      "loss": 32.5519,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970377353783447e-05,
+      "loss": 35.5655,
+      "step": 1221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970285520570609e-05,
+      "loss": 0.7661,
+      "step": 1222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970193545656075e-05,
+      "loss": 35.624,
+      "step": 1223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970101429042465e-05,
+      "loss": 33.1225,
+      "step": 1224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970009170732409e-05,
+      "loss": 0.8138,
+      "step": 1225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969916770728535e-05,
+      "loss": 0.7306,
+      "step": 1226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969824229033479e-05,
+      "loss": 35.0851,
+      "step": 1227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969731545649876e-05,
+      "loss": 37.0302,
+      "step": 1228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969638720580372e-05,
+      "loss": 32.1392,
+      "step": 1229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969545753827612e-05,
+      "loss": 0.0743,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969452645394246e-05,
+      "loss": 0.1071,
+      "step": 1231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969359395282928e-05,
+      "loss": 38.0738,
+      "step": 1232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96926600349632e-05,
+      "loss": 37.3595,
+      "step": 1233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96917247003708e-05,
+      "loss": 39.8027,
+      "step": 1234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969078794907878e-05,
+      "loss": 36.2625,
+      "step": 1235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968984978111382e-05,
+      "loss": 32.9242,
+      "step": 1236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968891019650268e-05,
+      "loss": 0.0943,
+      "step": 1237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968796919527214e-05,
+      "loss": 0.0857,
+      "step": 1238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968702677744906e-05,
+      "loss": 46.0455,
+      "step": 1239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968608294306025e-05,
+      "loss": 0.7881,
+      "step": 1240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968513769213268e-05,
+      "loss": 0.8085,
+      "step": 1241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968419102469326e-05,
+      "loss": 64.4274,
+      "step": 1242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968324294076898e-05,
+      "loss": 0.8075,
+      "step": 1243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96822934403869e-05,
+      "loss": 55.29,
+      "step": 1244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968134252357406e-05,
+      "loss": 50.5044,
+      "step": 1245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968039019035757e-05,
+      "loss": 35.5376,
+      "step": 1246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967943644076461e-05,
+      "loss": 34.6073,
+      "step": 1247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967848127482234e-05,
+      "loss": 0.841,
+      "step": 1248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9677524692558e-05,
+      "loss": 36.3338,
+      "step": 1249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967656669399887e-05,
+      "loss": 0.1045,
+      "step": 1250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967560727917224e-05,
+      "loss": 0.7881,
+      "step": 1251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967464644810552e-05,
+      "loss": 0.8107,
+      "step": 1252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967368420082602e-05,
+      "loss": 0.759,
+      "step": 1253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967272053736124e-05,
+      "loss": 0.7763,
+      "step": 1254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967175545773863e-05,
+      "loss": 0.8007,
+      "step": 1255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96707889619857e-05,
+      "loss": 49.4257,
+      "step": 1256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966982105013001e-05,
+      "loss": 50.5396,
+      "step": 1257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966885172219915e-05,
+      "loss": 47.7095,
+      "step": 1258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966788097822077e-05,
+      "loss": 0.7624,
+      "step": 1259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966690881822253e-05,
+      "loss": 0.0643,
+      "step": 1260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966593524223216e-05,
+      "loss": 0.7497,
+      "step": 1261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966496025027739e-05,
+      "loss": 31.9949,
+      "step": 1262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966398384238604e-05,
+      "loss": 37.028,
+      "step": 1263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966300601858596e-05,
+      "loss": 0.7889,
+      "step": 1264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966202677890499e-05,
+      "loss": 0.7894,
+      "step": 1265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966104612337108e-05,
+      "loss": 43.7419,
+      "step": 1266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966006405201215e-05,
+      "loss": 43.8021,
+      "step": 1267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965908056485626e-05,
+      "loss": 0.7392,
+      "step": 1268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965809566193138e-05,
+      "loss": 38.2233,
+      "step": 1269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965710934326564e-05,
+      "loss": 39.0314,
+      "step": 1270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965612160888715e-05,
+      "loss": 35.2953,
+      "step": 1271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965513245882405e-05,
+      "loss": 33.9938,
+      "step": 1272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965414189310456e-05,
+      "loss": 36.1571,
+      "step": 1273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965314991175691e-05,
+      "loss": 0.8277,
+      "step": 1274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965215651480937e-05,
+      "loss": 0.7918,
+      "step": 1275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96511617022903e-05,
+      "loss": 39.0192,
+      "step": 1276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965016547422803e-05,
+      "loss": 0.0801,
+      "step": 1277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964916783065097e-05,
+      "loss": 39.4654,
+      "step": 1278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964816877158756e-05,
+      "loss": 40.4523,
+      "step": 1279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964716829706628e-05,
+      "loss": 35.3714,
+      "step": 1280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964616640711565e-05,
+      "loss": 0.7477,
+      "step": 1281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964516310176427e-05,
+      "loss": 35.183,
+      "step": 1282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964415838104071e-05,
+      "loss": 0.0681,
+      "step": 1283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964315224497361e-05,
+      "loss": 35.0917,
+      "step": 1284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964214469359168e-05,
+      "loss": 0.706,
+      "step": 1285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964113572692361e-05,
+      "loss": 33.9968,
+      "step": 1286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96401253449982e-05,
+      "loss": 35.8046,
+      "step": 1287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963911354784425e-05,
+      "loss": 0.0964,
+      "step": 1288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963810033549059e-05,
+      "loss": 0.8118,
+      "step": 1289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963708570796611e-05,
+      "loss": 34.2565,
+      "step": 1290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963606966529975e-05,
+      "loss": 35.3381,
+      "step": 1291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963505220752046e-05,
+      "loss": 0.7785,
+      "step": 1292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963403333465726e-05,
+      "loss": 0.7896,
+      "step": 1293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963301304673919e-05,
+      "loss": 0.7891,
+      "step": 1294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963199134379535e-05,
+      "loss": 34.2142,
+      "step": 1295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963096822585484e-05,
+      "loss": 36.6389,
+      "step": 1296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962994369294687e-05,
+      "loss": 0.8947,
+      "step": 1297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962891774510063e-05,
+      "loss": 0.0705,
+      "step": 1298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962789038234537e-05,
+      "loss": 33.5011,
+      "step": 1299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962686160471036e-05,
+      "loss": 34.7782,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962583141222496e-05,
+      "loss": 33.984,
+      "step": 1301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962479980491852e-05,
+      "loss": 34.0491,
+      "step": 1302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962376678282048e-05,
+      "loss": 0.8912,
+      "step": 1303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962273234596024e-05,
+      "loss": 0.7679,
+      "step": 1304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962169649436736e-05,
+      "loss": 0.7616,
+      "step": 1305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962065922807131e-05,
+      "loss": 0.7676,
+      "step": 1306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961962054710168e-05,
+      "loss": 0.7585,
+      "step": 1307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961858045148811e-05,
+      "loss": 36.6079,
+      "step": 1308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961753894126022e-05,
+      "loss": 0.8174,
+      "step": 1309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961649601644771e-05,
+      "loss": 34.6063,
+      "step": 1310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961545167708033e-05,
+      "loss": 35.5035,
+      "step": 1311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961440592318784e-05,
+      "loss": 32.7541,
+      "step": 1312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961335875480007e-05,
+      "loss": 34.0843,
+      "step": 1313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961231017194684e-05,
+      "loss": 0.8078,
+      "step": 1314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961126017465807e-05,
+      "loss": 32.5249,
+      "step": 1315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961020876296369e-05,
+      "loss": 34.9243,
+      "step": 1316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960915593689369e-05,
+      "loss": 34.7826,
+      "step": 1317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960810169647806e-05,
+      "loss": 32.2834,
+      "step": 1318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960704604174688e-05,
+      "loss": 33.5919,
+      "step": 1319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960598897273023e-05,
+      "loss": 32.9346,
+      "step": 1320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960493048945825e-05,
+      "loss": 0.0932,
+      "step": 1321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960387059196111e-05,
+      "loss": 36.2161,
+      "step": 1322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960280928026905e-05,
+      "loss": 0.8371,
+      "step": 1323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960174655441231e-05,
+      "loss": 0.7639,
+      "step": 1324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96006824144212e-05,
+      "loss": 34.752,
+      "step": 1325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959961686032604e-05,
+      "loss": 33.4641,
+      "step": 1326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959854989215722e-05,
+      "loss": 34.3714,
+      "step": 1327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959748150994517e-05,
+      "loss": 0.0971,
+      "step": 1328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959641171372033e-05,
+      "loss": 0.0688,
+      "step": 1329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95953405035132e-05,
+      "loss": 0.122,
+      "step": 1330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959426787935434e-05,
+      "loss": 36.3872,
+      "step": 1331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959319384127431e-05,
+      "loss": 0.7785,
+      "step": 1332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959211838930375e-05,
+      "loss": 0.7553,
+      "step": 1333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959104152347329e-05,
+      "loss": 0.0577,
+      "step": 1334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958996324381366e-05,
+      "loss": 36.437,
+      "step": 1335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958888355035561e-05,
+      "loss": 0.8096,
+      "step": 1336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958780244312988e-05,
+      "loss": 0.0759,
+      "step": 1337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958671992216732e-05,
+      "loss": 35.7865,
+      "step": 1338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958563598749879e-05,
+      "loss": 0.8587,
+      "step": 1339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958455063915519e-05,
+      "loss": 32.784,
+      "step": 1340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958346387716748e-05,
+      "loss": 34.854,
+      "step": 1341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958237570156661e-05,
+      "loss": 0.8124,
+      "step": 1342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958128611238363e-05,
+      "loss": 31.6547,
+      "step": 1343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958019510964959e-05,
+      "loss": 34.5752,
+      "step": 1344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957910269339559e-05,
+      "loss": 0.0836,
+      "step": 1345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95780088636528e-05,
+      "loss": 33.892,
+      "step": 1346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957691362045239e-05,
+      "loss": 34.5005,
+      "step": 1347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957581696382558e-05,
+      "loss": 0.7529,
+      "step": 1348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957471889380364e-05,
+      "loss": 0.8287,
+      "step": 1349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957361941041784e-05,
+      "loss": 0.8281,
+      "step": 1350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95725185136996e-05,
+      "loss": 36.383,
+      "step": 1351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957141620368027e-05,
+      "loss": 36.3137,
+      "step": 1352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957031248039126e-05,
+      "loss": 32.7999,
+      "step": 1353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956920734386405e-05,
+      "loss": 31.8711,
+      "step": 1354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956810079413014e-05,
+      "loss": 37.2337,
+      "step": 1355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95669928312211e-05,
+      "loss": 30.8479,
+      "step": 1356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95658834551685e-05,
+      "loss": 0.795,
+      "step": 1357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956477266600398e-05,
+      "loss": 32.0896,
+      "step": 1358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95636604637592e-05,
+      "loss": 33.6922,
+      "step": 1359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956254684846585e-05,
+      "loss": 0.7922,
+      "step": 1360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956143182015571e-05,
+      "loss": 31.923,
+      "step": 1361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956031537886055e-05,
+      "loss": 0.8204,
+      "step": 1362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95591975246122e-05,
+      "loss": 0.7111,
+      "step": 1363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955807825744254e-05,
+      "loss": 35.4798,
+      "step": 1364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95569575773835e-05,
+      "loss": 37.6352,
+      "step": 1365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955583548446697e-05,
+      "loss": 34.5223,
+      "step": 1366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955471197872499e-05,
+      "loss": 0.0786,
+      "step": 1367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955358706018958e-05,
+      "loss": 0.0743,
+      "step": 1368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955246072889282e-05,
+      "loss": 0.0663,
+      "step": 1369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955133298486678e-05,
+      "loss": 34.9173,
+      "step": 1370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955020382814367e-05,
+      "loss": 33.7932,
+      "step": 1371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954907325875565e-05,
+      "loss": 35.4822,
+      "step": 1372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954794127673494e-05,
+      "loss": 35.5301,
+      "step": 1373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954680788211384e-05,
+      "loss": 0.0767,
+      "step": 1374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954567307492465e-05,
+      "loss": 36.6319,
+      "step": 1375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954453685519973e-05,
+      "loss": 32.8507,
+      "step": 1376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954339922297146e-05,
+      "loss": 32.5385,
+      "step": 1377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954226017827228e-05,
+      "loss": 0.0854,
+      "step": 1378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954111972113467e-05,
+      "loss": 33.4992,
+      "step": 1379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953997785159113e-05,
+      "loss": 36.7409,
+      "step": 1380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953883456967423e-05,
+      "loss": 32.4645,
+      "step": 1381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953768987541658e-05,
+      "loss": 33.0978,
+      "step": 1382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953654376885077e-05,
+      "loss": 0.8411,
+      "step": 1383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953539625000949e-05,
+      "loss": 34.0991,
+      "step": 1384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953424731892548e-05,
+      "loss": 33.1427,
+      "step": 1385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953309697563149e-05,
+      "loss": 0.0731,
+      "step": 1386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95319452201603e-05,
+      "loss": 0.7676,
+      "step": 1387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953079205254474e-05,
+      "loss": 0.0814,
+      "step": 1388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95296374728177e-05,
+      "loss": 0.0815,
+      "step": 1389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952848148101211e-05,
+      "loss": 35.1162,
+      "step": 1390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95273240771609e-05,
+      "loss": 0.7878,
+      "step": 1391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952616526129709e-05,
+      "loss": 35.3567,
+      "step": 1392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95250050334537e-05,
+      "loss": 32.4961,
+      "step": 1393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952384339366382e-05,
+      "loss": 32.4517,
+      "step": 1394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952268034196056e-05,
+      "loss": 33.7477,
+      "step": 1395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952151587837708e-05,
+      "loss": 35.3157,
+      "step": 1396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95203500029466e-05,
+      "loss": 33.1224,
+      "step": 1397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95191827157023e-05,
+      "loss": 31.2042,
+      "step": 1398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951801401667753e-05,
+      "loss": 31.1307,
+      "step": 1399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951684390590558e-05,
+      "loss": 35.3425,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95156723834198e-05,
+      "loss": 32.7824,
+      "step": 1401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95144994492536e-05,
+      "loss": 31.0063,
+      "step": 1402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95133251034404e-05,
+      "loss": 34.4653,
+      "step": 1403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951214934601371e-05,
+      "loss": 0.8217,
+      "step": 1404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951097217700704e-05,
+      "loss": 34.7447,
+      "step": 1405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950979359645396e-05,
+      "loss": 0.823,
+      "step": 1406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950861360438804e-05,
+      "loss": 30.3425,
+      "step": 1407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950743220084295e-05,
+      "loss": 34.5975,
+      "step": 1408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950624938585237e-05,
+      "loss": 0.8487,
+      "step": 1409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950506515945001e-05,
+      "loss": 30.9493,
+      "step": 1410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950387952166963e-05,
+      "loss": 32.0522,
+      "step": 1411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950269247254504e-05,
+      "loss": 35.0477,
+      "step": 1412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950150401211009e-05,
+      "loss": 31.2013,
+      "step": 1413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950031414039864e-05,
+      "loss": 35.1558,
+      "step": 1414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949912285744465e-05,
+      "loss": 30.4711,
+      "step": 1415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949793016328204e-05,
+      "loss": 32.9248,
+      "step": 1416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949673605794484e-05,
+      "loss": 31.291,
+      "step": 1417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949554054146709e-05,
+      "loss": 30.6942,
+      "step": 1418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949434361388288e-05,
+      "loss": 33.9734,
+      "step": 1419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94931452752263e-05,
+      "loss": 31.1318,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949194552553155e-05,
+      "loss": 0.8265,
+      "step": 1421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949074436483286e-05,
+      "loss": 33.7166,
+      "step": 1422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948954179316439e-05,
+      "loss": 0.8157,
+      "step": 1423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948833781056051e-05,
+      "loss": 33.1416,
+      "step": 1424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94871324170555e-05,
+      "loss": 34.7149,
+      "step": 1425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948592561268372e-05,
+      "loss": 0.8148,
+      "step": 1426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94847173974796e-05,
+      "loss": 31.0566,
+      "step": 1427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948350777147757e-05,
+      "loss": 0.8247,
+      "step": 1428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948229673471214e-05,
+      "loss": 0.7599,
+      "step": 1429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948108428721781e-05,
+      "loss": 0.8174,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947987042902915e-05,
+      "loss": 33.2776,
+      "step": 1431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947865516018078e-05,
+      "loss": 30.9298,
+      "step": 1432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947743848070735e-05,
+      "loss": 34.8035,
+      "step": 1433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947622039064351e-05,
+      "loss": 0.7842,
+      "step": 1434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947500089002404e-05,
+      "loss": 33.4188,
+      "step": 1435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947377997888367e-05,
+      "loss": 31.6099,
+      "step": 1436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947255765725723e-05,
+      "loss": 30.1875,
+      "step": 1437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947133392517954e-05,
+      "loss": 31.3303,
+      "step": 1438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.947010878268552e-05,
+      "loss": 0.7746,
+      "step": 1439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946888222981008e-05,
+      "loss": 0.0621,
+      "step": 1440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94676542665882e-05,
+      "loss": 0.7946,
+      "step": 1441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946642489305488e-05,
+      "loss": 33.8718,
+      "step": 1442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946519410924518e-05,
+      "loss": 31.2664,
+      "step": 1443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946396191519416e-05,
+      "loss": 31.5231,
+      "step": 1444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946272831093698e-05,
+      "loss": 31.3896,
+      "step": 1445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94614932965088e-05,
+      "loss": 0.8013,
+      "step": 1446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946025687194484e-05,
+      "loss": 0.8236,
+      "step": 1447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945901903728035e-05,
+      "loss": 0.8021,
+      "step": 1448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945777979255058e-05,
+      "loss": 38.0546,
+      "step": 1449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945653913779092e-05,
+      "loss": 32.3009,
+      "step": 1450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945529707303669e-05,
+      "loss": 0.7966,
+      "step": 1451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945405359832334e-05,
+      "loss": 0.0754,
+      "step": 1452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945280871368629e-05,
+      "loss": 0.0965,
+      "step": 1453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945156241916106e-05,
+      "loss": 35.2169,
+      "step": 1454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945031471478317e-05,
+      "loss": 34.0205,
+      "step": 1455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944906560058817e-05,
+      "loss": 0.7806,
+      "step": 1456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944781507661171e-05,
+      "loss": 0.7017,
+      "step": 1457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944656314288943e-05,
+      "loss": 0.798,
+      "step": 1458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944530979945698e-05,
+      "loss": 30.5951,
+      "step": 1459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944405504635017e-05,
+      "loss": 0.8626,
+      "step": 1460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944279888360471e-05,
+      "loss": 28.6406,
+      "step": 1461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944154131125642e-05,
+      "loss": 33.1966,
+      "step": 1462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944028232934119e-05,
+      "loss": 0.7895,
+      "step": 1463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94390219378949e-05,
+      "loss": 0.7526,
+      "step": 1464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943776013695345e-05,
+      "loss": 31.3659,
+      "step": 1465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943649692655283e-05,
+      "loss": 0.7972,
+      "step": 1466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943523230672908e-05,
+      "loss": 31.8487,
+      "step": 1467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943396627751824e-05,
+      "loss": 0.7898,
+      "step": 1468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94326988389564e-05,
+      "loss": 0.7896,
+      "step": 1469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943142999107968e-05,
+      "loss": 30.1404,
+      "step": 1470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943015973392428e-05,
+      "loss": 30.8197,
+      "step": 1471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942888806752639e-05,
+      "loss": 0.0634,
+      "step": 1472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942761499192229e-05,
+      "loss": 0.7864,
+      "step": 1473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942634050714825e-05,
+      "loss": 0.7774,
+      "step": 1474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942506461324064e-05,
+      "loss": 34.6191,
+      "step": 1475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94237873102358e-05,
+      "loss": 32.884,
+      "step": 1476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942250859817016e-05,
+      "loss": 32.411,
+      "step": 1477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942122847708017e-05,
+      "loss": 33.1288,
+      "step": 1478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941994694700234e-05,
+      "loss": 30.4712,
+      "step": 1479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94186640079732e-05,
+      "loss": 30.9965,
+      "step": 1480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941737966002931e-05,
+      "loss": 29.6869,
+      "step": 1481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941609390320732e-05,
+      "loss": 31.8737,
+      "step": 1482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941480673754387e-05,
+      "loss": 33.2696,
+      "step": 1483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941351816307563e-05,
+      "loss": 0.7884,
+      "step": 1484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941222817983938e-05,
+      "loss": 32.4656,
+      "step": 1485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941093678787186e-05,
+      "loss": 29.2325,
+      "step": 1486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940964398720991e-05,
+      "loss": 33.3983,
+      "step": 1487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940834977789039e-05,
+      "loss": 0.0831,
+      "step": 1488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940705415995018e-05,
+      "loss": 0.7756,
+      "step": 1489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940575713342624e-05,
+      "loss": 33.0291,
+      "step": 1490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940445869835553e-05,
+      "loss": 0.8111,
+      "step": 1491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940315885477507e-05,
+      "loss": 0.7975,
+      "step": 1492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940185760272194e-05,
+      "loss": 0.7727,
+      "step": 1493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940055494223319e-05,
+      "loss": 34.1094,
+      "step": 1494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939925087334601e-05,
+      "loss": 29.1066,
+      "step": 1495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939794539609756e-05,
+      "loss": 29.7832,
+      "step": 1496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939663851052505e-05,
+      "loss": 29.3033,
+      "step": 1497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939533021666574e-05,
+      "loss": 30.1991,
+      "step": 1498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939402051455694e-05,
+      "loss": 31.6041,
+      "step": 1499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9392709404236e-05,
+      "loss": 0.8622,
+      "step": 1500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939139688574027e-05,
+      "loss": 32.7551,
+      "step": 1501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939008295910716e-05,
+      "loss": 0.7325,
+      "step": 1502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938876762437419e-05,
+      "loss": 30.7456,
+      "step": 1503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938745088157879e-05,
+      "loss": 0.7974,
+      "step": 1504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938613273075857e-05,
+      "loss": 32.1726,
+      "step": 1505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938481317195104e-05,
+      "loss": 0.7538,
+      "step": 1506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938349220519387e-05,
+      "loss": 31.0877,
+      "step": 1507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93821698305247e-05,
+      "loss": 32.0999,
+      "step": 1508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938084604798124e-05,
+      "loss": 0.1003,
+      "step": 1509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93795208576012e-05,
+      "loss": 0.7301,
+      "step": 1510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937819425942241e-05,
+      "loss": 30.967,
+      "step": 1511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937686625348267e-05,
+      "loss": 0.8497,
+      "step": 1512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937553683981983e-05,
+      "loss": 31.4603,
+      "step": 1513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93742060184718e-05,
+      "loss": 28.1911,
+      "step": 1514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937287378947653e-05,
+      "loss": 31.4809,
+      "step": 1515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937154015287198e-05,
+      "loss": 33.7983,
+      "step": 1516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937020510869619e-05,
+      "loss": 30.0506,
+      "step": 1517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93688686569872e-05,
+      "loss": 32.7463,
+      "step": 1518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936753079778315e-05,
+      "loss": 29.3134,
+      "step": 1519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936619153112215e-05,
+      "loss": 27.0177,
+      "step": 1520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93648508570424e-05,
+      "loss": 0.8053,
+      "step": 1521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936350877558209e-05,
+      "loss": 0.7903,
+      "step": 1522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936216528677953e-05,
+      "loss": 0.7758,
+      "step": 1523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936082039067301e-05,
+      "loss": 0.7527,
+      "step": 1524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935947408730084e-05,
+      "loss": 0.7958,
+      "step": 1525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935812637670143e-05,
+      "loss": 36.6848,
+      "step": 1526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935677725891319e-05,
+      "loss": 31.8834,
+      "step": 1527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93554267339746e-05,
+      "loss": 0.8358,
+      "step": 1528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935407480192416e-05,
+      "loss": 0.7862,
+      "step": 1529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93527214628004e-05,
+      "loss": 33.4763,
+      "step": 1530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935136671664191e-05,
+      "loss": 32.4145,
+      "step": 1531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935001056348732e-05,
+      "loss": 33.6918,
+      "step": 1532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934865300337527e-05,
+      "loss": 32.2638,
+      "step": 1533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934729403634449e-05,
+      "loss": 0.0665,
+      "step": 1534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934593366243372e-05,
+      "loss": 30.5961,
+      "step": 1535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934457188168173e-05,
+      "loss": 0.7851,
+      "step": 1536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934320869412737e-05,
+      "loss": 0.7694,
+      "step": 1537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93418440998095e-05,
+      "loss": 31.8524,
+      "step": 1538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934047809876696e-05,
+      "loss": 33.147,
+      "step": 1539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933911069103878e-05,
+      "loss": 0.7934,
+      "step": 1540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933774187666392e-05,
+      "loss": 0.0809,
+      "step": 1541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933637165568137e-05,
+      "loss": 31.8379,
+      "step": 1542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933500002813023e-05,
+      "loss": 0.7613,
+      "step": 1543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93336269940496e-05,
+      "loss": 28.8416,
+      "step": 1544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933225255347862e-05,
+      "loss": 0.7542,
+      "step": 1545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933087670645647e-05,
+      "loss": 0.8196,
+      "step": 1546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93294994530224e-05,
+      "loss": 33.4297,
+      "step": 1547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932812079321562e-05,
+      "loss": 0.0837,
+      "step": 1548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932674072707548e-05,
+      "loss": 0.0929,
+      "step": 1549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932535925464132e-05,
+      "loss": 0.7473,
+      "step": 1550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932397637595253e-05,
+      "loss": 30.4621,
+      "step": 1551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932259209104851e-05,
+      "loss": 32.1594,
+      "step": 1552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932120639996876e-05,
+      "loss": 30.5091,
+      "step": 1553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931981930275275e-05,
+      "loss": 31.8763,
+      "step": 1554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931843079944005e-05,
+      "loss": 30.8669,
+      "step": 1555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931704089007023e-05,
+      "loss": 0.0898,
+      "step": 1556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931564957468295e-05,
+      "loss": 30.816,
+      "step": 1557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931425685331784e-05,
+      "loss": 34.0093,
+      "step": 1558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931286272601461e-05,
+      "loss": 0.7498,
+      "step": 1559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931146719281301e-05,
+      "loss": 0.8,
+      "step": 1560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931007025375286e-05,
+      "loss": 31.3586,
+      "step": 1561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930867190887392e-05,
+      "loss": 29.217,
+      "step": 1562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930727215821612e-05,
+      "loss": 31.899,
+      "step": 1563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930587100181932e-05,
+      "loss": 0.7333,
+      "step": 1564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930446843972349e-05,
+      "loss": 31.8741,
+      "step": 1565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930306447196862e-05,
+      "loss": 34.7696,
+      "step": 1566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930165909859473e-05,
+      "loss": 29.8177,
+      "step": 1567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930025231964188e-05,
+      "loss": 0.0814,
+      "step": 1568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92988441351502e-05,
+      "loss": 0.8167,
+      "step": 1569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92974345451598e-05,
+      "loss": 0.8023,
+      "step": 1570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929602354971091e-05,
+      "loss": 0.7564,
+      "step": 1571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929461114884372e-05,
+      "loss": 34.0465,
+      "step": 1572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92931973425985e-05,
+      "loss": 33.4621,
+      "step": 1573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92917821310156e-05,
+      "loss": 30.3572,
+      "step": 1574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929036551413532e-05,
+      "loss": 0.7648,
+      "step": 1575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928894749199806e-05,
+      "loss": 32.2232,
+      "step": 1576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928752806464426e-05,
+      "loss": 33.2637,
+      "step": 1577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928610723211438e-05,
+      "loss": 0.0851,
+      "step": 1578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928468499444892e-05,
+      "loss": 31.8244,
+      "step": 1579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928326135168844e-05,
+      "loss": 32.8568,
+      "step": 1580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928183630387352e-05,
+      "loss": 0.7766,
+      "step": 1581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928040985104478e-05,
+      "loss": 31.1347,
+      "step": 1582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927898199324293e-05,
+      "loss": 0.7819,
+      "step": 1583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927755273050861e-05,
+      "loss": 30.7136,
+      "step": 1584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927612206288263e-05,
+      "loss": 0.7889,
+      "step": 1585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927468999040574e-05,
+      "loss": 0.0875,
+      "step": 1586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927325651311878e-05,
+      "loss": 0.7797,
+      "step": 1587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927182163106262e-05,
+      "loss": 0.6846,
+      "step": 1588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927038534427816e-05,
+      "loss": 32.652,
+      "step": 1589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926894765280636e-05,
+      "loss": 0.732,
+      "step": 1590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92675085566882e-05,
+      "loss": 31.0064,
+      "step": 1591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926606805596471e-05,
+      "loss": 0.0966,
+      "step": 1592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926462615067697e-05,
+      "loss": 33.4592,
+      "step": 1593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926318284086607e-05,
+      "loss": 32.1152,
+      "step": 1594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926173812657315e-05,
+      "loss": 30.5497,
+      "step": 1595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926029200783944e-05,
+      "loss": 0.7801,
+      "step": 1596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925884448470613e-05,
+      "loss": 0.7537,
+      "step": 1597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925739555721451e-05,
+      "loss": 29.4721,
+      "step": 1598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925594522540586e-05,
+      "loss": 33.5477,
+      "step": 1599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925449348932157e-05,
+      "loss": 33.2041,
+      "step": 1600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9253040349003e-05,
+      "loss": 31.388,
+      "step": 1601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925158580449159e-05,
+      "loss": 0.7534,
+      "step": 1602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92501298558288e-05,
+      "loss": 0.7677,
+      "step": 1603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924867250305615e-05,
+      "loss": 34.4374,
+      "step": 1604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924721374621518e-05,
+      "loss": 29.8912,
+      "step": 1605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924575358534747e-05,
+      "loss": 33.8084,
+      "step": 1606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924429202049467e-05,
+      "loss": 0.7423,
+      "step": 1607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924282905169845e-05,
+      "loss": 32.4196,
+      "step": 1608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924136467900049e-05,
+      "loss": 0.8143,
+      "step": 1609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923989890244257e-05,
+      "loss": 30.2503,
+      "step": 1610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923843172206646e-05,
+      "loss": 0.7447,
+      "step": 1611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9236963137914e-05,
+      "loss": 31.5954,
+      "step": 1612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923549315002705e-05,
+      "loss": 32.7693,
+      "step": 1613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923402175844752e-05,
+      "loss": 0.7726,
+      "step": 1614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923254896321735e-05,
+      "loss": 29.5215,
+      "step": 1615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923107476437857e-05,
+      "loss": 33.0484,
+      "step": 1616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922959916197315e-05,
+      "loss": 32.0898,
+      "step": 1617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922812215604321e-05,
+      "loss": 0.0894,
+      "step": 1618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922664374663082e-05,
+      "loss": 30.3387,
+      "step": 1619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922516393377816e-05,
+      "loss": 0.7977,
+      "step": 1620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92236827175274e-05,
+      "loss": 0.7396,
+      "step": 1621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922220009792078e-05,
+      "loss": 0.1169,
+      "step": 1622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922071607500057e-05,
+      "loss": 35.0437,
+      "step": 1623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921923064880907e-05,
+      "loss": 31.0644,
+      "step": 1624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921774381938863e-05,
+      "loss": 31.3289,
+      "step": 1625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921625558678164e-05,
+      "loss": 0.7608,
+      "step": 1626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921476595103054e-05,
+      "loss": 30.5688,
+      "step": 1627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921327491217778e-05,
+      "loss": 0.7541,
+      "step": 1628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921178247026588e-05,
+      "loss": 0.7452,
+      "step": 1629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92102886253374e-05,
+      "loss": 0.0796,
+      "step": 1630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920879337743492e-05,
+      "loss": 0.8006,
+      "step": 1631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920729672660105e-05,
+      "loss": 36.7335,
+      "step": 1632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92057986728785e-05,
+      "loss": 0.7809,
+      "step": 1633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920429921630993e-05,
+      "loss": 0.7995,
+      "step": 1634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920279835693812e-05,
+      "loss": 0.7871,
+      "step": 1635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920129609480586e-05,
+      "loss": 0.0625,
+      "step": 1636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919979242995597e-05,
+      "loss": 29.9098,
+      "step": 1637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919828736243133e-05,
+      "loss": 0.075,
+      "step": 1638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919678089227484e-05,
+      "loss": 28.9781,
+      "step": 1639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919527301952943e-05,
+      "loss": 0.0831,
+      "step": 1640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919376374423812e-05,
+      "loss": 0.0723,
+      "step": 1641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919225306644393e-05,
+      "loss": 0.795,
+      "step": 1642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919074098618995e-05,
+      "loss": 0.0952,
+      "step": 1643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918922750351924e-05,
+      "loss": 0.0799,
+      "step": 1644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9187712618475e-05,
+      "loss": 35.9681,
+      "step": 1645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918619633110035e-05,
+      "loss": 0.8046,
+      "step": 1646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918467864143861e-05,
+      "loss": 33.6839,
+      "step": 1647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918315954953298e-05,
+      "loss": 28.5463,
+      "step": 1648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918163905542679e-05,
+      "loss": 28.7586,
+      "step": 1649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91801171591634e-05,
+      "loss": 29.8517,
+      "step": 1650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917859386078618e-05,
+      "loss": 31.1995,
+      "step": 1651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917706916033859e-05,
+      "loss": 0.7414,
+      "step": 1652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917554305786406e-05,
+      "loss": 30.0245,
+      "step": 1653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917401555340611e-05,
+      "loss": 0.0559,
+      "step": 1654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917248664700831e-05,
+      "loss": 0.7492,
+      "step": 1655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917095633871422e-05,
+      "loss": 34.6942,
+      "step": 1656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91694246285675e-05,
+      "loss": 31.4298,
+      "step": 1657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916789151661179e-05,
+      "loss": 31.0334,
+      "step": 1658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916635700289081e-05,
+      "loss": 0.7747,
+      "step": 1659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91648210874483e-05,
+      "loss": 32.1741,
+      "step": 1660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916328377032806e-05,
+      "loss": 29.9297,
+      "step": 1661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916174505157394e-05,
+      "loss": 31.2522,
+      "step": 1662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916020493122975e-05,
+      "loss": 0.7483,
+      "step": 1663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915866340933945e-05,
+      "loss": 31.4457,
+      "step": 1664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915712048594697e-05,
+      "loss": 31.3716,
+      "step": 1665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91555761610963e-05,
+      "loss": 31.4505,
+      "step": 1666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915403043483147e-05,
+      "loss": 0.8012,
+      "step": 1667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915248330719654e-05,
+      "loss": 0.785,
+      "step": 1668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915093477823562e-05,
+      "loss": 33.2448,
+      "step": 1669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914938484799288e-05,
+      "loss": 0.7662,
+      "step": 1670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914783351651247e-05,
+      "loss": 0.7967,
+      "step": 1671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914628078383865e-05,
+      "loss": 29.7728,
+      "step": 1672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914472665001568e-05,
+      "loss": 30.3315,
+      "step": 1673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914317111508786e-05,
+      "loss": 0.0786,
+      "step": 1674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914161417909955e-05,
+      "loss": 32.5599,
+      "step": 1675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914005584209511e-05,
+      "loss": 32.2555,
+      "step": 1676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9138496104119e-05,
+      "loss": 33.9527,
+      "step": 1677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913693496521569e-05,
+      "loss": 0.0649,
+      "step": 1678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913537242542965e-05,
+      "loss": 0.7905,
+      "step": 1679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913380848480547e-05,
+      "loss": 0.0604,
+      "step": 1680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913224314338772e-05,
+      "loss": 34.0939,
+      "step": 1681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913067640122103e-05,
+      "loss": 0.7449,
+      "step": 1682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912910825835004e-05,
+      "loss": 31.6504,
+      "step": 1683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91275387148195e-05,
+      "loss": 0.8051,
+      "step": 1684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912596777067413e-05,
+      "loss": 31.2143,
+      "step": 1685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912439542595872e-05,
+      "loss": 0.7702,
+      "step": 1686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912282168071811e-05,
+      "loss": 28.7869,
+      "step": 1687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912124653499717e-05,
+      "loss": 0.0824,
+      "step": 1688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911966998884079e-05,
+      "loss": 0.7575,
+      "step": 1689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911809204229395e-05,
+      "loss": 0.7901,
+      "step": 1690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911651269540156e-05,
+      "loss": 28.0283,
+      "step": 1691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911493194820874e-05,
+      "loss": 28.9736,
+      "step": 1692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91133498007605e-05,
+      "loss": 31.6358,
+      "step": 1693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911176625310196e-05,
+      "loss": 0.0927,
+      "step": 1694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911018130527828e-05,
+      "loss": 28.2472,
+      "step": 1695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910859495733462e-05,
+      "loss": 0.0939,
+      "step": 1696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910700720931623e-05,
+      "loss": 0.8024,
+      "step": 1697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910541806126835e-05,
+      "loss": 34.5015,
+      "step": 1698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910382751323632e-05,
+      "loss": 32.5254,
+      "step": 1699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910223556526547e-05,
+      "loss": 29.4793,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910064221740119e-05,
+      "loss": 30.9669,
+      "step": 1701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909904746968887e-05,
+      "loss": 0.776,
+      "step": 1702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909745132217405e-05,
+      "loss": 35.2334,
+      "step": 1703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909585377490218e-05,
+      "loss": 0.0799,
+      "step": 1704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909425482791881e-05,
+      "loss": 31.514,
+      "step": 1705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909265448126955e-05,
+      "loss": 31.3024,
+      "step": 1706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909105273499999e-05,
+      "loss": 31.7608,
+      "step": 1707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908944958915582e-05,
+      "loss": 26.9841,
+      "step": 1708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908784504378275e-05,
+      "loss": 0.7795,
+      "step": 1709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908623909892651e-05,
+      "loss": 29.7716,
+      "step": 1710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908463175463289e-05,
+      "loss": 0.7953,
+      "step": 1711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908302301094773e-05,
+      "loss": 30.3439,
+      "step": 1712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908141286791688e-05,
+      "loss": 0.7987,
+      "step": 1713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907980132558623e-05,
+      "loss": 0.0827,
+      "step": 1714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907818838400175e-05,
+      "loss": 27.3074,
+      "step": 1715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907657404320941e-05,
+      "loss": 30.8112,
+      "step": 1716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907495830325525e-05,
+      "loss": 30.7421,
+      "step": 1717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907334116418531e-05,
+      "loss": 0.7176,
+      "step": 1718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907172262604573e-05,
+      "loss": 32.6026,
+      "step": 1719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907010268888261e-05,
+      "loss": 0.8274,
+      "step": 1720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906848135274218e-05,
+      "loss": 0.7312,
+      "step": 1721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906685861767062e-05,
+      "loss": 32.4094,
+      "step": 1722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906523448371423e-05,
+      "loss": 31.7448,
+      "step": 1723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906360895091928e-05,
+      "loss": 30.2241,
+      "step": 1724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906198201933215e-05,
+      "loss": 30.4925,
+      "step": 1725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90603536889992e-05,
+      "loss": 0.757,
+      "step": 1726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905872395996684e-05,
+      "loss": 0.1047,
+      "step": 1727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905709283228157e-05,
+      "loss": 0.7458,
+      "step": 1728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905546030598989e-05,
+      "loss": 0.7487,
+      "step": 1729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90538263811383e-05,
+      "loss": 34.0441,
+      "step": 1730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905219105777343e-05,
+      "loss": 0.0778,
+      "step": 1731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905055433594186e-05,
+      "loss": 0.7748,
+      "step": 1732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90489162156903e-05,
+      "loss": 32.8256,
+      "step": 1733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904727669706542e-05,
+      "loss": 30.6492,
+      "step": 1734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904563578011396e-05,
+      "loss": 30.7568,
+      "step": 1735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904399346488273e-05,
+      "loss": 0.7432,
+      "step": 1736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904234975141853e-05,
+      "loss": 31.1588,
+      "step": 1737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90407046397682e-05,
+      "loss": 0.085,
+      "step": 1738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903905812997871e-05,
+      "loss": 0.8194,
+      "step": 1739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903741022209694e-05,
+      "loss": 29.9643,
+      "step": 1740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903576091616988e-05,
+      "loss": 0.7725,
+      "step": 1741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903411021224458e-05,
+      "loss": 0.7578,
+      "step": 1742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903245811036808e-05,
+      "loss": 0.792,
+      "step": 1743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903080461058749e-05,
+      "loss": 32.2028,
+      "step": 1744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902914971294994e-05,
+      "loss": 0.7468,
+      "step": 1745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902749341750262e-05,
+      "loss": 32.562,
+      "step": 1746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902583572429274e-05,
+      "loss": 29.5233,
+      "step": 1747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902417663336757e-05,
+      "loss": 30.3586,
+      "step": 1748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902251614477441e-05,
+      "loss": 30.4896,
+      "step": 1749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90208542585606e-05,
+      "loss": 31.9853,
+      "step": 1750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901919097477352e-05,
+      "loss": 0.7779,
+      "step": 1751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901752629346058e-05,
+      "loss": 0.7538,
+      "step": 1752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901586021466926e-05,
+      "loss": 29.6984,
+      "step": 1753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901419273844703e-05,
+      "loss": 28.571,
+      "step": 1754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901252386484146e-05,
+      "loss": 27.4777,
+      "step": 1755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901085359390011e-05,
+      "loss": 31.3838,
+      "step": 1756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900918192567061e-05,
+      "loss": 30.5271,
+      "step": 1757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900750886020061e-05,
+      "loss": 31.9942,
+      "step": 1758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900583439753782e-05,
+      "loss": 0.7501,
+      "step": 1759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900415853772996e-05,
+      "loss": 0.8332,
+      "step": 1760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900248128082482e-05,
+      "loss": 0.0833,
+      "step": 1761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900080262687022e-05,
+      "loss": 30.3992,
+      "step": 1762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899912257591401e-05,
+      "loss": 28.4932,
+      "step": 1763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899744112800411e-05,
+      "loss": 0.0769,
+      "step": 1764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899575828318842e-05,
+      "loss": 0.0911,
+      "step": 1765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899407404151495e-05,
+      "loss": 30.0312,
+      "step": 1766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89923884030317e-05,
+      "loss": 0.8036,
+      "step": 1767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899070136778674e-05,
+      "loss": 30.2336,
+      "step": 1768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898901293582816e-05,
+      "loss": 31.6689,
+      "step": 1769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89873231072041e-05,
+      "loss": 31.2459,
+      "step": 1770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898563188196273e-05,
+      "loss": 0.7911,
+      "step": 1771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898393926015226e-05,
+      "loss": 28.5742,
+      "step": 1772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898224524182099e-05,
+      "loss": 0.7127,
+      "step": 1773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898054982701714e-05,
+      "loss": 30.4784,
+      "step": 1774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897885301578911e-05,
+      "loss": 0.8303,
+      "step": 1775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897715480818526e-05,
+      "loss": 28.3389,
+      "step": 1776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897545520425398e-05,
+      "loss": 30.0523,
+      "step": 1777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897375420404376e-05,
+      "loss": 32.4505,
+      "step": 1778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897205180760308e-05,
+      "loss": 29.1588,
+      "step": 1779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897034801498046e-05,
+      "loss": 29.4584,
+      "step": 1780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89686428262245e-05,
+      "loss": 0.7703,
+      "step": 1781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89669362413838e-05,
+      "loss": 30.153,
+      "step": 1782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896522826050702e-05,
+      "loss": 0.8257,
+      "step": 1783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896351888364286e-05,
+      "loss": 31.2316,
+      "step": 1784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896180811084003e-05,
+      "loss": 0.8061,
+      "step": 1785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896009594214734e-05,
+      "loss": 0.0793,
+      "step": 1786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895838237761356e-05,
+      "loss": 31.4293,
+      "step": 1787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89566674172876e-05,
+      "loss": 29.4923,
+      "step": 1788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895495106121829e-05,
+      "loss": 0.7649,
+      "step": 1789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895323330945461e-05,
+      "loss": 30.1621,
+      "step": 1790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895151416204551e-05,
+      "loss": 31.2068,
+      "step": 1791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894979361904001e-05,
+      "loss": 28.6276,
+      "step": 1792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894807168048716e-05,
+      "loss": 30.9064,
+      "step": 1793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894634834643606e-05,
+      "loss": 28.4495,
+      "step": 1794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894462361693582e-05,
+      "loss": 30.178,
+      "step": 1795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894289749203562e-05,
+      "loss": 29.5185,
+      "step": 1796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894116997178471e-05,
+      "loss": 0.8269,
+      "step": 1797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893944105623228e-05,
+      "loss": 0.8178,
+      "step": 1798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893771074542766e-05,
+      "loss": 0.7704,
+      "step": 1799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893597903942017e-05,
+      "loss": 0.7557,
+      "step": 1800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893424593825919e-05,
+      "loss": 28.242,
+      "step": 1801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893251144199411e-05,
+      "loss": 0.7842,
+      "step": 1802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89307755506744e-05,
+      "loss": 0.7954,
+      "step": 1803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892903826434952e-05,
+      "loss": 0.7939,
+      "step": 1804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892729958306904e-05,
+      "loss": 31.9385,
+      "step": 1805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892555950688251e-05,
+      "loss": 0.7975,
+      "step": 1806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892381803583955e-05,
+      "loss": 31.647,
+      "step": 1807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892207516998977e-05,
+      "loss": 30.1176,
+      "step": 1808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892033090938291e-05,
+      "loss": 28.8804,
+      "step": 1809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891858525406867e-05,
+      "loss": 0.0843,
+      "step": 1810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891683820409682e-05,
+      "loss": 0.7886,
+      "step": 1811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891508975951717e-05,
+      "loss": 27.5503,
+      "step": 1812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891333992037957e-05,
+      "loss": 30.526,
+      "step": 1813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891158868673391e-05,
+      "loss": 0.0851,
+      "step": 1814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890983605863012e-05,
+      "loss": 29.3194,
+      "step": 1815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890808203611817e-05,
+      "loss": 31.6564,
+      "step": 1816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890632661924804e-05,
+      "loss": 27.8953,
+      "step": 1817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890456980806979e-05,
+      "loss": 28.854,
+      "step": 1818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890281160263352e-05,
+      "loss": 0.8152,
+      "step": 1819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890105200298937e-05,
+      "loss": 0.7477,
+      "step": 1820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.889929100918744e-05,
+      "loss": 28.4573,
+      "step": 1821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.889752862127802e-05,
+      "loss": 30.7302,
+      "step": 1822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889576483931128e-05,
+      "loss": 0.803,
+      "step": 1823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889399966333756e-05,
+      "loss": 27.2035,
+      "step": 1824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889223309340715e-05,
+      "loss": 26.8137,
+      "step": 1825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889046512957044e-05,
+      "loss": 0.7652,
+      "step": 1826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888869577187781e-05,
+      "loss": 27.2732,
+      "step": 1827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888692502037973e-05,
+      "loss": 0.8417,
+      "step": 1828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888515287512665e-05,
+      "loss": 29.1816,
+      "step": 1829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888337933616913e-05,
+      "loss": 30.5908,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888160440355771e-05,
+      "loss": 0.7722,
+      "step": 1831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8879828077343e-05,
+      "loss": 30.2472,
+      "step": 1832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887805035757565e-05,
+      "loss": 0.8371,
+      "step": 1833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887627124430632e-05,
+      "loss": 29.108,
+      "step": 1834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887449073758577e-05,
+      "loss": 0.733,
+      "step": 1835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887270883746472e-05,
+      "loss": 30.8036,
+      "step": 1836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887092554399399e-05,
+      "loss": 0.8055,
+      "step": 1837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886914085722444e-05,
+      "loss": 29.9244,
+      "step": 1838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88673547772069e-05,
+      "loss": 27.5401,
+      "step": 1839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886556730399234e-05,
+      "loss": 28.6821,
+      "step": 1840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88637784376317e-05,
+      "loss": 31.9564,
+      "step": 1841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886198817817599e-05,
+      "loss": 0.7309,
+      "step": 1842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886019652567624e-05,
+      "loss": 0.8315,
+      "step": 1843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885840348018353e-05,
+      "loss": 31.215,
+      "step": 1844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885660904174899e-05,
+      "loss": 0.768,
+      "step": 1845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885481321042376e-05,
+      "loss": 26.2005,
+      "step": 1846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885301598625907e-05,
+      "loss": 0.0753,
+      "step": 1847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885121736930613e-05,
+      "loss": 0.082,
+      "step": 1848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884941735961622e-05,
+      "loss": 26.5187,
+      "step": 1849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884761595724067e-05,
+      "loss": 0.7897,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884581316223082e-05,
+      "loss": 32.1987,
+      "step": 1851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884400897463811e-05,
+      "loss": 29.0525,
+      "step": 1852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88422033945139e-05,
+      "loss": 33.309,
+      "step": 1853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884039642190977e-05,
+      "loss": 24.0778,
+      "step": 1854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883858805687714e-05,
+      "loss": 28.5219,
+      "step": 1855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883677829946761e-05,
+      "loss": 30.9919,
+      "step": 1856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883496714973277e-05,
+      "loss": 0.773,
+      "step": 1857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883315460772427e-05,
+      "loss": 31.4622,
+      "step": 1858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883134067349377e-05,
+      "loss": 29.7424,
+      "step": 1859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882952534709297e-05,
+      "loss": 0.7895,
+      "step": 1860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882770862857366e-05,
+      "loss": 0.0645,
+      "step": 1861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88258905179876e-05,
+      "loss": 0.8142,
+      "step": 1862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882407101538666e-05,
+      "loss": 0.7695,
+      "step": 1863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882225012082269e-05,
+      "loss": 0.8377,
+      "step": 1864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88204278343476e-05,
+      "loss": 0.0855,
+      "step": 1865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881860415601335e-05,
+      "loss": 0.085,
+      "step": 1866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881677908587193e-05,
+      "loss": 38.1658,
+      "step": 1867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881495262397537e-05,
+      "loss": 31.7658,
+      "step": 1868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881312477037574e-05,
+      "loss": 0.0754,
+      "step": 1869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881129552512518e-05,
+      "loss": 29.3138,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88094648882758e-05,
+      "loss": 29.4575,
+      "step": 1871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880763285987982e-05,
+      "loss": 0.7327,
+      "step": 1872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880579943998947e-05,
+      "loss": 32.0361,
+      "step": 1873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8803964628657e-05,
+      "loss": 0.0811,
+      "step": 1874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880212842593473e-05,
+      "loss": 0.7612,
+      "step": 1875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880029083187501e-05,
+      "loss": 28.5769,
+      "step": 1876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879845184653024e-05,
+      "loss": 29.2023,
+      "step": 1877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879661146995283e-05,
+      "loss": 0.7565,
+      "step": 1878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879476970219527e-05,
+      "loss": 28.9101,
+      "step": 1879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879292654331005e-05,
+      "loss": 0.7253,
+      "step": 1880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879108199334971e-05,
+      "loss": 28.2359,
+      "step": 1881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878923605236688e-05,
+      "loss": 0.0857,
+      "step": 1882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878738872041415e-05,
+      "loss": 0.759,
+      "step": 1883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87855399975442e-05,
+      "loss": 33.3593,
+      "step": 1884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878368988380973e-05,
+      "loss": 26.7668,
+      "step": 1885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878183837926348e-05,
+      "loss": 27.9142,
+      "step": 1886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877998548395826e-05,
+      "loss": 0.8205,
+      "step": 1887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877813119794688e-05,
+      "loss": 27.7277,
+      "step": 1888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877627552128219e-05,
+      "loss": 0.7822,
+      "step": 1889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877441845401712e-05,
+      "loss": 0.7064,
+      "step": 1890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877255999620462e-05,
+      "loss": 0.0943,
+      "step": 1891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877070014789764e-05,
+      "loss": 0.7763,
+      "step": 1892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876883890914925e-05,
+      "loss": 27.2898,
+      "step": 1893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876697628001246e-05,
+      "loss": 0.0631,
+      "step": 1894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876511226054042e-05,
+      "loss": 0.7575,
+      "step": 1895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876324685078623e-05,
+      "loss": 0.7874,
+      "step": 1896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876138005080314e-05,
+      "loss": 0.7528,
+      "step": 1897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875951186064429e-05,
+      "loss": 32.7959,
+      "step": 1898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8757642280363e-05,
+      "loss": 29.5844,
+      "step": 1899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875577131001254e-05,
+      "loss": 0.8156,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875389894964626e-05,
+      "loss": 0.76,
+      "step": 1901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875202519931755e-05,
+      "loss": 0.8192,
+      "step": 1902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875015005907981e-05,
+      "loss": 29.8879,
+      "step": 1903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874827352898653e-05,
+      "loss": 27.7566,
+      "step": 1904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874639560909117e-05,
+      "loss": 26.8149,
+      "step": 1905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874451629944732e-05,
+      "loss": 29.6784,
+      "step": 1906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874263560010853e-05,
+      "loss": 27.4292,
+      "step": 1907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874075351112839e-05,
+      "loss": 0.7565,
+      "step": 1908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873887003256059e-05,
+      "loss": 26.6042,
+      "step": 1909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873698516445884e-05,
+      "loss": 0.0712,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873509890687685e-05,
+      "loss": 0.8546,
+      "step": 1911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873321125986842e-05,
+      "loss": 28.9471,
+      "step": 1912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873132222348734e-05,
+      "loss": 23.9034,
+      "step": 1913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872943179778749e-05,
+      "loss": 0.8139,
+      "step": 1914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872753998282274e-05,
+      "loss": 0.8458,
+      "step": 1915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872564677864707e-05,
+      "loss": 29.1063,
+      "step": 1916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87237521853144e-05,
+      "loss": 0.0689,
+      "step": 1917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872185620287877e-05,
+      "loss": 0.0854,
+      "step": 1918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871995883139425e-05,
+      "loss": 29.3733,
+      "step": 1919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87180600709149e-05,
+      "loss": 30.1514,
+      "step": 1920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871615992149488e-05,
+      "loss": 25.7167,
+      "step": 1921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871425838318836e-05,
+      "loss": 0.7753,
+      "step": 1922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871235545604954e-05,
+      "loss": 0.8059,
+      "step": 1923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871045114013267e-05,
+      "loss": 0.0593,
+      "step": 1924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870854543549206e-05,
+      "loss": 0.766,
+      "step": 1925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870663834218202e-05,
+      "loss": 26.9452,
+      "step": 1926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870472986025694e-05,
+      "loss": 0.0931,
+      "step": 1927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870281998977122e-05,
+      "loss": 28.0533,
+      "step": 1928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87009087307793e-05,
+      "loss": 0.8007,
+      "step": 1929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869899608333571e-05,
+      "loss": 29.4149,
+      "step": 1930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869708204749493e-05,
+      "loss": 30.9195,
+      "step": 1931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869516662331155e-05,
+      "loss": 31.0063,
+      "step": 1932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869324981084018e-05,
+      "loss": 0.777,
+      "step": 1933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869133161013548e-05,
+      "loss": 26.6873,
+      "step": 1934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868941202125211e-05,
+      "loss": 0.8157,
+      "step": 1935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86874910442448e-05,
+      "loss": 0.7554,
+      "step": 1936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868556867916834e-05,
+      "loss": 0.0904,
+      "step": 1937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868364492607753e-05,
+      "loss": 0.7661,
+      "step": 1938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86817197850272e-05,
+      "loss": 0.7632,
+      "step": 1939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867979325607223e-05,
+      "loss": 0.8223,
+      "step": 1940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867786533926757e-05,
+      "loss": 0.789,
+      "step": 1941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867593603466818e-05,
+      "loss": 31.4677,
+      "step": 1942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867400534232905e-05,
+      "loss": 0.7559,
+      "step": 1943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867207326230523e-05,
+      "loss": 0.7575,
+      "step": 1944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867013979465181e-05,
+      "loss": 0.7263,
+      "step": 1945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866820493942391e-05,
+      "loss": 0.776,
+      "step": 1946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866626869667668e-05,
+      "loss": 0.1099,
+      "step": 1947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866433106646533e-05,
+      "loss": 0.7771,
+      "step": 1948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866239204884511e-05,
+      "loss": 32.3533,
+      "step": 1949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86604516438713e-05,
+      "loss": 29.6279,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865850985159919e-05,
+      "loss": 28.5697,
+      "step": 1951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865656667208419e-05,
+      "loss": 29.3149,
+      "step": 1952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865462210538166e-05,
+      "loss": 29.9714,
+      "step": 1953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865267615154705e-05,
+      "loss": 28.7773,
+      "step": 1954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865072881063584e-05,
+      "loss": 0.0938,
+      "step": 1955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864878008270355e-05,
+      "loss": 28.1198,
+      "step": 1956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864682996780575e-05,
+      "loss": 0.7094,
+      "step": 1957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864487846599801e-05,
+      "loss": 27.0631,
+      "step": 1958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864292557733599e-05,
+      "loss": 0.791,
+      "step": 1959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864097130187535e-05,
+      "loss": 0.7304,
+      "step": 1960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863901563967181e-05,
+      "loss": 26.6108,
+      "step": 1961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863705859078115e-05,
+      "loss": 0.8264,
+      "step": 1962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863510015525915e-05,
+      "loss": 0.7645,
+      "step": 1963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86331403331616e-05,
+      "loss": 0.7466,
+      "step": 1964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863117912454445e-05,
+      "loss": 30.8648,
+      "step": 1965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862921652946356e-05,
+      "loss": 0.0854,
+      "step": 1966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862725254797489e-05,
+      "loss": 29.4238,
+      "step": 1967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862528718013446e-05,
+      "loss": 0.7773,
+      "step": 1968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862332042599828e-05,
+      "loss": 27.091,
+      "step": 1969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862135228562242e-05,
+      "loss": 0.7516,
+      "step": 1970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8619382759063e-05,
+      "loss": 27.1684,
+      "step": 1971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861741184637617e-05,
+      "loss": 28.8928,
+      "step": 1972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861543954761811e-05,
+      "loss": 27.2785,
+      "step": 1973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861346586284507e-05,
+      "loss": 0.7646,
+      "step": 1974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861149079211331e-05,
+      "loss": 0.0781,
+      "step": 1975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860951433547913e-05,
+      "loss": 27.1812,
+      "step": 1976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860753649299888e-05,
+      "loss": 0.0798,
+      "step": 1977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860555726472896e-05,
+      "loss": 28.9198,
+      "step": 1978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860357665072579e-05,
+      "loss": 28.3353,
+      "step": 1979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860159465104584e-05,
+      "loss": 29.466,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859961126574559e-05,
+      "loss": 28.3327,
+      "step": 1981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859762649488164e-05,
+      "loss": 0.0805,
+      "step": 1982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859564033851053e-05,
+      "loss": 29.3553,
+      "step": 1983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85936527966889e-05,
+      "loss": 0.7724,
+      "step": 1984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859166386947341e-05,
+      "loss": 28.9337,
+      "step": 1985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858967355692077e-05,
+      "loss": 0.7731,
+      "step": 1986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858768185908773e-05,
+      "loss": 0.1025,
+      "step": 1987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858568877603105e-05,
+      "loss": 28.5358,
+      "step": 1988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858369430780758e-05,
+      "loss": 28.1834,
+      "step": 1989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858169845447416e-05,
+      "loss": 0.8158,
+      "step": 1990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857970121608768e-05,
+      "loss": 26.5583,
+      "step": 1991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857770259270514e-05,
+      "loss": 29.713,
+      "step": 1992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857570258438345e-05,
+      "loss": 30.8973,
+      "step": 1993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857370119117966e-05,
+      "loss": 0.8319,
+      "step": 1994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857169841315084e-05,
+      "loss": 0.7418,
+      "step": 1995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856969425035408e-05,
+      "loss": 0.7699,
+      "step": 1996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856768870284651e-05,
+      "loss": 0.7317,
+      "step": 1997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85656817706853e-05,
+      "loss": 28.9465,
+      "step": 1998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856367345392771e-05,
+      "loss": 30.1459,
+      "step": 1999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856166375263095e-05,
+      "loss": 27.1493,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855965266685234e-05,
+      "loss": 30.1862,
+      "step": 2001
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85576401966492e-05,
+      "loss": 0.0671,
+      "step": 2002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855562634207893e-05,
+      "loss": 30.2643,
+      "step": 2003
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855361110319892e-05,
+      "loss": 0.8247,
+      "step": 2004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855159448006664e-05,
+      "loss": 28.9495,
+      "step": 2005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854957647273957e-05,
+      "loss": 26.1262,
+      "step": 2006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854755708127524e-05,
+      "loss": 0.7804,
+      "step": 2007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854553630573124e-05,
+      "loss": 30.4607,
+      "step": 2008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854351414616519e-05,
+      "loss": 29.9712,
+      "step": 2009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85414906026347e-05,
+      "loss": 29.7696,
+      "step": 2010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85394656751975e-05,
+      "loss": 0.1042,
+      "step": 2011
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853743936391128e-05,
+      "loss": 0.7724,
+      "step": 2012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853541166883386e-05,
+      "loss": 28.6075,
+      "step": 2013
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853338259002303e-05,
+      "loss": 28.3255,
+      "step": 2014
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853135212753662e-05,
+      "loss": 0.0591,
+      "step": 2015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852932028143254e-05,
+      "loss": 30.4881,
+      "step": 2016
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852728705176871e-05,
+      "loss": 27.725,
+      "step": 2017
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852525243860308e-05,
+      "loss": 0.7535,
+      "step": 2018
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852321644199369e-05,
+      "loss": 0.8014,
+      "step": 2019
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852117906199856e-05,
+      "loss": 28.0005,
+      "step": 2020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851914029867578e-05,
+      "loss": 0.7726,
+      "step": 2021
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851710015208349e-05,
+      "loss": 29.2281,
+      "step": 2022
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851505862227983e-05,
+      "loss": 27.5366,
+      "step": 2023
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851301570932301e-05,
+      "loss": 0.7738,
+      "step": 2024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851097141327129e-05,
+      "loss": 28.7316,
+      "step": 2025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850892573418293e-05,
+      "loss": 27.9035,
+      "step": 2026
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850687867211628e-05,
+      "loss": 0.7317,
+      "step": 2027
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850483022712966e-05,
+      "loss": 0.1,
+      "step": 2028
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850278039928152e-05,
+      "loss": 27.972,
+      "step": 2029
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850072918863026e-05,
+      "loss": 0.7832,
+      "step": 2030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849867659523438e-05,
+      "loss": 30.8887,
+      "step": 2031
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849662261915239e-05,
+      "loss": 0.747,
+      "step": 2032
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849456726044285e-05,
+      "loss": 0.753,
+      "step": 2033
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849251051916436e-05,
+      "loss": 0.8219,
+      "step": 2034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849045239537556e-05,
+      "loss": 25.7462,
+      "step": 2035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848839288913512e-05,
+      "loss": 0.6907,
+      "step": 2036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848633200050176e-05,
+      "loss": 27.2833,
+      "step": 2037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848426972953424e-05,
+      "loss": 29.4131,
+      "step": 2038
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848220607629134e-05,
+      "loss": 0.744,
+      "step": 2039
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84801410408319e-05,
+      "loss": 31.0463,
+      "step": 2040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847807462321481e-05,
+      "loss": 0.7835,
+      "step": 2041
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847600682349897e-05,
+      "loss": 28.2363,
+      "step": 2042
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847393764174333e-05,
+      "loss": 29.7002,
+      "step": 2043
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847186707800689e-05,
+      "loss": 28.8361,
+      "step": 2044
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846979513234866e-05,
+      "loss": 0.7754,
+      "step": 2045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846772180482775e-05,
+      "loss": 29.7025,
+      "step": 2046
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846564709550324e-05,
+      "loss": 27.6573,
+      "step": 2047
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846357100443427e-05,
+      "loss": 28.577,
+      "step": 2048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846149353168006e-05,
+      "loss": 28.3585,
+      "step": 2049
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845941467729982e-05,
+      "loss": 28.1332,
+      "step": 2050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845733444135281e-05,
+      "loss": 28.632,
+      "step": 2051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845525282389836e-05,
+      "loss": 0.702,
+      "step": 2052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845316982499582e-05,
+      "loss": 0.0751,
+      "step": 2053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845108544470454e-05,
+      "loss": 26.7242,
+      "step": 2054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844899968308397e-05,
+      "loss": 0.7069,
+      "step": 2055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844691254019358e-05,
+      "loss": 0.7415,
+      "step": 2056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844482401609285e-05,
+      "loss": 0.0754,
+      "step": 2057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844273411084135e-05,
+      "loss": 27.0913,
+      "step": 2058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844064282449865e-05,
+      "loss": 28.4868,
+      "step": 2059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843855015712436e-05,
+      "loss": 28.3718,
+      "step": 2060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843645610877816e-05,
+      "loss": 26.1875,
+      "step": 2061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843436067951975e-05,
+      "loss": 29.7298,
+      "step": 2062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843226386940888e-05,
+      "loss": 0.0653,
+      "step": 2063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843016567850528e-05,
+      "loss": 0.7807,
+      "step": 2064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842806610686883e-05,
+      "loss": 0.755,
+      "step": 2065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842596515455935e-05,
+      "loss": 31.1114,
+      "step": 2066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842386282163675e-05,
+      "loss": 0.7969,
+      "step": 2067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842175910816096e-05,
+      "loss": 31.0299,
+      "step": 2068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841965401419197e-05,
+      "loss": 30.9995,
+      "step": 2069
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84175475397898e-05,
+      "loss": 31.2587,
+      "step": 2070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841543968501447e-05,
+      "loss": 25.5666,
+      "step": 2071
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841333044992614e-05,
+      "loss": 0.7344,
+      "step": 2072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841121983458485e-05,
+      "loss": 0.7377,
+      "step": 2073
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840910783905087e-05,
+      "loss": 28.6669,
+      "step": 2074
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840699446338434e-05,
+      "loss": 0.7981,
+      "step": 2075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840487970764554e-05,
+      "loss": 0.765,
+      "step": 2076
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840276357189477e-05,
+      "loss": 28.0149,
+      "step": 2077
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840064605619236e-05,
+      "loss": 0.7194,
+      "step": 2078
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839852716059865e-05,
+      "loss": 28.3029,
+      "step": 2079
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839640688517408e-05,
+      "loss": 0.0791,
+      "step": 2080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83942852299791e-05,
+      "loss": 0.8228,
+      "step": 2081
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839216219507417e-05,
+      "loss": 27.6089,
+      "step": 2082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839003778051985e-05,
+      "loss": 26.2777,
+      "step": 2083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83879119863767e-05,
+      "loss": 0.0746,
+      "step": 2084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838578481270529e-05,
+      "loss": 26.8697,
+      "step": 2085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83836562595663e-05,
+      "loss": 0.0688,
+      "step": 2086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83815263270204e-05,
+      "loss": 29.5097,
+      "step": 2087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837939501512834e-05,
+      "loss": 30.2776,
+      "step": 2088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837726232395083e-05,
+      "loss": 28.1069,
+      "step": 2089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837512825354874e-05,
+      "loss": 27.0927,
+      "step": 2090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837299280398286e-05,
+      "loss": 28.3726,
+      "step": 2091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83708559753141e-05,
+      "loss": 28.9579,
+      "step": 2092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836871776760337e-05,
+      "loss": 26.8086,
+      "step": 2093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836657818091162e-05,
+      "loss": 28.1291,
+      "step": 2094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836443721529986e-05,
+      "loss": 27.3631,
+      "step": 2095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836229487082913e-05,
+      "loss": 28.4169,
+      "step": 2096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83601511475605e-05,
+      "loss": 27.4756,
+      "step": 2097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83580060455551e-05,
+      "loss": 25.3787,
+      "step": 2098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835585956487407e-05,
+      "loss": 28.2372,
+      "step": 2099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835371170557864e-05,
+      "loss": 0.7806,
+      "step": 2100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835156246772998e-05,
+      "loss": 0.7943,
+      "step": 2101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834941185138942e-05,
+      "loss": 0.0808,
+      "step": 2102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834725985661825e-05,
+      "loss": 0.7831,
+      "step": 2103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834510648347785e-05,
+      "loss": 27.9322,
+      "step": 2104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834295173202957e-05,
+      "loss": 0.8225,
+      "step": 2105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834079560233486e-05,
+      "loss": 29.8936,
+      "step": 2106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833863809445522e-05,
+      "loss": 28.3724,
+      "step": 2107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833647920845212e-05,
+      "loss": 0.815,
+      "step": 2108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833431894438712e-05,
+      "loss": 27.8995,
+      "step": 2109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833215730232181e-05,
+      "loss": 0.0597,
+      "step": 2110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832999428231783e-05,
+      "loss": 25.7977,
+      "step": 2111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832782988443683e-05,
+      "loss": 30.0631,
+      "step": 2112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832566410874054e-05,
+      "loss": 0.7488,
+      "step": 2113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832349695529067e-05,
+      "loss": 28.7418,
+      "step": 2114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832132842414901e-05,
+      "loss": 28.919,
+      "step": 2115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831915851537742e-05,
+      "loss": 0.7668,
+      "step": 2116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831698722903775e-05,
+      "loss": 0.7732,
+      "step": 2117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831481456519188e-05,
+      "loss": 28.1052,
+      "step": 2118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831264052390177e-05,
+      "loss": 0.7336,
+      "step": 2119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831046510522941e-05,
+      "loss": 0.7669,
+      "step": 2120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830828830923679e-05,
+      "loss": 28.3282,
+      "step": 2121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8306110135986e-05,
+      "loss": 0.7166,
+      "step": 2122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830393058553913e-05,
+      "loss": 27.841,
+      "step": 2123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830174965795832e-05,
+      "loss": 27.0146,
+      "step": 2124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829956735330575e-05,
+      "loss": 24.7929,
+      "step": 2125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829738367164361e-05,
+      "loss": 0.78,
+      "step": 2126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829519861303419e-05,
+      "loss": 0.7829,
+      "step": 2127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829301217753977e-05,
+      "loss": 0.0667,
+      "step": 2128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82908243652227e-05,
+      "loss": 26.6107,
+      "step": 2129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828863517614533e-05,
+      "loss": 30.7272,
+      "step": 2130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828644461037009e-05,
+      "loss": 28.6459,
+      "step": 2131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828425266795943e-05,
+      "loss": 31.9369,
+      "step": 2132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828205934897583e-05,
+      "loss": 0.7904,
+      "step": 2133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827986465348184e-05,
+      "loss": 0.0716,
+      "step": 2134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827766858154003e-05,
+      "loss": 0.8081,
+      "step": 2135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827547113321299e-05,
+      "loss": 30.8062,
+      "step": 2136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827327230856338e-05,
+      "loss": 30.2467,
+      "step": 2137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82710721076539e-05,
+      "loss": 0.837,
+      "step": 2138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826887053054724e-05,
+      "loss": 0.0771,
+      "step": 2139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82666675773062e-05,
+      "loss": 25.7831,
+      "step": 2140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826446324799359e-05,
+      "loss": 0.076,
+      "step": 2141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826225754267224e-05,
+      "loss": 27.7938,
+      "step": 2142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826005046140502e-05,
+      "loss": 26.4494,
+      "step": 2143
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825784200425487e-05,
+      "loss": 27.0265,
+      "step": 2144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825563217128478e-05,
+      "loss": 0.7511,
+      "step": 2145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82534209625577e-05,
+      "loss": 26.47,
+      "step": 2146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825120837813669e-05,
+      "loss": 26.0767,
+      "step": 2147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824899441808484e-05,
+      "loss": 0.7814,
+      "step": 2148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824677908246527e-05,
+      "loss": 27.4386,
+      "step": 2149
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824456237134115e-05,
+      "loss": 26.8139,
+      "step": 2150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824234428477563e-05,
+      "loss": 26.392,
+      "step": 2151
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8240124822832e-05,
+      "loss": 24.9561,
+      "step": 2152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823790398557349e-05,
+      "loss": 26.7587,
+      "step": 2153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823568177306346e-05,
+      "loss": 29.6528,
+      "step": 2154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823345818536525e-05,
+      "loss": 0.7868,
+      "step": 2155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823123322254225e-05,
+      "loss": 0.7945,
+      "step": 2156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822900688465785e-05,
+      "loss": 0.7385,
+      "step": 2157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822677917177561e-05,
+      "loss": 28.0535,
+      "step": 2158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822455008395899e-05,
+      "loss": 0.7482,
+      "step": 2159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822231962127153e-05,
+      "loss": 0.0746,
+      "step": 2160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822008778377685e-05,
+      "loss": 26.5574,
+      "step": 2161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821785457153856e-05,
+      "loss": 0.7047,
+      "step": 2162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821561998462034e-05,
+      "loss": 0.7387,
+      "step": 2163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821338402308589e-05,
+      "loss": 0.8514,
+      "step": 2164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821114668699896e-05,
+      "loss": 0.0775,
+      "step": 2165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820890797642334e-05,
+      "loss": 0.7602,
+      "step": 2166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820666789142283e-05,
+      "loss": 0.7909,
+      "step": 2167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820442643206133e-05,
+      "loss": 0.7676,
+      "step": 2168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820218359840274e-05,
+      "loss": 0.732,
+      "step": 2169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819993939051096e-05,
+      "loss": 0.0782,
+      "step": 2170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819769380845002e-05,
+      "loss": 28.576,
+      "step": 2171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819544685228393e-05,
+      "loss": 0.7432,
+      "step": 2172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819319852207672e-05,
+      "loss": 0.7624,
+      "step": 2173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819094881789252e-05,
+      "loss": 28.5736,
+      "step": 2174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818869773979548e-05,
+      "loss": 28.6554,
+      "step": 2175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818644528784975e-05,
+      "loss": 0.7837,
+      "step": 2176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818419146211955e-05,
+      "loss": 0.7139,
+      "step": 2177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818193626266913e-05,
+      "loss": 28.8516,
+      "step": 2178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817967968956282e-05,
+      "loss": 29.6889,
+      "step": 2179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817742174286493e-05,
+      "loss": 26.8321,
+      "step": 2180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817516242263982e-05,
+      "loss": 0.7626,
+      "step": 2181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817290172895193e-05,
+      "loss": 27.8966,
+      "step": 2182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817063966186569e-05,
+      "loss": 0.7854,
+      "step": 2183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816837622144561e-05,
+      "loss": 29.4429,
+      "step": 2184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81661114077562e-05,
+      "loss": 29.2623,
+      "step": 2185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816384522086205e-05,
+      "loss": 0.7495,
+      "step": 2186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816157766082775e-05,
+      "loss": 26.8777,
+      "step": 2187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815930872771797e-05,
+      "loss": 0.736,
+      "step": 2188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815703842159737e-05,
+      "loss": 27.2967,
+      "step": 2189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81547667425307e-05,
+      "loss": 0.7713,
+      "step": 2190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815249369058271e-05,
+      "loss": 0.8179,
+      "step": 2191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81502192658182e-05,
+      "loss": 27.1367,
+      "step": 2192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814794346830203e-05,
+      "loss": 25.8832,
+      "step": 2193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814566629809907e-05,
+      "loss": 26.5305,
+      "step": 2194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814338775527424e-05,
+      "loss": 26.295,
+      "step": 2195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814110783989251e-05,
+      "loss": 25.1452,
+      "step": 2196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813882655201889e-05,
+      "loss": 26.385,
+      "step": 2197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813654389171838e-05,
+      "loss": 27.7272,
+      "step": 2198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81342598590561e-05,
+      "loss": 28.1948,
+      "step": 2199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813197445409714e-05,
+      "loss": 0.0753,
+      "step": 2200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812968767690669e-05,
+      "loss": 25.7313,
+      "step": 2201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81273995275499e-05,
+      "loss": 29.1963,
+      "step": 2202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812511000609203e-05,
+      "loss": 0.806,
+      "step": 2203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812281911259835e-05,
+      "loss": 28.5754,
+      "step": 2204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812052684713417e-05,
+      "loss": 28.0766,
+      "step": 2205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811823320976487e-05,
+      "loss": 25.7069,
+      "step": 2206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.81159382005558e-05,
+      "loss": 24.1032,
+      "step": 2207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.81136418195724e-05,
+      "loss": 24.5496,
+      "step": 2208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811134406688015e-05,
+      "loss": 29.076,
+      "step": 2209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810904494254456e-05,
+      "loss": 28.6123,
+      "step": 2210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810674444663117e-05,
+      "loss": 0.7893,
+      "step": 2211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810444257920556e-05,
+      "loss": 0.114,
+      "step": 2212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810213934033338e-05,
+      "loss": 0.0801,
+      "step": 2213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809983473008026e-05,
+      "loss": 28.4777,
+      "step": 2214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809752874851194e-05,
+      "loss": 27.4314,
+      "step": 2215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809522139569414e-05,
+      "loss": 0.7789,
+      "step": 2216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809291267169265e-05,
+      "loss": 27.1755,
+      "step": 2217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809060257657328e-05,
+      "loss": 0.8037,
+      "step": 2218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808829111040191e-05,
+      "loss": 0.7756,
+      "step": 2219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808597827324441e-05,
+      "loss": 26.628,
+      "step": 2220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808366406516676e-05,
+      "loss": 25.3022,
+      "step": 2221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808134848623492e-05,
+      "loss": 28.0143,
+      "step": 2222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807903153651488e-05,
+      "loss": 0.8063,
+      "step": 2223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807671321607272e-05,
+      "loss": 0.7841,
+      "step": 2224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807439352497454e-05,
+      "loss": 0.0863,
+      "step": 2225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807207246328646e-05,
+      "loss": 0.8162,
+      "step": 2226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806975003107467e-05,
+      "loss": 0.8043,
+      "step": 2227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806742622840536e-05,
+      "loss": 0.7682,
+      "step": 2228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806510105534479e-05,
+      "loss": 0.8226,
+      "step": 2229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806277451195927e-05,
+      "loss": 34.7353,
+      "step": 2230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80604465983151e-05,
+      "loss": 28.8429,
+      "step": 2231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805811731447865e-05,
+      "loss": 26.9308,
+      "step": 2232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805578666051633e-05,
+      "loss": 26.1224,
+      "step": 2233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805345463649462e-05,
+      "loss": 0.6813,
+      "step": 2234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805112124247998e-05,
+      "loss": 0.0706,
+      "step": 2235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80487864785389e-05,
+      "loss": 29.8854,
+      "step": 2236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804645034473801e-05,
+      "loss": 0.7911,
+      "step": 2237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804411284114386e-05,
+      "loss": 29.4186,
+      "step": 2238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804177396782312e-05,
+      "loss": 30.9603,
+      "step": 2239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803943372484246e-05,
+      "loss": 0.7568,
+      "step": 2240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803709211226859e-05,
+      "loss": 0.709,
+      "step": 2241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80347491301683e-05,
+      "loss": 0.0696,
+      "step": 2242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803240477860835e-05,
+      "loss": 0.7851,
+      "step": 2243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803005905765561e-05,
+      "loss": 0.7527,
+      "step": 2244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802771196737694e-05,
+      "loss": 30.225,
+      "step": 2245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802536350783925e-05,
+      "loss": 0.7199,
+      "step": 2246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802301367910951e-05,
+      "loss": 0.7667,
+      "step": 2247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80206624812547e-05,
+      "loss": 29.5765,
+      "step": 2248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801830991434184e-05,
+      "loss": 0.0508,
+      "step": 2249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801595597843803e-05,
+      "loss": 30.0605,
+      "step": 2250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801360067361038e-05,
+      "loss": 28.3143,
+      "step": 2251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801124399992601e-05,
+      "loss": 0.7529,
+      "step": 2252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800888595745212e-05,
+      "loss": 0.7831,
+      "step": 2253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800652654625595e-05,
+      "loss": 30.4136,
+      "step": 2254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800416576640475e-05,
+      "loss": 0.8468,
+      "step": 2255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800180361796583e-05,
+      "loss": 0.7237,
+      "step": 2256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799944010100656e-05,
+      "loss": 31.6028,
+      "step": 2257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799707521559428e-05,
+      "loss": 0.7515,
+      "step": 2258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799470896179644e-05,
+      "loss": 0.7838,
+      "step": 2259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79923413396805e-05,
+      "loss": 0.8059,
+      "step": 2260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798997234931395e-05,
+      "loss": 0.6953,
+      "step": 2261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798760199076433e-05,
+      "loss": 27.3275,
+      "step": 2262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798523026409923e-05,
+      "loss": 27.7104,
+      "step": 2263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798285716938625e-05,
+      "loss": 24.1354,
+      "step": 2264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798048270669306e-05,
+      "loss": 0.0868,
+      "step": 2265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797810687608736e-05,
+      "loss": 27.5092,
+      "step": 2266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797572967763686e-05,
+      "loss": 31.0933,
+      "step": 2267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797335111140936e-05,
+      "loss": 0.085,
+      "step": 2268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797097117747266e-05,
+      "loss": 27.1652,
+      "step": 2269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796858987589461e-05,
+      "loss": 0.753,
+      "step": 2270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796620720674311e-05,
+      "loss": 28.5597,
+      "step": 2271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796382317008607e-05,
+      "loss": 27.0925,
+      "step": 2272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79614377659915e-05,
+      "loss": 0.8021,
+      "step": 2273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795905099452736e-05,
+      "loss": 26.0493,
+      "step": 2274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79566628557617e-05,
+      "loss": 0.7698,
+      "step": 2275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795427334976264e-05,
+      "loss": 25.4766,
+      "step": 2276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795188247659828e-05,
+      "loss": 28.0142,
+      "step": 2277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794949023633679e-05,
+      "loss": 23.8952,
+      "step": 2278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794709662904636e-05,
+      "loss": 26.8977,
+      "step": 2279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794470165479526e-05,
+      "loss": 0.7249,
+      "step": 2280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794230531365174e-05,
+      "loss": 25.6056,
+      "step": 2281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793990760568413e-05,
+      "loss": 0.7607,
+      "step": 2282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793750853096077e-05,
+      "loss": 0.8259,
+      "step": 2283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79351080895501e-05,
+      "loss": 0.7747,
+      "step": 2284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793270628152053e-05,
+      "loss": 0.0864,
+      "step": 2285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793030310694054e-05,
+      "loss": 0.7534,
+      "step": 2286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792789856587863e-05,
+      "loss": 28.7264,
+      "step": 2287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792549265840337e-05,
+      "loss": 26.6722,
+      "step": 2288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792308538458332e-05,
+      "loss": 26.5692,
+      "step": 2289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792067674448716e-05,
+      "loss": 25.5776,
+      "step": 2290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791826673818354e-05,
+      "loss": 30.9254,
+      "step": 2291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791585536574115e-05,
+      "loss": 25.8446,
+      "step": 2292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791344262722877e-05,
+      "loss": 0.715,
+      "step": 2293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791102852271515e-05,
+      "loss": 0.7794,
+      "step": 2294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790861305226914e-05,
+      "loss": 0.0644,
+      "step": 2295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79061962159596e-05,
+      "loss": 27.3946,
+      "step": 2296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790377801385544e-05,
+      "loss": 25.6041,
+      "step": 2297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79013584460256e-05,
+      "loss": 0.823,
+      "step": 2298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789893751253904e-05,
+      "loss": 0.7602,
+      "step": 2299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789651521346479e-05,
+      "loss": 0.7312,
+      "step": 2300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789409154887193e-05,
+      "loss": 27.5159,
+      "step": 2301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789166651882955e-05,
+      "loss": 0.7714,
+      "step": 2302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788924012340677e-05,
+      "loss": 29.2644,
+      "step": 2303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788681236267277e-05,
+      "loss": 25.8887,
+      "step": 2304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788438323669677e-05,
+      "loss": 26.184,
+      "step": 2305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788195274554804e-05,
+      "loss": 0.0732,
+      "step": 2306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787952088929586e-05,
+      "loss": 0.809,
+      "step": 2307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787708766800956e-05,
+      "loss": 30.4693,
+      "step": 2308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787465308175849e-05,
+      "loss": 25.5813,
+      "step": 2309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78722171306121e-05,
+      "loss": 25.9319,
+      "step": 2310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78697798146398e-05,
+      "loss": 25.9066,
+      "step": 2311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786734113391112e-05,
+      "loss": 27.728,
+      "step": 2312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786490108849555e-05,
+      "loss": 0.7344,
+      "step": 2313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786245967846266e-05,
+      "loss": 0.7549,
+      "step": 2314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786001690388206e-05,
+      "loss": 0.8167,
+      "step": 2315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785757276482341e-05,
+      "loss": 0.081,
+      "step": 2316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785512726135638e-05,
+      "loss": 30.9768,
+      "step": 2317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785268039355066e-05,
+      "loss": 32.3297,
+      "step": 2318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785023216147606e-05,
+      "loss": 31.3784,
+      "step": 2319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784778256520234e-05,
+      "loss": 29.615,
+      "step": 2320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784533160479936e-05,
+      "loss": 28.959,
+      "step": 2321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784287928033698e-05,
+      "loss": 0.8137,
+      "step": 2322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784042559188514e-05,
+      "loss": 28.8198,
+      "step": 2323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783797053951375e-05,
+      "loss": 0.7685,
+      "step": 2324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783551412329286e-05,
+      "loss": 26.6559,
+      "step": 2325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783305634329245e-05,
+      "loss": 28.3758,
+      "step": 2326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783059719958264e-05,
+      "loss": 29.66,
+      "step": 2327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782813669223349e-05,
+      "loss": 0.7203,
+      "step": 2328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782567482131517e-05,
+      "loss": 0.8156,
+      "step": 2329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782321158689788e-05,
+      "loss": 0.7007,
+      "step": 2330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782074698905184e-05,
+      "loss": 0.0878,
+      "step": 2331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781828102784729e-05,
+      "loss": 24.9176,
+      "step": 2332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781581370335457e-05,
+      "loss": 0.0906,
+      "step": 2333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7813345015644e-05,
+      "loss": 0.0595,
+      "step": 2334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781087496478598e-05,
+      "loss": 0.8052,
+      "step": 2335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780840355085091e-05,
+      "loss": 26.1806,
+      "step": 2336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780593077390924e-05,
+      "loss": 27.5659,
+      "step": 2337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780345663403152e-05,
+      "loss": 0.8156,
+      "step": 2338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780098113128824e-05,
+      "loss": 27.3913,
+      "step": 2339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779850426575e-05,
+      "loss": 0.7914,
+      "step": 2340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779602603748739e-05,
+      "loss": 0.7328,
+      "step": 2341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779354644657109e-05,
+      "loss": 0.0869,
+      "step": 2342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779106549307178e-05,
+      "loss": 26.412,
+      "step": 2343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778858317706018e-05,
+      "loss": 26.3824,
+      "step": 2344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778609949860708e-05,
+      "loss": 27.0939,
+      "step": 2345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77836144577833e-05,
+      "loss": 0.8173,
+      "step": 2346
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778112805465964e-05,
+      "loss": 25.6208,
+      "step": 2347
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777864028930705e-05,
+      "loss": 0.0717,
+      "step": 2348
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77761511617964e-05,
+      "loss": 26.3806,
+      "step": 2349
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777366067219869e-05,
+      "loss": 0.7637,
+      "step": 2350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77711688205849e-05,
+      "loss": 0.7448,
+      "step": 2351
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776867560702607e-05,
+      "loss": 25.7818,
+      "step": 2352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776618103159332e-05,
+      "loss": 29.5325,
+      "step": 2353
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776368509435771e-05,
+      "loss": 0.8311,
+      "step": 2354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776118779539046e-05,
+      "loss": 26.9641,
+      "step": 2355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775868913476273e-05,
+      "loss": 0.7789,
+      "step": 2356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775618911254578e-05,
+      "loss": 26.4968,
+      "step": 2357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775368772881085e-05,
+      "loss": 27.5425,
+      "step": 2358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775118498362928e-05,
+      "loss": 0.7332,
+      "step": 2359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774868087707242e-05,
+      "loss": 0.817,
+      "step": 2360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774617540921166e-05,
+      "loss": 0.7217,
+      "step": 2361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774366858011844e-05,
+      "loss": 26.9873,
+      "step": 2362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774116038986419e-05,
+      "loss": 0.0981,
+      "step": 2363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773865083852045e-05,
+      "loss": 0.7809,
+      "step": 2364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773613992615878e-05,
+      "loss": 0.7803,
+      "step": 2365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773362765285075e-05,
+      "loss": 25.4338,
+      "step": 2366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773111401866796e-05,
+      "loss": 24.9854,
+      "step": 2367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772859902368211e-05,
+      "loss": 0.7415,
+      "step": 2368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772608266796489e-05,
+      "loss": 0.7802,
+      "step": 2369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772356495158803e-05,
+      "loss": 24.853,
+      "step": 2370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772104587462332e-05,
+      "loss": 0.7098,
+      "step": 2371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771852543714257e-05,
+      "loss": 0.7456,
+      "step": 2372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771600363921767e-05,
+      "loss": 24.5277,
+      "step": 2373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771348048092046e-05,
+      "loss": 26.1807,
+      "step": 2374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771095596232292e-05,
+      "loss": 25.0904,
+      "step": 2375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7708430083497e-05,
+      "loss": 26.2529,
+      "step": 2376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770590284451471e-05,
+      "loss": 25.5135,
+      "step": 2377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770337424544812e-05,
+      "loss": 0.7181,
+      "step": 2378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770084428636932e-05,
+      "loss": 0.8088,
+      "step": 2379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769831296735041e-05,
+      "loss": 0.7483,
+      "step": 2380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76957802884636e-05,
+      "loss": 0.7724,
+      "step": 2381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769324624978104e-05,
+      "loss": 0.0954,
+      "step": 2382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769071085137502e-05,
+      "loss": 0.7702,
+      "step": 2383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768817409331784e-05,
+      "loss": 0.6917,
+      "step": 2384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768563597568174e-05,
+      "loss": 29.8144,
+      "step": 2385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768309649853917e-05,
+      "loss": 0.7029,
+      "step": 2386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768055566196249e-05,
+      "loss": 28.9447,
+      "step": 2387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767801346602415e-05,
+      "loss": 0.7797,
+      "step": 2388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76754699107966e-05,
+      "loss": 27.0297,
+      "step": 2389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767292499635238e-05,
+      "loss": 29.6907,
+      "step": 2390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767037872276404e-05,
+      "loss": 0.0864,
+      "step": 2391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766783109010418e-05,
+      "loss": 29.7648,
+      "step": 2392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766528209844543e-05,
+      "loss": 28.031,
+      "step": 2393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766273174786045e-05,
+      "loss": 0.7724,
+      "step": 2394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766018003842197e-05,
+      "loss": 29.419,
+      "step": 2395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765762697020273e-05,
+      "loss": 27.0437,
+      "step": 2396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76550725432755e-05,
+      "loss": 24.217,
+      "step": 2397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765251675771312e-05,
+      "loss": 26.3186,
+      "step": 2398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764995961358847e-05,
+      "loss": 0.8057,
+      "step": 2399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764740111097444e-05,
+      "loss": 29.8028,
+      "step": 2400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764484124994395e-05,
+      "loss": 26.154,
+      "step": 2401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764228003057002e-05,
+      "loss": 0.7096,
+      "step": 2402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763971745292565e-05,
+      "loss": 0.7504,
+      "step": 2403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763715351708389e-05,
+      "loss": 23.8673,
+      "step": 2404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763458822311786e-05,
+      "loss": 26.5038,
+      "step": 2405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763202157110068e-05,
+      "loss": 24.9868,
+      "step": 2406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762945356110553e-05,
+      "loss": 0.0753,
+      "step": 2407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762688419320561e-05,
+      "loss": 26.4064,
+      "step": 2408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76243134674742e-05,
+      "loss": 27.4227,
+      "step": 2409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762174138398456e-05,
+      "loss": 26.1052,
+      "step": 2410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761916794281005e-05,
+      "loss": 28.3618,
+      "step": 2411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7616593144024e-05,
+      "loss": 25.7201,
+      "step": 2412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761401698769988e-05,
+      "loss": 26.5448,
+      "step": 2413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761143947391105e-05,
+      "loss": 29.045,
+      "step": 2414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760886060273106e-05,
+      "loss": 0.0937,
+      "step": 2415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76062803742334e-05,
+      "loss": 0.8219,
+      "step": 2416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760369878849166e-05,
+      "loss": 26.0908,
+      "step": 2417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760111584557943e-05,
+      "loss": 0.7913,
+      "step": 2418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759853154557033e-05,
+      "loss": 0.8016,
+      "step": 2419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759594588853806e-05,
+      "loss": 24.923,
+      "step": 2420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759335887455632e-05,
+      "loss": 28.4051,
+      "step": 2421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759077050369887e-05,
+      "loss": 0.79,
+      "step": 2422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758818077603951e-05,
+      "loss": 26.4084,
+      "step": 2423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758558969165207e-05,
+      "loss": 26.629,
+      "step": 2424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758299725061042e-05,
+      "loss": 26.1713,
+      "step": 2425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758040345298846e-05,
+      "loss": 25.157,
+      "step": 2426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757780829886017e-05,
+      "loss": 0.0741,
+      "step": 2427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757521178829952e-05,
+      "loss": 26.8284,
+      "step": 2428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757261392138051e-05,
+      "loss": 26.9655,
+      "step": 2429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757001469817723e-05,
+      "loss": 0.794,
+      "step": 2430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75674141187638e-05,
+      "loss": 0.8015,
+      "step": 2431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756481218321432e-05,
+      "loss": 0.8245,
+      "step": 2432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756220889160302e-05,
+      "loss": 24.9317,
+      "step": 2433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755960424400405e-05,
+      "loss": 0.0781,
+      "step": 2434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755699824049173e-05,
+      "loss": 0.7636,
+      "step": 2435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755439088114036e-05,
+      "loss": 26.6976,
+      "step": 2436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755178216602423e-05,
+      "loss": 0.8017,
+      "step": 2437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754917209521773e-05,
+      "loss": 25.121,
+      "step": 2438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754656066879529e-05,
+      "loss": 0.8004,
+      "step": 2439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754394788683135e-05,
+      "loss": 29.3306,
+      "step": 2440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754133374940037e-05,
+      "loss": 24.0274,
+      "step": 2441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753871825657695e-05,
+      "loss": 0.0852,
+      "step": 2442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753610140843559e-05,
+      "loss": 25.0117,
+      "step": 2443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753348320505091e-05,
+      "loss": 0.7718,
+      "step": 2444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753086364649759e-05,
+      "loss": 27.7775,
+      "step": 2445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752824273285028e-05,
+      "loss": 0.7607,
+      "step": 2446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752562046418369e-05,
+      "loss": 0.769,
+      "step": 2447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752299684057262e-05,
+      "loss": 25.1061,
+      "step": 2448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752037186209185e-05,
+      "loss": 0.7875,
+      "step": 2449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751774552881621e-05,
+      "loss": 0.8415,
+      "step": 2450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751511784082057e-05,
+      "loss": 0.8351,
+      "step": 2451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751248879817987e-05,
+      "loss": 25.8369,
+      "step": 2452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750985840096904e-05,
+      "loss": 26.9645,
+      "step": 2453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75072266492631e-05,
+      "loss": 0.8144,
+      "step": 2454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750459354313705e-05,
+      "loss": 0.0665,
+      "step": 2455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750195908266599e-05,
+      "loss": 25.8427,
+      "step": 2456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7499323267925e-05,
+      "loss": 0.7873,
+      "step": 2457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749668609898922e-05,
+      "loss": 0.8516,
+      "step": 2458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749404757593387e-05,
+      "loss": 26.5362,
+      "step": 2459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749140769883413e-05,
+      "loss": 24.3865,
+      "step": 2460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748876646776533e-05,
+      "loss": 25.7964,
+      "step": 2461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748612388280269e-05,
+      "loss": 26.6587,
+      "step": 2462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74834799440216e-05,
+      "loss": 0.707,
+      "step": 2463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748083465149745e-05,
+      "loss": 0.055,
+      "step": 2464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74781880053056e-05,
+      "loss": 26.7878,
+      "step": 2465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747554000552155e-05,
+      "loss": 0.758,
+      "step": 2466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747289065222079e-05,
+      "loss": 0.0795,
+      "step": 2467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747023994547884e-05,
+      "loss": 26.5583,
+      "step": 2468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746758788537126e-05,
+      "loss": 0.8218,
+      "step": 2469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746493447197369e-05,
+      "loss": 0.0673,
+      "step": 2470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746227970536176e-05,
+      "loss": 0.0884,
+      "step": 2471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745962358561116e-05,
+      "loss": 0.7653,
+      "step": 2472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745696611279761e-05,
+      "loss": 24.8241,
+      "step": 2473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745430728699689e-05,
+      "loss": 29.3458,
+      "step": 2474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745164710828479e-05,
+      "loss": 29.1167,
+      "step": 2475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744898557673714e-05,
+      "loss": 25.0861,
+      "step": 2476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744632269242984e-05,
+      "loss": 0.7506,
+      "step": 2477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74436584554388e-05,
+      "loss": 0.7662,
+      "step": 2478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744099286583999e-05,
+      "loss": 0.7494,
+      "step": 2479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743832592370938e-05,
+      "loss": 23.8589,
+      "step": 2480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743565762912301e-05,
+      "loss": 0.7098,
+      "step": 2481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743298798215698e-05,
+      "loss": 28.1505,
+      "step": 2482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743031698288736e-05,
+      "loss": 26.3043,
+      "step": 2483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742764463139034e-05,
+      "loss": 27.3457,
+      "step": 2484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742497092774208e-05,
+      "loss": 0.0881,
+      "step": 2485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74222958720188e-05,
+      "loss": 26.4219,
+      "step": 2486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741961946429679e-05,
+      "loss": 24.3888,
+      "step": 2487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741694170465234e-05,
+      "loss": 0.0759,
+      "step": 2488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74142625931618e-05,
+      "loss": 0.0754,
+      "step": 2489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741158212990155e-05,
+      "loss": 28.5621,
+      "step": 2490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7408900314948e-05,
+      "loss": 29.857,
+      "step": 2491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740621714837762e-05,
+      "loss": 0.0687,
+      "step": 2492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740353263026691e-05,
+      "loss": 28.8563,
+      "step": 2493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740084676069236e-05,
+      "loss": 0.7366,
+      "step": 2494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739815953973062e-05,
+      "loss": 28.2202,
+      "step": 2495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739547096745824e-05,
+      "loss": 28.0477,
+      "step": 2496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739278104395189e-05,
+      "loss": 25.8894,
+      "step": 2497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739008976928828e-05,
+      "loss": 23.7256,
+      "step": 2498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73873971435441e-05,
+      "loss": 0.7937,
+      "step": 2499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738470316679613e-05,
+      "loss": 29.1072,
+      "step": 2500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73820078391212e-05,
+      "loss": 26.051,
+      "step": 2501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737931116059611e-05,
+      "loss": 0.7376,
+      "step": 2502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737661313129778e-05,
+      "loss": 0.7422,
+      "step": 2503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737391375130311e-05,
+      "loss": 0.7234,
+      "step": 2504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737121302068906e-05,
+      "loss": 27.7691,
+      "step": 2505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736851093953263e-05,
+      "loss": 0.7454,
+      "step": 2506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736580750791087e-05,
+      "loss": 28.6699,
+      "step": 2507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736310272590081e-05,
+      "loss": 28.2148,
+      "step": 2508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736039659357962e-05,
+      "loss": 25.7401,
+      "step": 2509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735768911102441e-05,
+      "loss": 0.7423,
+      "step": 2510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73549802783124e-05,
+      "loss": 27.3657,
+      "step": 2511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735227009552079e-05,
+      "loss": 26.3248,
+      "step": 2512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734955856272685e-05,
+      "loss": 25.5652,
+      "step": 2513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734684568000791e-05,
+      "loss": 0.7264,
+      "step": 2514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734413144744129e-05,
+      "loss": 0.7809,
+      "step": 2515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734141586510439e-05,
+      "loss": 0.7657,
+      "step": 2516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73386989330746e-05,
+      "loss": 0.8108,
+      "step": 2517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733598065142942e-05,
+      "loss": 0.7324,
+      "step": 2518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733326102024631e-05,
+      "loss": 27.3769,
+      "step": 2519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733054003960283e-05,
+      "loss": 0.0999,
+      "step": 2520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732781770957654e-05,
+      "loss": 26.6835,
+      "step": 2521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732509403024506e-05,
+      "loss": 0.0619,
+      "step": 2522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732236900168606e-05,
+      "loss": 27.3531,
+      "step": 2523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731964262397719e-05,
+      "loss": 26.1341,
+      "step": 2524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731691489719623e-05,
+      "loss": 23.9565,
+      "step": 2525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731418582142088e-05,
+      "loss": 0.121,
+      "step": 2526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731145539672901e-05,
+      "loss": 0.7005,
+      "step": 2527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730872362319843e-05,
+      "loss": 0.7274,
+      "step": 2528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730599050090703e-05,
+      "loss": 0.8246,
+      "step": 2529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730325602993271e-05,
+      "loss": 26.9899,
+      "step": 2530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730052021035348e-05,
+      "loss": 0.7786,
+      "step": 2531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729778304224728e-05,
+      "loss": 0.7675,
+      "step": 2532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729504452569218e-05,
+      "loss": 0.7625,
+      "step": 2533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729230466076625e-05,
+      "loss": 25.523,
+      "step": 2534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728956344754761e-05,
+      "loss": 0.767,
+      "step": 2535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728682088611438e-05,
+      "loss": 25.3619,
+      "step": 2536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728407697654478e-05,
+      "loss": 0.7999,
+      "step": 2537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728133171891703e-05,
+      "loss": 24.1421,
+      "step": 2538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727858511330938e-05,
+      "loss": 0.7944,
+      "step": 2539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727583715980017e-05,
+      "loss": 24.1608,
+      "step": 2540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727308785846771e-05,
+      "loss": 25.7188,
+      "step": 2541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727033720939041e-05,
+      "loss": 0.7843,
+      "step": 2542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726758521264667e-05,
+      "loss": 0.7516,
+      "step": 2543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726483186831496e-05,
+      "loss": 27.974,
+      "step": 2544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726207717647377e-05,
+      "loss": 0.798,
+      "step": 2545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725932113720163e-05,
+      "loss": 23.2635,
+      "step": 2546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725656375057712e-05,
+      "loss": 29.1325,
+      "step": 2547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725380501667887e-05,
+      "loss": 0.7815,
+      "step": 2548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72510449355855e-05,
+      "loss": 0.703,
+      "step": 2549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724828350737573e-05,
+      "loss": 0.0799,
+      "step": 2550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724552073212827e-05,
+      "loss": 26.7008,
+      "step": 2551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724275660992188e-05,
+      "loss": 26.7933,
+      "step": 2552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723999114083537e-05,
+      "loss": 0.7374,
+      "step": 2553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723722432494759e-05,
+      "loss": 25.6201,
+      "step": 2554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723445616233742e-05,
+      "loss": 0.7799,
+      "step": 2555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723168665308378e-05,
+      "loss": 24.8662,
+      "step": 2556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72289157972656e-05,
+      "loss": 0.0781,
+      "step": 2557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722614359496194e-05,
+      "loss": 0.7393,
+      "step": 2558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722337004625177e-05,
+      "loss": 28.1693,
+      "step": 2559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72205951512142e-05,
+      "loss": 26.0736,
+      "step": 2560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721781890992831e-05,
+      "loss": 0.7152,
+      "step": 2561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72150413224733e-05,
+      "loss": 0.7685,
+      "step": 2562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721226238892832e-05,
+      "loss": 0.0742,
+      "step": 2563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720948210937259e-05,
+      "loss": 0.7397,
+      "step": 2564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720670048388542e-05,
+      "loss": 0.0824,
+      "step": 2565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720391751254607e-05,
+      "loss": 27.9006,
+      "step": 2566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72011331954339e-05,
+      "loss": 27.6138,
+      "step": 2567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719834753262827e-05,
+      "loss": 27.4064,
+      "step": 2568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719556052420864e-05,
+      "loss": 26.8989,
+      "step": 2569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719277217025441e-05,
+      "loss": 28.6538,
+      "step": 2570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718998247084514e-05,
+      "loss": 0.0666,
+      "step": 2571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718719142606032e-05,
+      "loss": 25.3037,
+      "step": 2572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718439903597954e-05,
+      "loss": 0.0805,
+      "step": 2573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718160530068238e-05,
+      "loss": 28.3192,
+      "step": 2574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717881022024856e-05,
+      "loss": 0.8006,
+      "step": 2575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717601379475767e-05,
+      "loss": 24.658,
+      "step": 2576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717321602428951e-05,
+      "loss": 25.2104,
+      "step": 2577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717041690892381e-05,
+      "loss": 27.6142,
+      "step": 2578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71676164487404e-05,
+      "loss": 25.829,
+      "step": 2579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716481464381908e-05,
+      "loss": 26.6679,
+      "step": 2580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716201149423975e-05,
+      "loss": 0.7903,
+      "step": 2581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715920700008234e-05,
+      "loss": 0.7512,
+      "step": 2582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71564011614268e-05,
+      "loss": 0.8087,
+      "step": 2583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715359397835308e-05,
+      "loss": 0.0803,
+      "step": 2584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715078545094128e-05,
+      "loss": 0.7691,
+      "step": 2585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714797557927143e-05,
+      "loss": 28.3018,
+      "step": 2586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714516436342366e-05,
+      "loss": 0.8154,
+      "step": 2587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71423518034781e-05,
+      "loss": 24.3196,
+      "step": 2588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.713953789951493e-05,
+      "loss": 26.2899,
+      "step": 2589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71367226516144e-05,
+      "loss": 25.6858,
+      "step": 2590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713390605985673e-05,
+      "loss": 0.7221,
+      "step": 2591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713108812432228e-05,
+      "loss": 24.6856,
+      "step": 2592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712826884509134e-05,
+      "loss": 27.343,
+      "step": 2593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71254482222443e-05,
+      "loss": 26.7835,
+      "step": 2594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712262625586158e-05,
+      "loss": 23.8772,
+      "step": 2595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711980294602362e-05,
+      "loss": 0.7207,
+      "step": 2596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711697829281095e-05,
+      "loss": 24.5527,
+      "step": 2597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711415229630406e-05,
+      "loss": 0.0876,
+      "step": 2598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711132495658353e-05,
+      "loss": 27.8585,
+      "step": 2599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710849627372998e-05,
+      "loss": 25.361,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710566624782404e-05,
+      "loss": 0.0675,
+      "step": 2601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710283487894638e-05,
+      "loss": 0.7927,
+      "step": 2602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710000216717776e-05,
+      "loss": 0.7153,
+      "step": 2603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709716811259891e-05,
+      "loss": 0.7631,
+      "step": 2604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709433271529065e-05,
+      "loss": 26.8499,
+      "step": 2605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709149597533378e-05,
+      "loss": 0.8014,
+      "step": 2606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708865789280922e-05,
+      "loss": 0.0448,
+      "step": 2607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708581846779785e-05,
+      "loss": 0.0634,
+      "step": 2608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708297770038063e-05,
+      "loss": 0.7801,
+      "step": 2609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708013559063858e-05,
+      "loss": 0.0697,
+      "step": 2610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707729213865267e-05,
+      "loss": 0.0876,
+      "step": 2611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707444734450402e-05,
+      "loss": 25.6666,
+      "step": 2612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707160120827368e-05,
+      "loss": 0.0805,
+      "step": 2613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706875373004284e-05,
+      "loss": 26.5264,
+      "step": 2614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706590490989268e-05,
+      "loss": 28.3371,
+      "step": 2615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706305474790439e-05,
+      "loss": 0.8002,
+      "step": 2616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706020324415924e-05,
+      "loss": 27.8741,
+      "step": 2617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705735039873854e-05,
+      "loss": 27.18,
+      "step": 2618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70544962117236e-05,
+      "loss": 25.0203,
+      "step": 2619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705164068319582e-05,
+      "loss": 22.9535,
+      "step": 2620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704878381323656e-05,
+      "loss": 24.3743,
+      "step": 2621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704592560192733e-05,
+      "loss": 27.6209,
+      "step": 2622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704306604934958e-05,
+      "loss": 0.689,
+      "step": 2623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704020515558485e-05,
+      "loss": 24.9019,
+      "step": 2624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703734292071469e-05,
+      "loss": 0.0761,
+      "step": 2625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703447934482072e-05,
+      "loss": 0.7563,
+      "step": 2626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703161442798457e-05,
+      "loss": 0.7553,
+      "step": 2627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702874817028791e-05,
+      "loss": 0.0782,
+      "step": 2628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702588057181246e-05,
+      "loss": 0.7768,
+      "step": 2629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702301163263998e-05,
+      "loss": 27.424,
+      "step": 2630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702014135285226e-05,
+      "loss": 26.3532,
+      "step": 2631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701726973253116e-05,
+      "loss": 26.0946,
+      "step": 2632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70143967717585e-05,
+      "loss": 0.8508,
+      "step": 2633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70115224706162e-05,
+      "loss": 0.7641,
+      "step": 2634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700864682918622e-05,
+      "loss": 27.443,
+      "step": 2635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700576984755056e-05,
+      "loss": 0.7626,
+      "step": 2636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700289152579121e-05,
+      "loss": 0.099,
+      "step": 2637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700001186399023e-05,
+      "loss": 31.0653,
+      "step": 2638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699713086222974e-05,
+      "loss": 27.0792,
+      "step": 2639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699424852059186e-05,
+      "loss": 0.0732,
+      "step": 2640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699136483915877e-05,
+      "loss": 27.3574,
+      "step": 2641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698847981801269e-05,
+      "loss": 0.084,
+      "step": 2642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698559345723586e-05,
+      "loss": 27.1515,
+      "step": 2643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698270575691058e-05,
+      "loss": 0.7659,
+      "step": 2644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697981671711918e-05,
+      "loss": 25.9824,
+      "step": 2645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6976926337944e-05,
+      "loss": 0.7121,
+      "step": 2646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697403461946749e-05,
+      "loss": 0.8213,
+      "step": 2647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697114156177203e-05,
+      "loss": 27.221,
+      "step": 2648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696824716494018e-05,
+      "loss": 0.8174,
+      "step": 2649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696535142905437e-05,
+      "loss": 0.7583,
+      "step": 2650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696245435419722e-05,
+      "loss": 0.762,
+      "step": 2651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695955594045131e-05,
+      "loss": 0.0647,
+      "step": 2652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695665618789926e-05,
+      "loss": 0.7657,
+      "step": 2653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695375509662376e-05,
+      "loss": 0.7569,
+      "step": 2654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695085266670749e-05,
+      "loss": 0.0731,
+      "step": 2655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694794889823324e-05,
+      "loss": 26.2473,
+      "step": 2656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694504379128376e-05,
+      "loss": 25.5686,
+      "step": 2657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694213734594189e-05,
+      "loss": 29.417,
+      "step": 2658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693922956229048e-05,
+      "loss": 0.7831,
+      "step": 2659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693632044041245e-05,
+      "loss": 0.7925,
+      "step": 2660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693340998039073e-05,
+      "loss": 0.7582,
+      "step": 2661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69304981823083e-05,
+      "loss": 28.6764,
+      "step": 2662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692758504624814e-05,
+      "loss": 0.7751,
+      "step": 2663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692467057229335e-05,
+      "loss": 24.8854,
+      "step": 2664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6921754760527e-05,
+      "loss": 28.3531,
+      "step": 2665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691883761103225e-05,
+      "loss": 25.5435,
+      "step": 2666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691591912389219e-05,
+      "loss": 0.766,
+      "step": 2667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69129992991901e-05,
+      "loss": 0.0801,
+      "step": 2668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691007813700921e-05,
+      "loss": 0.079,
+      "step": 2669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690715563743276e-05,
+      "loss": 25.9139,
+      "step": 2670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690423180054412e-05,
+      "loss": 0.6601,
+      "step": 2671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690130662642663e-05,
+      "loss": 30.3323,
+      "step": 2672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689838011516366e-05,
+      "loss": 26.7167,
+      "step": 2673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68954522668387e-05,
+      "loss": 0.7846,
+      "step": 2674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689252308153515e-05,
+      "loss": 0.777,
+      "step": 2675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68895925593366e-05,
+      "loss": 0.762,
+      "step": 2676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688666070032654e-05,
+      "loss": 26.4427,
+      "step": 2677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688372750458857e-05,
+      "loss": 28.8765,
+      "step": 2678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688079297220634e-05,
+      "loss": 0.7442,
+      "step": 2679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687785710326347e-05,
+      "loss": 0.0919,
+      "step": 2680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68749198978437e-05,
+      "loss": 28.7692,
+      "step": 2681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687198135603074e-05,
+      "loss": 25.1548,
+      "step": 2682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686904147790839e-05,
+      "loss": 0.8339,
+      "step": 2683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686610026356044e-05,
+      "loss": 25.6864,
+      "step": 2684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686315771307078e-05,
+      "loss": 27.8729,
+      "step": 2685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686021382652326e-05,
+      "loss": 0.7743,
+      "step": 2686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685726860400185e-05,
+      "loss": 0.7676,
+      "step": 2687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685432204559049e-05,
+      "loss": 0.8136,
+      "step": 2688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685137415137317e-05,
+      "loss": 0.7026,
+      "step": 2689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6848424921434e-05,
+      "loss": 26.2933,
+      "step": 2690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684547435585697e-05,
+      "loss": 25.0641,
+      "step": 2691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68425224547263e-05,
+      "loss": 0.082,
+      "step": 2692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683956921812606e-05,
+      "loss": 27.3079,
+      "step": 2693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68366146461405e-05,
+      "loss": 26.6142,
+      "step": 2694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683365873885382e-05,
+      "loss": 25.8142,
+      "step": 2695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683070149635031e-05,
+      "loss": 25.534,
+      "step": 2696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68277429187143e-05,
+      "loss": 0.7841,
+      "step": 2697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682478300603011e-05,
+      "loss": 30.0322,
+      "step": 2698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682182175838213e-05,
+      "loss": 26.8469,
+      "step": 2699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681885917585478e-05,
+      "loss": 25.8384,
+      "step": 2700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681589525853254e-05,
+      "loss": 26.7766,
+      "step": 2701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681293000649989e-05,
+      "loss": 25.5233,
+      "step": 2702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68099634198414e-05,
+      "loss": 26.5281,
+      "step": 2703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680699549864162e-05,
+      "loss": 24.251,
+      "step": 2704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680402624298516e-05,
+      "loss": 25.9585,
+      "step": 2705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68010556529567e-05,
+      "loss": 25.4394,
+      "step": 2706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679808372864091e-05,
+      "loss": 0.8316,
+      "step": 2707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679511047012254e-05,
+      "loss": 23.7415,
+      "step": 2708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679213587748631e-05,
+      "loss": 23.5196,
+      "step": 2709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678915995081707e-05,
+      "loss": 0.7646,
+      "step": 2710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678618269019966e-05,
+      "loss": 0.6967,
+      "step": 2711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678320409571893e-05,
+      "loss": 26.9107,
+      "step": 2712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678022416745984e-05,
+      "loss": 24.9719,
+      "step": 2713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677724290550731e-05,
+      "loss": 27.075,
+      "step": 2714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677426030994636e-05,
+      "loss": 0.8162,
+      "step": 2715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677127638086202e-05,
+      "loss": 0.7883,
+      "step": 2716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676829111833935e-05,
+      "loss": 26.1241,
+      "step": 2717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676530452246347e-05,
+      "loss": 0.0915,
+      "step": 2718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676231659331952e-05,
+      "loss": 0.8415,
+      "step": 2719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675932733099266e-05,
+      "loss": 27.606,
+      "step": 2720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675633673556817e-05,
+      "loss": 0.0921,
+      "step": 2721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675334480713128e-05,
+      "loss": 24.4537,
+      "step": 2722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675035154576728e-05,
+      "loss": 0.7227,
+      "step": 2723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674735695156153e-05,
+      "loss": 0.7683,
+      "step": 2724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674436102459938e-05,
+      "loss": 0.0768,
+      "step": 2725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674136376496627e-05,
+      "loss": 0.7349,
+      "step": 2726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673836517274764e-05,
+      "loss": 24.3395,
+      "step": 2727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673536524802895e-05,
+      "loss": 0.0887,
+      "step": 2728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673236399089578e-05,
+      "loss": 26.8804,
+      "step": 2729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672936140143366e-05,
+      "loss": 0.7523,
+      "step": 2730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67263574797282e-05,
+      "loss": 0.7398,
+      "step": 2731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672335222586505e-05,
+      "loss": 26.4617,
+      "step": 2732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672034563992988e-05,
+      "loss": 0.7699,
+      "step": 2733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67173377220084e-05,
+      "loss": 25.5533,
+      "step": 2734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671432847218638e-05,
+      "loss": 24.8174,
+      "step": 2735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671131789054961e-05,
+      "loss": 25.96,
+      "step": 2736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670830597718392e-05,
+      "loss": 0.7741,
+      "step": 2737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670529273217517e-05,
+      "loss": 25.6331,
+      "step": 2738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670227815560929e-05,
+      "loss": 25.2017,
+      "step": 2739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66992622475722e-05,
+      "loss": 26.7507,
+      "step": 2740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669624500814989e-05,
+      "loss": 25.6228,
+      "step": 2741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669322643742837e-05,
+      "loss": 25.8671,
+      "step": 2742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669020653549373e-05,
+      "loss": 25.3088,
+      "step": 2743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668718530243205e-05,
+      "loss": 0.741,
+      "step": 2744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668416273832946e-05,
+      "loss": 0.7551,
+      "step": 2745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668113884327212e-05,
+      "loss": 26.405,
+      "step": 2746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667811361734627e-05,
+      "loss": 24.833,
+      "step": 2747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667508706063815e-05,
+      "loss": 0.7001,
+      "step": 2748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667205917323403e-05,
+      "loss": 24.4839,
+      "step": 2749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666902995522025e-05,
+      "loss": 27.1285,
+      "step": 2750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666599940668317e-05,
+      "loss": 0.7902,
+      "step": 2751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666296752770919e-05,
+      "loss": 0.0921,
+      "step": 2752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665993431838474e-05,
+      "loss": 23.7769,
+      "step": 2753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665689977879629e-05,
+      "loss": 25.3465,
+      "step": 2754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665386390903038e-05,
+      "loss": 22.1721,
+      "step": 2755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665082670917356e-05,
+      "loss": 0.7824,
+      "step": 2756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66477881793124e-05,
+      "loss": 0.0899,
+      "step": 2757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664474831953351e-05,
+      "loss": 26.3165,
+      "step": 2758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664170712992363e-05,
+      "loss": 24.2325,
+      "step": 2759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663866461056936e-05,
+      "loss": 22.1442,
+      "step": 2760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663562076155754e-05,
+      "loss": 25.7958,
+      "step": 2761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66325755829749e-05,
+      "loss": 0.7911,
+      "step": 2762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662952907490825e-05,
+      "loss": 0.8078,
+      "step": 2763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662648123744446e-05,
+      "loss": 24.4788,
+      "step": 2764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662343207067042e-05,
+      "loss": 0.7664,
+      "step": 2765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662038157467307e-05,
+      "loss": 25.0089,
+      "step": 2766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661732974953936e-05,
+      "loss": 27.2805,
+      "step": 2767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661427659535634e-05,
+      "loss": 26.1637,
+      "step": 2768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661122211221099e-05,
+      "loss": 25.0589,
+      "step": 2769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660816630019045e-05,
+      "loss": 26.746,
+      "step": 2770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660510915938178e-05,
+      "loss": 26.5902,
+      "step": 2771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660205068987223e-05,
+      "loss": 27.3336,
+      "step": 2772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65989908917489e-05,
+      "loss": 0.7382,
+      "step": 2773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659592976509908e-05,
+      "loss": 25.0694,
+      "step": 2774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659286731001002e-05,
+      "loss": 0.8051,
+      "step": 2775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658980352656903e-05,
+      "loss": 0.0783,
+      "step": 2776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658673841486349e-05,
+      "loss": 0.8192,
+      "step": 2777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658367197498075e-05,
+      "loss": 26.3191,
+      "step": 2778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658060420700822e-05,
+      "loss": 23.8937,
+      "step": 2779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657753511103343e-05,
+      "loss": 0.7355,
+      "step": 2780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65744646871438e-05,
+      "loss": 0.7371,
+      "step": 2781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.657139293542692e-05,
+      "loss": 0.0955,
+      "step": 2782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656831985597034e-05,
+      "loss": 24.6902,
+      "step": 2783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656524544886167e-05,
+      "loss": 26.2119,
+      "step": 2784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656216971418859e-05,
+      "loss": 24.7697,
+      "step": 2785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655909265203875e-05,
+      "loss": 0.7979,
+      "step": 2786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65560142624999e-05,
+      "loss": 21.8444,
+      "step": 2787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655293454565982e-05,
+      "loss": 0.8215,
+      "step": 2788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654985350160626e-05,
+      "loss": 28.209,
+      "step": 2789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654677113042711e-05,
+      "loss": 0.7452,
+      "step": 2790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654368743221022e-05,
+      "loss": 0.773,
+      "step": 2791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654060240704351e-05,
+      "loss": 25.8857,
+      "step": 2792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653751605501495e-05,
+      "loss": 28.27,
+      "step": 2793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653442837621252e-05,
+      "loss": 0.8182,
+      "step": 2794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653133937072423e-05,
+      "loss": 26.6134,
+      "step": 2795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652824903863818e-05,
+      "loss": 0.0773,
+      "step": 2796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652515738004246e-05,
+      "loss": 0.778,
+      "step": 2797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65220643950252e-05,
+      "loss": 23.9503,
+      "step": 2798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65189700836746e-05,
+      "loss": 0.8125,
+      "step": 2799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651587444607887e-05,
+      "loss": 0.0484,
+      "step": 2800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651277748232627e-05,
+      "loss": 0.7824,
+      "step": 2801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650967919250509e-05,
+      "loss": 28.1034,
+      "step": 2802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650657957670365e-05,
+      "loss": 24.1636,
+      "step": 2803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650347863501034e-05,
+      "loss": 27.272,
+      "step": 2804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650037636751354e-05,
+      "loss": 0.0763,
+      "step": 2805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649727277430174e-05,
+      "loss": 24.9976,
+      "step": 2806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649416785546338e-05,
+      "loss": 28.2465,
+      "step": 2807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6491061611087e-05,
+      "loss": 26.3652,
+      "step": 2808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648795404126115e-05,
+      "loss": 0.7193,
+      "step": 2809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648484514607443e-05,
+      "loss": 22.9265,
+      "step": 2810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648173492561548e-05,
+      "loss": 0.0831,
+      "step": 2811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647862337997296e-05,
+      "loss": 24.1706,
+      "step": 2812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647551050923561e-05,
+      "loss": 0.7937,
+      "step": 2813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647239631349212e-05,
+      "loss": 0.7979,
+      "step": 2814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646928079283132e-05,
+      "loss": 0.0852,
+      "step": 2815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646616394734201e-05,
+      "loss": 0.7442,
+      "step": 2816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646304577711307e-05,
+      "loss": 26.8771,
+      "step": 2817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645992628223339e-05,
+      "loss": 23.7187,
+      "step": 2818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645680546279191e-05,
+      "loss": 23.2626,
+      "step": 2819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64536833188776e-05,
+      "loss": 0.0903,
+      "step": 2820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645055985057947e-05,
+      "loss": 24.8811,
+      "step": 2821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644743505798655e-05,
+      "loss": 26.1146,
+      "step": 2822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644430894118797e-05,
+      "loss": 24.8153,
+      "step": 2823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644118150027283e-05,
+      "loss": 24.9649,
+      "step": 2824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643805273533029e-05,
+      "loss": 0.7731,
+      "step": 2825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643492264644955e-05,
+      "loss": 25.6118,
+      "step": 2826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643179123371986e-05,
+      "loss": 27.7615,
+      "step": 2827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642865849723049e-05,
+      "loss": 0.761,
+      "step": 2828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642552443707074e-05,
+      "loss": 24.5831,
+      "step": 2829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642238905332999e-05,
+      "loss": 23.9701,
+      "step": 2830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641925234609762e-05,
+      "loss": 25.2599,
+      "step": 2831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641611431546303e-05,
+      "loss": 0.7069,
+      "step": 2832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641297496151571e-05,
+      "loss": 0.7231,
+      "step": 2833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640983428434516e-05,
+      "loss": 0.7716,
+      "step": 2834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640669228404091e-05,
+      "loss": 23.7572,
+      "step": 2835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640354896069254e-05,
+      "loss": 23.1164,
+      "step": 2836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640040431438968e-05,
+      "loss": 25.4713,
+      "step": 2837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639725834522197e-05,
+      "loss": 0.785,
+      "step": 2838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639411105327911e-05,
+      "loss": 23.7978,
+      "step": 2839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639096243865082e-05,
+      "loss": 23.1984,
+      "step": 2840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638781250142685e-05,
+      "loss": 0.0672,
+      "step": 2841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638466124169705e-05,
+      "loss": 0.7378,
+      "step": 2842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638150865955122e-05,
+      "loss": 0.7732,
+      "step": 2843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637835475507925e-05,
+      "loss": 0.7765,
+      "step": 2844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637519952837109e-05,
+      "loss": 23.5968,
+      "step": 2845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637204297951662e-05,
+      "loss": 27.6254,
+      "step": 2846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636888510860591e-05,
+      "loss": 0.072,
+      "step": 2847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636572591572895e-05,
+      "loss": 0.782,
+      "step": 2848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636256540097582e-05,
+      "loss": 26.2757,
+      "step": 2849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635940356443662e-05,
+      "loss": 0.6758,
+      "step": 2850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63562404062015e-05,
+      "loss": 0.0586,
+      "step": 2851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635307592636062e-05,
+      "loss": 25.4701,
+      "step": 2852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634991012500423e-05,
+      "loss": 26.7093,
+      "step": 2853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634674300222255e-05,
+      "loss": 0.7527,
+      "step": 2854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634357455810592e-05,
+      "loss": 25.1612,
+      "step": 2855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634040479274461e-05,
+      "loss": 0.7883,
+      "step": 2856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633723370622907e-05,
+      "loss": 24.1936,
+      "step": 2857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633406129864963e-05,
+      "loss": 0.7717,
+      "step": 2858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633088757009679e-05,
+      "loss": 26.6975,
+      "step": 2859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632771252066099e-05,
+      "loss": 0.0596,
+      "step": 2860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632453615043277e-05,
+      "loss": 25.7667,
+      "step": 2861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632135845950271e-05,
+      "loss": 0.7684,
+      "step": 2862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631817944796137e-05,
+      "loss": 23.5305,
+      "step": 2863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631499911589939e-05,
+      "loss": 27.5668,
+      "step": 2864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631181746340745e-05,
+      "loss": 24.2593,
+      "step": 2865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630863449057625e-05,
+      "loss": 0.0993,
+      "step": 2866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630545019749653e-05,
+      "loss": 26.3647,
+      "step": 2867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630226458425911e-05,
+      "loss": 0.826,
+      "step": 2868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629907765095477e-05,
+      "loss": 25.3904,
+      "step": 2869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629588939767436e-05,
+      "loss": 24.4015,
+      "step": 2870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629269982450883e-05,
+      "loss": 0.7842,
+      "step": 2871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628950893154907e-05,
+      "loss": 25.5233,
+      "step": 2872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628631671888605e-05,
+      "loss": 0.7993,
+      "step": 2873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628312318661083e-05,
+      "loss": 25.1928,
+      "step": 2874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62799283348144e-05,
+      "loss": 0.0725,
+      "step": 2875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627673216358785e-05,
+      "loss": 23.514,
+      "step": 2876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627353467302235e-05,
+      "loss": 0.7566,
+      "step": 2877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627033586320901e-05,
+      "loss": 0.0946,
+      "step": 2878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626713573423904e-05,
+      "loss": 24.6403,
+      "step": 2879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62639342862037e-05,
+      "loss": 26.2771,
+      "step": 2880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626073151919421e-05,
+      "loss": 0.7313,
+      "step": 2881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625752743330195e-05,
+      "loss": 0.8165,
+      "step": 2882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62543220286182e-05,
+      "loss": 0.0807,
+      "step": 2883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625111530523438e-05,
+      "loss": 0.7547,
+      "step": 2884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624790726324193e-05,
+      "loss": 0.8089,
+      "step": 2885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624469790273226e-05,
+      "loss": 24.6982,
+      "step": 2886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624148722379691e-05,
+      "loss": 0.8332,
+      "step": 2887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62382752265274e-05,
+      "loss": 23.8688,
+      "step": 2888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623506191101531e-05,
+      "loss": 24.5233,
+      "step": 2889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623184727735225e-05,
+      "loss": 23.3146,
+      "step": 2890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622863132562986e-05,
+      "loss": 25.4444,
+      "step": 2891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622541405593984e-05,
+      "loss": 0.7115,
+      "step": 2892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622219546837388e-05,
+      "loss": 0.7351,
+      "step": 2893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621897556302379e-05,
+      "loss": 0.7382,
+      "step": 2894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621575433998133e-05,
+      "loss": 0.7325,
+      "step": 2895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621253179933835e-05,
+      "loss": 0.7133,
+      "step": 2896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620930794118674e-05,
+      "loss": 0.7348,
+      "step": 2897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62060827656184e-05,
+      "loss": 23.1865,
+      "step": 2898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620285627272526e-05,
+      "loss": 24.5884,
+      "step": 2899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619962846259934e-05,
+      "loss": 26.2256,
+      "step": 2900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619639933533262e-05,
+      "loss": 26.6315,
+      "step": 2901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619316889101721e-05,
+      "loss": 0.7935,
+      "step": 2902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618993712974518e-05,
+      "loss": 25.966,
+      "step": 2903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618670405160867e-05,
+      "loss": 0.7043,
+      "step": 2904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618346965669987e-05,
+      "loss": 24.1181,
+      "step": 2905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618023394511097e-05,
+      "loss": 24.2151,
+      "step": 2906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617699691693423e-05,
+      "loss": 23.9425,
+      "step": 2907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617375857226193e-05,
+      "loss": 27.6099,
+      "step": 2908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617051891118641e-05,
+      "loss": 0.7361,
+      "step": 2909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616727793380003e-05,
+      "loss": 28.4092,
+      "step": 2910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616403564019516e-05,
+      "loss": 23.0583,
+      "step": 2911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616079203046427e-05,
+      "loss": 0.7268,
+      "step": 2912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615754710469984e-05,
+      "loss": 26.6785,
+      "step": 2913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615430086299435e-05,
+      "loss": 24.4948,
+      "step": 2914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615105330544037e-05,
+      "loss": 0.7778,
+      "step": 2915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614780443213047e-05,
+      "loss": 23.8842,
+      "step": 2916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614455424315732e-05,
+      "loss": 0.0813,
+      "step": 2917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614130273861352e-05,
+      "loss": 0.7954,
+      "step": 2918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613804991859181e-05,
+      "loss": 25.6711,
+      "step": 2919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613479578318491e-05,
+      "loss": 26.1353,
+      "step": 2920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61315403324856e-05,
+      "loss": 0.7921,
+      "step": 2921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61282835665867e-05,
+      "loss": 25.5327,
+      "step": 2922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612502548558106e-05,
+      "loss": 0.7329,
+      "step": 2923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612176608956154e-05,
+      "loss": 25.1403,
+      "step": 2924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61185053786211e-05,
+      "loss": 0.0792,
+      "step": 2925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611524335285269e-05,
+      "loss": 22.9902,
+      "step": 2926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61119800123493e-05,
+      "loss": 0.7682,
+      "step": 2927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610871535720396e-05,
+      "loss": 23.9399,
+      "step": 2928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610544938750978e-05,
+      "loss": 0.783,
+      "step": 2929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610218210335984e-05,
+      "loss": 26.8012,
+      "step": 2930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60989135048473e-05,
+      "loss": 24.3294,
+      "step": 2931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609564359206534e-05,
+      "loss": 0.7674,
+      "step": 2932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60923723651072e-05,
+      "loss": 0.7595,
+      "step": 2933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608909982406614e-05,
+      "loss": 0.7709,
+      "step": 2934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608582596903544e-05,
+      "loss": 27.733,
+      "step": 2935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608255080010844e-05,
+      "loss": 22.7157,
+      "step": 2936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607927431737853e-05,
+      "loss": 25.5771,
+      "step": 2937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607599652093912e-05,
+      "loss": 0.0814,
+      "step": 2938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607271741088367e-05,
+      "loss": 0.7473,
+      "step": 2939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606943698730563e-05,
+      "loss": 0.0896,
+      "step": 2940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606615525029855e-05,
+      "loss": 0.8103,
+      "step": 2941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606287219995599e-05,
+      "loss": 0.7277,
+      "step": 2942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605958783637155e-05,
+      "loss": 29.8766,
+      "step": 2943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605630215963885e-05,
+      "loss": 0.8522,
+      "step": 2944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605301516985158e-05,
+      "loss": 27.6373,
+      "step": 2945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604972686710345e-05,
+      "loss": 25.2514,
+      "step": 2946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604643725148821e-05,
+      "loss": 23.0382,
+      "step": 2947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604314632309963e-05,
+      "loss": 0.7838,
+      "step": 2948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603985408203156e-05,
+      "loss": 0.0828,
+      "step": 2949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603656052837784e-05,
+      "loss": 0.7243,
+      "step": 2950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603326566223239e-05,
+      "loss": 27.0041,
+      "step": 2951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602996948368911e-05,
+      "loss": 0.09,
+      "step": 2952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6026671992842e-05,
+      "loss": 0.7757,
+      "step": 2953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602337318978508e-05,
+      "loss": 0.0754,
+      "step": 2954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602007307461237e-05,
+      "loss": 30.1695,
+      "step": 2955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601677164741796e-05,
+      "loss": 0.7647,
+      "step": 2956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601346890829598e-05,
+      "loss": 25.7967,
+      "step": 2957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601016485734063e-05,
+      "loss": 27.7372,
+      "step": 2958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600685949464604e-05,
+      "loss": 25.3784,
+      "step": 2959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600355282030648e-05,
+      "loss": 0.0784,
+      "step": 2960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600024483441623e-05,
+      "loss": 0.7312,
+      "step": 2961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599693553706956e-05,
+      "loss": 28.1703,
+      "step": 2962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599362492836086e-05,
+      "loss": 26.2542,
+      "step": 2963
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59903130083845e-05,
+      "loss": 0.7477,
+      "step": 2964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59869997772349e-05,
+      "loss": 25.9809,
+      "step": 2965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598368523500653e-05,
+      "loss": 0.6368,
+      "step": 2966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598036938179387e-05,
+      "loss": 26.0495,
+      "step": 2967
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597705221769147e-05,
+      "loss": 26.789,
+      "step": 2968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597373374279388e-05,
+      "loss": 25.3995,
+      "step": 2969
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597041395719573e-05,
+      "loss": 0.0828,
+      "step": 2970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596709286099165e-05,
+      "loss": 23.6332,
+      "step": 2971
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596377045427636e-05,
+      "loss": 0.0858,
+      "step": 2972
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596044673714453e-05,
+      "loss": 0.7505,
+      "step": 2973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595712170969095e-05,
+      "loss": 23.0388,
+      "step": 2974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595379537201041e-05,
+      "loss": 27.8342,
+      "step": 2975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595046772419774e-05,
+      "loss": 0.0818,
+      "step": 2976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59471387663478e-05,
+      "loss": 0.7683,
+      "step": 2977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594380849855553e-05,
+      "loss": 0.7266,
+      "step": 2978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594047692091584e-05,
+      "loss": 0.7581,
+      "step": 2979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593714403352371e-05,
+      "loss": 0.7761,
+      "step": 2980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593380983647421e-05,
+      "loss": 27.8315,
+      "step": 2981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593047432986235e-05,
+      "loss": 28.8308,
+      "step": 2982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592713751378324e-05,
+      "loss": 0.042,
+      "step": 2983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5923799388332e-05,
+      "loss": 0.0873,
+      "step": 2984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592045995360383e-05,
+      "loss": 22.6856,
+      "step": 2985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591711920969391e-05,
+      "loss": 22.7333,
+      "step": 2986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591377715669748e-05,
+      "loss": 25.5767,
+      "step": 2987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591043379470982e-05,
+      "loss": 23.5882,
+      "step": 2988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590708912382628e-05,
+      "loss": 0.0938,
+      "step": 2989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59037431441422e-05,
+      "loss": 25.72,
+      "step": 2990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590039585575296e-05,
+      "loss": 26.1016,
+      "step": 2991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5897047258754e-05,
+      "loss": 0.7495,
+      "step": 2992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589369735324078e-05,
+      "loss": 23.6017,
+      "step": 2993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589034613930883e-05,
+      "loss": 22.825,
+      "step": 2994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588699361705366e-05,
+      "loss": 0.7698,
+      "step": 2995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588363978657087e-05,
+      "loss": 0.805,
+      "step": 2996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588028464795607e-05,
+      "loss": 26.7525,
+      "step": 2997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587692820130491e-05,
+      "loss": 0.775,
+      "step": 2998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587357044671308e-05,
+      "loss": 26.068,
+      "step": 2999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587021138427632e-05,
+      "loss": 23.9801,
+      "step": 3000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58668510140904e-05,
+      "loss": 25.919,
+      "step": 3001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586348933625111e-05,
+      "loss": 0.0656,
+      "step": 3002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586012635085428e-05,
+      "loss": 25.5201,
+      "step": 3003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585676205799583e-05,
+      "loss": 0.751,
+      "step": 3004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585339645777163e-05,
+      "loss": 0.7817,
+      "step": 3005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585002955027767e-05,
+      "loss": 24.1107,
+      "step": 3006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584666133560991e-05,
+      "loss": 0.7839,
+      "step": 3007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584329181386438e-05,
+      "loss": 0.8217,
+      "step": 3008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583992098513717e-05,
+      "loss": 24.6784,
+      "step": 3009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583654884952434e-05,
+      "loss": 25.7961,
+      "step": 3010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583317540712207e-05,
+      "loss": 24.3621,
+      "step": 3011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582980065802651e-05,
+      "loss": 25.0008,
+      "step": 3012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582642460233388e-05,
+      "loss": 0.7825,
+      "step": 3013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582304724014044e-05,
+      "loss": 24.3226,
+      "step": 3014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581966857154246e-05,
+      "loss": 26.585,
+      "step": 3015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58162885966363e-05,
+      "loss": 24.6808,
+      "step": 3016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581290731551826e-05,
+      "loss": 26.098,
+      "step": 3017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580952472828482e-05,
+      "loss": 25.5271,
+      "step": 3018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580614083503234e-05,
+      "loss": 0.0909,
+      "step": 3019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580275563585734e-05,
+      "loss": 26.7599,
+      "step": 3020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579936913085632e-05,
+      "loss": 24.6112,
+      "step": 3021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579598132012581e-05,
+      "loss": 26.9298,
+      "step": 3022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579259220376245e-05,
+      "loss": 21.3686,
+      "step": 3023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57892017818628e-05,
+      "loss": 0.0693,
+      "step": 3024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578581005452356e-05,
+      "loss": 0.734,
+      "step": 3025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57824170218414e-05,
+      "loss": 25.3093,
+      "step": 3026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577902268391308e-05,
+      "loss": 0.7163,
+      "step": 3027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577562704083535e-05,
+      "loss": 22.8486,
+      "step": 3028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577223009270504e-05,
+      "loss": 24.1514,
+      "step": 3029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576883183961898e-05,
+      "loss": 0.1014,
+      "step": 3030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576543228167405e-05,
+      "loss": 24.6623,
+      "step": 3031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576203141896718e-05,
+      "loss": 23.9778,
+      "step": 3032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575862925159534e-05,
+      "loss": 24.4175,
+      "step": 3033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57552257796555e-05,
+      "loss": 0.8124,
+      "step": 3034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575182100324473e-05,
+      "loss": 24.3988,
+      "step": 3035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574841492246006e-05,
+      "loss": 26.2543,
+      "step": 3036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574500753739859e-05,
+      "loss": 25.5863,
+      "step": 3037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57415988481575e-05,
+      "loss": 23.5529,
+      "step": 3038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573818885483396e-05,
+      "loss": 0.0876,
+      "step": 3039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573477755752518e-05,
+      "loss": 0.8191,
+      "step": 3040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573136495632842e-05,
+      "loss": 25.9192,
+      "step": 3041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572795105134095e-05,
+      "loss": 24.5435,
+      "step": 3042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572453584266014e-05,
+      "loss": 0.8047,
+      "step": 3043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572111933038334e-05,
+      "loss": 24.9548,
+      "step": 3044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571770151460796e-05,
+      "loss": 0.7513,
+      "step": 3045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571428239543142e-05,
+      "loss": 23.8188,
+      "step": 3046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57108619729512e-05,
+      "loss": 24.221,
+      "step": 3047
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570744024726486e-05,
+      "loss": 25.4812,
+      "step": 3048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570401721846989e-05,
+      "loss": 0.8109,
+      "step": 3049
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570059288666391e-05,
+      "loss": 24.4912,
+      "step": 3050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569716725194456e-05,
+      "loss": 0.7751,
+      "step": 3051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569374031440947e-05,
+      "loss": 23.7807,
+      "step": 3052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569031207415638e-05,
+      "loss": 26.8168,
+      "step": 3053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568688253128299e-05,
+      "loss": 0.7957,
+      "step": 3054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568345168588711e-05,
+      "loss": 0.7199,
+      "step": 3055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568001953806654e-05,
+      "loss": 25.1115,
+      "step": 3056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567658608791912e-05,
+      "loss": 0.7687,
+      "step": 3057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567315133554275e-05,
+      "loss": 23.8956,
+      "step": 3058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566971528103534e-05,
+      "loss": 0.784,
+      "step": 3059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566627792449485e-05,
+      "loss": 0.0645,
+      "step": 3060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56628392660193e-05,
+      "loss": 0.766,
+      "step": 3061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56593993057067e-05,
+      "loss": 0.0629,
+      "step": 3062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565595804365517e-05,
+      "loss": 22.7771,
+      "step": 3063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565251547996275e-05,
+      "loss": 0.7188,
+      "step": 3064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564907161472764e-05,
+      "loss": 0.7485,
+      "step": 3065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5645626448048e-05,
+      "loss": 0.8146,
+      "step": 3066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564217998002205e-05,
+      "loss": 25.6526,
+      "step": 3067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563873221074806e-05,
+      "loss": 24.7832,
+      "step": 3068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563528314032431e-05,
+      "loss": 0.7879,
+      "step": 3069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563183276884915e-05,
+      "loss": 24.8398,
+      "step": 3070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562838109642095e-05,
+      "loss": 25.5124,
+      "step": 3071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56249281231381e-05,
+      "loss": 0.7881,
+      "step": 3072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562147384909904e-05,
+      "loss": 0.0859,
+      "step": 3073
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561801827440226e-05,
+      "loss": 0.8484,
+      "step": 3074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56145613991463e-05,
+      "loss": 23.2766,
+      "step": 3075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561110322342967e-05,
+      "loss": 24.5348,
+      "step": 3076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5607643747351e-05,
+      "loss": 0.7872,
+      "step": 3077
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560418297100891e-05,
+      "loss": 25.8807,
+      "step": 3078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560072089450204e-05,
+      "loss": 24.9749,
+      "step": 3079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559725751792914e-05,
+      "loss": 0.728,
+      "step": 3080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55937928413889e-05,
+      "loss": 22.9012,
+      "step": 3081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559032686498014e-05,
+      "loss": 25.3098,
+      "step": 3082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558685958880163e-05,
+      "loss": 28.4584,
+      "step": 3083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558339101295226e-05,
+      "loss": 0.8024,
+      "step": 3084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55799211375309e-05,
+      "loss": 22.7035,
+      "step": 3085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557644996263648e-05,
+      "loss": 0.0715,
+      "step": 3086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557297748836797e-05,
+      "loss": 0.0559,
+      "step": 3087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556950371482436e-05,
+      "loss": 24.6875,
+      "step": 3088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556602864210469e-05,
+      "loss": 0.7603,
+      "step": 3089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556255227030801e-05,
+      "loss": 23.8629,
+      "step": 3090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555907459953347e-05,
+      "loss": 25.4572,
+      "step": 3091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55555956298802e-05,
+      "loss": 0.743,
+      "step": 3092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555211536144739e-05,
+      "loss": 23.3076,
+      "step": 3093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554863379433424e-05,
+      "loss": 0.0601,
+      "step": 3094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554515092864002e-05,
+      "loss": 24.6404,
+      "step": 3095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554166676446405e-05,
+      "loss": 0.787,
+      "step": 3096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553818130190562e-05,
+      "loss": 26.3544,
+      "step": 3097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553469454106414e-05,
+      "loss": 23.0544,
+      "step": 3098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553120648203899e-05,
+      "loss": 25.099,
+      "step": 3099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552771712492964e-05,
+      "loss": 0.7597,
+      "step": 3100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552422646983555e-05,
+      "loss": 26.3073,
+      "step": 3101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552073451685623e-05,
+      "loss": 25.528,
+      "step": 3102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551724126609125e-05,
+      "loss": 0.1098,
+      "step": 3103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55137467176402e-05,
+      "loss": 24.3437,
+      "step": 3104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55102508716027e-05,
+      "loss": 25.7671,
+      "step": 3105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550675372807842e-05,
+      "loss": 0.0823,
+      "step": 3106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550325528716708e-05,
+      "loss": 25.5076,
+      "step": 3107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549975554896839e-05,
+      "loss": 26.9812,
+      "step": 3108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549625451358215e-05,
+      "loss": 24.7272,
+      "step": 3109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549275218110818e-05,
+      "loss": 0.0814,
+      "step": 3110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54892485516463e-05,
+      "loss": 0.7449,
+      "step": 3111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548574362529642e-05,
+      "loss": 23.1385,
+      "step": 3112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548223740215844e-05,
+      "loss": 24.9353,
+      "step": 3113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547872988233236e-05,
+      "loss": 23.6285,
+      "step": 3114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547522106591815e-05,
+      "loss": 0.7396,
+      "step": 3115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547171095301585e-05,
+      "loss": 23.662,
+      "step": 3116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546819954372557e-05,
+      "loss": 23.3933,
+      "step": 3117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546468683814736e-05,
+      "loss": 0.809,
+      "step": 3118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546117283638138e-05,
+      "loss": 0.0756,
+      "step": 3119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545765753852785e-05,
+      "loss": 0.8048,
+      "step": 3120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545414094468695e-05,
+      "loss": 26.0247,
+      "step": 3121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545062305495896e-05,
+      "loss": 23.5253,
+      "step": 3122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544710386944418e-05,
+      "loss": 0.0683,
+      "step": 3123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544358338824291e-05,
+      "loss": 23.7618,
+      "step": 3124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544006161145556e-05,
+      "loss": 23.7001,
+      "step": 3125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543653853918249e-05,
+      "loss": 0.7436,
+      "step": 3126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543301417152418e-05,
+      "loss": 0.7756,
+      "step": 3127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542948850858108e-05,
+      "loss": 25.7097,
+      "step": 3128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542596155045372e-05,
+      "loss": 24.1237,
+      "step": 3129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542243329724266e-05,
+      "loss": 27.8772,
+      "step": 3130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541890374904847e-05,
+      "loss": 0.7794,
+      "step": 3131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54153729059718e-05,
+      "loss": 0.7224,
+      "step": 3132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54118407681133e-05,
+      "loss": 24.0372,
+      "step": 3133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540830733557366e-05,
+      "loss": 24.1489,
+      "step": 3134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540477260845365e-05,
+      "loss": 0.8248,
+      "step": 3135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540123658685402e-05,
+      "loss": 26.8812,
+      "step": 3136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539769927087557e-05,
+      "loss": 25.6126,
+      "step": 3137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539416066061917e-05,
+      "loss": 23.4573,
+      "step": 3138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53906207561857e-05,
+      "loss": 23.6371,
+      "step": 3139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538707955767608e-05,
+      "loss": 0.8217,
+      "step": 3140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538353706519127e-05,
+      "loss": 24.7084,
+      "step": 3141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537999327883227e-05,
+      "loss": 27.3415,
+      "step": 3142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537644819870008e-05,
+      "loss": 20.9162,
+      "step": 3143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537290182489583e-05,
+      "loss": 25.9164,
+      "step": 3144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536935415752057e-05,
+      "loss": 25.0231,
+      "step": 3145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536580519667547e-05,
+      "loss": 0.7428,
+      "step": 3146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536225494246171e-05,
+      "loss": 22.5276,
+      "step": 3147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53587033949805e-05,
+      "loss": 0.7178,
+      "step": 3148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535515055433309e-05,
+      "loss": 25.9867,
+      "step": 3149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535159642062079e-05,
+      "loss": 0.7701,
+      "step": 3150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534804099394492e-05,
+      "loss": 0.7803,
+      "step": 3151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534448427440682e-05,
+      "loss": 0.1017,
+      "step": 3152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534092626210791e-05,
+      "loss": 0.0743,
+      "step": 3153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533736695714964e-05,
+      "loss": 22.8597,
+      "step": 3154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533380635963347e-05,
+      "loss": 25.4947,
+      "step": 3155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533024446966093e-05,
+      "loss": 0.7337,
+      "step": 3156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532668128733353e-05,
+      "loss": 0.7375,
+      "step": 3157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532311681275288e-05,
+      "loss": 22.1459,
+      "step": 3158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531955104602062e-05,
+      "loss": 24.2735,
+      "step": 3159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531598398723837e-05,
+      "loss": 0.7548,
+      "step": 3160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531241563650786e-05,
+      "loss": 0.7485,
+      "step": 3161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530884599393081e-05,
+      "loss": 23.0412,
+      "step": 3162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530527505960898e-05,
+      "loss": 23.6074,
+      "step": 3163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530170283364419e-05,
+      "loss": 23.7544,
+      "step": 3164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.529812931613829e-05,
+      "loss": 0.7275,
+      "step": 3165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529455450719313e-05,
+      "loss": 0.8278,
+      "step": 3166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529097840691066e-05,
+      "loss": 0.6761,
+      "step": 3167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528740101539282e-05,
+      "loss": 23.4581,
+      "step": 3168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52838223327416e-05,
+      "loss": 0.0802,
+      "step": 3169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528024235905903e-05,
+      "loss": 0.7455,
+      "step": 3170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527666109444717e-05,
+      "loss": 25.0016,
+      "step": 3171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527307853900811e-05,
+      "loss": 26.0336,
+      "step": 3172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526949469284402e-05,
+      "loss": 22.8234,
+      "step": 3173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526590955605705e-05,
+      "loss": 23.1936,
+      "step": 3174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526232312874942e-05,
+      "loss": 0.774,
+      "step": 3175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525873541102337e-05,
+      "loss": 0.0858,
+      "step": 3176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525514640298119e-05,
+      "loss": 23.2256,
+      "step": 3177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52515561047252e-05,
+      "loss": 25.1485,
+      "step": 3178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524796451635778e-05,
+      "loss": 0.7717,
+      "step": 3179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524437163798129e-05,
+      "loss": 21.0683,
+      "step": 3180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524077746969819e-05,
+      "loss": 0.7804,
+      "step": 3181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523718201161091e-05,
+      "loss": 24.1739,
+      "step": 3182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523358526382201e-05,
+      "loss": 22.9415,
+      "step": 3183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522998722643399e-05,
+      "loss": 24.8338,
+      "step": 3184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522638789954946e-05,
+      "loss": 0.7817,
+      "step": 3185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522278728327101e-05,
+      "loss": 23.6968,
+      "step": 3186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521918537770132e-05,
+      "loss": 22.725,
+      "step": 3187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521558218294304e-05,
+      "loss": 0.7788,
+      "step": 3188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521197769909893e-05,
+      "loss": 23.7437,
+      "step": 3189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520837192627175e-05,
+      "loss": 0.1051,
+      "step": 3190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520476486456427e-05,
+      "loss": 25.608,
+      "step": 3191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520115651407936e-05,
+      "loss": 24.895,
+      "step": 3192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519754687491989e-05,
+      "loss": 20.7296,
+      "step": 3193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519393594718875e-05,
+      "loss": 24.0786,
+      "step": 3194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51903237309889e-05,
+      "loss": 25.183,
+      "step": 3195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518671022642332e-05,
+      "loss": 25.2224,
+      "step": 3196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518309543359505e-05,
+      "loss": 0.7744,
+      "step": 3197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517947935260712e-05,
+      "loss": 24.6557,
+      "step": 3198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517586198356262e-05,
+      "loss": 0.7324,
+      "step": 3199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51722433265647e-05,
+      "loss": 23.712,
+      "step": 3200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516862338171652e-05,
+      "loss": 0.7312,
+      "step": 3201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516500214912128e-05,
+      "loss": 22.1088,
+      "step": 3202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516137962888224e-05,
+      "loss": 0.1057,
+      "step": 3203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515775582110264e-05,
+      "loss": 26.6305,
+      "step": 3204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515413072588584e-05,
+      "loss": 26.8636,
+      "step": 3205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515050434333514e-05,
+      "loss": 24.8341,
+      "step": 3206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514687667355397e-05,
+      "loss": 0.7328,
+      "step": 3207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514324771664573e-05,
+      "loss": 25.1541,
+      "step": 3208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51396174727139e-05,
+      "loss": 0.8328,
+      "step": 3209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513598594186196e-05,
+      "loss": 0.7536,
+      "step": 3210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513235312419344e-05,
+      "loss": 23.7532,
+      "step": 3211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512871901981193e-05,
+      "loss": 24.0489,
+      "step": 3212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512508362882103e-05,
+      "loss": 25.2543,
+      "step": 3213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512144695132437e-05,
+      "loss": 23.6076,
+      "step": 3214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511780898742565e-05,
+      "loss": 23.7866,
+      "step": 3215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511416973722857e-05,
+      "loss": 26.5974,
+      "step": 3216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511052920083692e-05,
+      "loss": 0.0882,
+      "step": 3217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510688737835445e-05,
+      "loss": 0.7947,
+      "step": 3218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.5103244269885e-05,
+      "loss": 25.93,
+      "step": 3219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509959987553245e-05,
+      "loss": 27.2859,
+      "step": 3220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509595419540067e-05,
+      "loss": 26.7785,
+      "step": 3221
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509230722959363e-05,
+      "loss": 23.9093,
+      "step": 3222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508865897821528e-05,
+      "loss": 0.0538,
+      "step": 3223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508500944136964e-05,
+      "loss": 24.0366,
+      "step": 3224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508135861916077e-05,
+      "loss": 24.5557,
+      "step": 3225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507770651169275e-05,
+      "loss": 25.5193,
+      "step": 3226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507405311906967e-05,
+      "loss": 25.5327,
+      "step": 3227
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507039844139572e-05,
+      "loss": 23.4817,
+      "step": 3228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506674247877508e-05,
+      "loss": 21.9942,
+      "step": 3229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506308523131198e-05,
+      "loss": 0.7042,
+      "step": 3230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50594266991107e-05,
+      "loss": 0.7442,
+      "step": 3231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505576688227554e-05,
+      "loss": 0.7401,
+      "step": 3232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505210578091084e-05,
+      "loss": 24.7444,
+      "step": 3233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504844339512095e-05,
+      "loss": 23.8744,
+      "step": 3234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504477972501034e-05,
+      "loss": 23.6042,
+      "step": 3235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504111477068341e-05,
+      "loss": 25.5327,
+      "step": 3236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503744853224466e-05,
+      "loss": 24.0494,
+      "step": 3237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503378100979863e-05,
+      "loss": 22.991,
+      "step": 3238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503011220344986e-05,
+      "loss": 21.9577,
+      "step": 3239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502644211330297e-05,
+      "loss": 22.326,
+      "step": 3240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502277073946257e-05,
+      "loss": 0.8162,
+      "step": 3241
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501909808203333e-05,
+      "loss": 22.9335,
+      "step": 3242
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501542414111997e-05,
+      "loss": 24.1144,
+      "step": 3243
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501174891682724e-05,
+      "loss": 0.7763,
+      "step": 3244
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500807240925991e-05,
+      "loss": 24.5302,
+      "step": 3245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500439461852277e-05,
+      "loss": 24.4472,
+      "step": 3246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500071554472072e-05,
+      "loss": 25.7443,
+      "step": 3247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499703518795861e-05,
+      "loss": 24.0368,
+      "step": 3248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49933535483414e-05,
+      "loss": 0.7864,
+      "step": 3249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498967062597403e-05,
+      "loss": 0.8,
+      "step": 3250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498598642096153e-05,
+      "loss": 24.6811,
+      "step": 3251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498230093340887e-05,
+      "loss": 0.7164,
+      "step": 3252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49786141634212e-05,
+      "loss": 24.3098,
+      "step": 3253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497492611110357e-05,
+      "loss": 21.7127,
+      "step": 3254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497123677656115e-05,
+      "loss": 23.8906,
+      "step": 3255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496754615989914e-05,
+      "loss": 0.0779,
+      "step": 3256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496385426122272e-05,
+      "loss": 23.1473,
+      "step": 3257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496016108063716e-05,
+      "loss": 0.7949,
+      "step": 3258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495646661824776e-05,
+      "loss": 0.726,
+      "step": 3259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495277087415986e-05,
+      "loss": 0.0855,
+      "step": 3260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49490738484788e-05,
+      "loss": 23.6185,
+      "step": 3261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494537554130998e-05,
+      "loss": 24.4304,
+      "step": 3262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494167595275885e-05,
+      "loss": 0.7669,
+      "step": 3263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493797508293089e-05,
+      "loss": 0.7397,
+      "step": 3264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493427293193161e-05,
+      "loss": 27.2818,
+      "step": 3265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493056949986654e-05,
+      "loss": 0.7692,
+      "step": 3266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492686478684129e-05,
+      "loss": 0.0508,
+      "step": 3267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492315879296146e-05,
+      "loss": 0.7128,
+      "step": 3268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491945151833271e-05,
+      "loss": 0.7391,
+      "step": 3269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491574296306072e-05,
+      "loss": 23.1239,
+      "step": 3270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491203312725126e-05,
+      "loss": 24.7826,
+      "step": 3271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490832201101005e-05,
+      "loss": 22.5193,
+      "step": 3272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490460961444294e-05,
+      "loss": 0.8154,
+      "step": 3273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490089593765575e-05,
+      "loss": 22.4279,
+      "step": 3274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.489718098075433e-05,
+      "loss": 0.7451,
+      "step": 3275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.489346474384462e-05,
+      "loss": 24.0483,
+      "step": 3276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488974722703256e-05,
+      "loss": 26.1199,
+      "step": 3277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488602843042414e-05,
+      "loss": 24.2995,
+      "step": 3278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48823083541254e-05,
+      "loss": 0.0724,
+      "step": 3279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487858699824235e-05,
+      "loss": 23.9578,
+      "step": 3280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487486436288113e-05,
+      "loss": 24.0896,
+      "step": 3281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487114044814784e-05,
+      "loss": 24.5113,
+      "step": 3282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486741525414868e-05,
+      "loss": 22.0439,
+      "step": 3283
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486368878098985e-05,
+      "loss": 24.3348,
+      "step": 3284
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485996102877756e-05,
+      "loss": 24.3239,
+      "step": 3285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485623199761811e-05,
+      "loss": 0.0701,
+      "step": 3286
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48525016876178e-05,
+      "loss": 24.5131,
+      "step": 3287
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484877009888302e-05,
+      "loss": 23.7673,
+      "step": 3288
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484503723152009e-05,
+      "loss": 0.0861,
+      "step": 3289
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48413030856355e-05,
+      "loss": 0.8449,
+      "step": 3290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483756766133566e-05,
+      "loss": 0.7717,
+      "step": 3291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483383095872711e-05,
+      "loss": 25.2738,
+      "step": 3292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483009297791634e-05,
+      "loss": 0.7876,
+      "step": 3293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482635371900994e-05,
+      "loss": 0.0797,
+      "step": 3294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482261318211452e-05,
+      "loss": 23.3777,
+      "step": 3295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481887136733671e-05,
+      "loss": 0.783,
+      "step": 3296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481512827478319e-05,
+      "loss": 24.3942,
+      "step": 3297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481138390456069e-05,
+      "loss": 23.147,
+      "step": 3298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480763825677594e-05,
+      "loss": 23.51,
+      "step": 3299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480389133153573e-05,
+      "loss": 0.8205,
+      "step": 3300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48001431289469e-05,
+      "loss": 22.8302,
+      "step": 3301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47963936491163e-05,
+      "loss": 0.768,
+      "step": 3302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479264289215083e-05,
+      "loss": 0.7254,
+      "step": 3303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478889085815743e-05,
+      "loss": 0.7234,
+      "step": 3304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478513754724303e-05,
+      "loss": 25.4797,
+      "step": 3305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478138295951469e-05,
+      "loss": 0.7147,
+      "step": 3306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477762709507943e-05,
+      "loss": 0.8416,
+      "step": 3307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477386995404432e-05,
+      "loss": 25.4845,
+      "step": 3308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477011153651648e-05,
+      "loss": 0.07,
+      "step": 3309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476635184260307e-05,
+      "loss": 23.7831,
+      "step": 3310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476259087241129e-05,
+      "loss": 0.0786,
+      "step": 3311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475882862604833e-05,
+      "loss": 24.569,
+      "step": 3312
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475506510362145e-05,
+      "loss": 22.9071,
+      "step": 3313
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4751300305238e-05,
+      "loss": 25.667,
+      "step": 3314
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474753423100525e-05,
+      "loss": 21.4247,
+      "step": 3315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474376688103061e-05,
+      "loss": 0.0633,
+      "step": 3316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473999825542148e-05,
+      "loss": 23.06,
+      "step": 3317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473622835428528e-05,
+      "loss": 21.9781,
+      "step": 3318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473245717772953e-05,
+      "loss": 0.7441,
+      "step": 3319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.472868472586172e-05,
+      "loss": 22.5957,
+      "step": 3320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.472491099878938e-05,
+      "loss": 0.8235,
+      "step": 3321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.472113599662014e-05,
+      "loss": 25.2368,
+      "step": 3322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471735971946161e-05,
+      "loss": 25.112,
+      "step": 3323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471358216742143e-05,
+      "loss": 0.7651,
+      "step": 3324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470980334060732e-05,
+      "loss": 0.0737,
+      "step": 3325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470602323912702e-05,
+      "loss": 25.234,
+      "step": 3326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470224186308828e-05,
+      "loss": 0.6954,
+      "step": 3327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469845921259892e-05,
+      "loss": 25.9482,
+      "step": 3328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469467528776677e-05,
+      "loss": 26.211,
+      "step": 3329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46908900886997e-05,
+      "loss": 0.084,
+      "step": 3330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468710361550566e-05,
+      "loss": 0.7286,
+      "step": 3331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468331586829258e-05,
+      "loss": 0.0636,
+      "step": 3332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467952684716844e-05,
+      "loss": 23.8697,
+      "step": 3333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467573655224127e-05,
+      "loss": 0.7807,
+      "step": 3334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467194498361915e-05,
+      "loss": 23.6407,
+      "step": 3335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466815214141014e-05,
+      "loss": 0.8098,
+      "step": 3336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466435802572242e-05,
+      "loss": 21.5029,
+      "step": 3337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466056263666412e-05,
+      "loss": 0.0726,
+      "step": 3338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465676597434345e-05,
+      "loss": 0.7845,
+      "step": 3339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465296803886867e-05,
+      "loss": 0.8104,
+      "step": 3340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464916883034804e-05,
+      "loss": 24.0195,
+      "step": 3341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46453683488899e-05,
+      "loss": 24.3614,
+      "step": 3342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464156659460257e-05,
+      "loss": 0.7759,
+      "step": 3343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463776356759446e-05,
+      "loss": 24.7029,
+      "step": 3344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463395926797399e-05,
+      "loss": 24.6035,
+      "step": 3345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463015369584959e-05,
+      "loss": 24.1049,
+      "step": 3346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46263468513298e-05,
+      "loss": 23.2307,
+      "step": 3347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462253873452314e-05,
+      "loss": 22.1157,
+      "step": 3348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461872934553815e-05,
+      "loss": 0.7924,
+      "step": 3349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461491868448347e-05,
+      "loss": 23.9523,
+      "step": 3350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461110675146774e-05,
+      "loss": 22.0746,
+      "step": 3351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46072935465996e-05,
+      "loss": 25.7659,
+      "step": 3352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460347906998778e-05,
+      "loss": 0.7974,
+      "step": 3353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459966332174107e-05,
+      "loss": 0.7703,
+      "step": 3354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.45958463019682e-05,
+      "loss": 0.0615,
+      "step": 3355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459202801077804e-05,
+      "loss": 25.8779,
+      "step": 3356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45882084482794e-05,
+      "loss": 23.4534,
+      "step": 3357
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458438761458122e-05,
+      "loss": 22.9696,
+      "step": 3358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45805655097924e-05,
+      "loss": 0.097,
+      "step": 3359
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457674213402192e-05,
+      "loss": 23.3278,
+      "step": 3360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457291748737878e-05,
+      "loss": 25.2973,
+      "step": 3361
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456909156997202e-05,
+      "loss": 0.7458,
+      "step": 3362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456526438191073e-05,
+      "loss": 25.3146,
+      "step": 3363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4561435923304e-05,
+      "loss": 22.1602,
+      "step": 3364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455760619426099e-05,
+      "loss": 25.4758,
+      "step": 3365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455377519489089e-05,
+      "loss": 0.7794,
+      "step": 3366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454994292530291e-05,
+      "loss": 24.2166,
+      "step": 3367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454610938560631e-05,
+      "loss": 21.6811,
+      "step": 3368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454227457591038e-05,
+      "loss": 21.9649,
+      "step": 3369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453843849632446e-05,
+      "loss": 23.5814,
+      "step": 3370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453460114695791e-05,
+      "loss": 0.7186,
+      "step": 3371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453076252792012e-05,
+      "loss": 0.0828,
+      "step": 3372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452692263932056e-05,
+      "loss": 0.7379,
+      "step": 3373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452308148126868e-05,
+      "loss": 0.7352,
+      "step": 3374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451923905387398e-05,
+      "loss": 0.757,
+      "step": 3375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451539535724603e-05,
+      "loss": 26.6033,
+      "step": 3376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45115503914944e-05,
+      "loss": 0.1524,
+      "step": 3377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450770415672872e-05,
+      "loss": 23.9505,
+      "step": 3378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450385665305864e-05,
+      "loss": 0.0594,
+      "step": 3379
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450000788059385e-05,
+      "loss": 27.143,
+      "step": 3380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449615783944408e-05,
+      "loss": 25.9464,
+      "step": 3381
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449230652971908e-05,
+      "loss": 0.0823,
+      "step": 3382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448845395152866e-05,
+      "loss": 25.8169,
+      "step": 3383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448460010498267e-05,
+      "loss": 22.8116,
+      "step": 3384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448074499019097e-05,
+      "loss": 0.7916,
+      "step": 3385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447688860726346e-05,
+      "loss": 0.7145,
+      "step": 3386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447303095631011e-05,
+      "loss": 21.8872,
+      "step": 3387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446917203744086e-05,
+      "loss": 0.1927,
+      "step": 3388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446531185076578e-05,
+      "loss": 24.3893,
+      "step": 3389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446145039639485e-05,
+      "loss": 23.5491,
+      "step": 3390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445758767443824e-05,
+      "loss": 0.1199,
+      "step": 3391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445372368500603e-05,
+      "loss": 22.0764,
+      "step": 3392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444985842820838e-05,
+      "loss": 24.3002,
+      "step": 3393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444599190415551e-05,
+      "loss": 22.3188,
+      "step": 3394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444212411295762e-05,
+      "loss": 0.8199,
+      "step": 3395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443825505472503e-05,
+      "loss": 0.8182,
+      "step": 3396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4434384729568e-05,
+      "loss": 0.7824,
+      "step": 3397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443051313759688e-05,
+      "loss": 0.8109,
+      "step": 3398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442664027892208e-05,
+      "loss": 0.6943,
+      "step": 3399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442276615365397e-05,
+      "loss": 0.7896,
+      "step": 3400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441889076190303e-05,
+      "loss": 25.5319,
+      "step": 3401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441501410377975e-05,
+      "loss": 0.8232,
+      "step": 3402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441113617939463e-05,
+      "loss": 0.0609,
+      "step": 3403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440725698885823e-05,
+      "loss": 0.7888,
+      "step": 3404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440337653228117e-05,
+      "loss": 21.6847,
+      "step": 3405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439949480977406e-05,
+      "loss": 25.6565,
+      "step": 3406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439561182144756e-05,
+      "loss": 22.9001,
+      "step": 3407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43917275674124e-05,
+      "loss": 21.5391,
+      "step": 3408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438784204777931e-05,
+      "loss": 0.7662,
+      "step": 3409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438395526265903e-05,
+      "loss": 0.758,
+      "step": 3410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438006721216243e-05,
+      "loss": 21.4799,
+      "step": 3411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437617789640033e-05,
+      "loss": 0.8088,
+      "step": 3412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43722873154836e-05,
+      "loss": 26.9426,
+      "step": 3413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436839546952318e-05,
+      "loss": 25.0553,
+      "step": 3414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436450235863002e-05,
+      "loss": 0.7695,
+      "step": 3415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43606079829151e-05,
+      "loss": 0.708,
+      "step": 3416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435671234248945e-05,
+      "loss": 22.2978,
+      "step": 3417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435281543746414e-05,
+      "loss": 25.0914,
+      "step": 3418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43489172679503e-05,
+      "loss": 26.1147,
+      "step": 3419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4345017834059e-05,
+      "loss": 25.8011,
+      "step": 3420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434111713590148e-05,
+      "loss": 20.6516,
+      "step": 3421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433721517358888e-05,
+      "loss": 0.093,
+      "step": 3422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433331194723251e-05,
+      "loss": 22.4545,
+      "step": 3423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432940745694362e-05,
+      "loss": 0.7795,
+      "step": 3424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432550170283352e-05,
+      "loss": 0.0921,
+      "step": 3425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432159468501356e-05,
+      "loss": 0.731,
+      "step": 3426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431768640359515e-05,
+      "loss": 26.0256,
+      "step": 3427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43137768586897e-05,
+      "loss": 0.756,
+      "step": 3428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430986605040866e-05,
+      "loss": 0.789,
+      "step": 3429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430595397886354e-05,
+      "loss": 0.0724,
+      "step": 3430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430204064416588e-05,
+      "loss": 24.5221,
+      "step": 3431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429812604642724e-05,
+      "loss": 23.5534,
+      "step": 3432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429421018575921e-05,
+      "loss": 22.8132,
+      "step": 3433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429029306227344e-05,
+      "loss": 0.7691,
+      "step": 3434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428637467608164e-05,
+      "loss": 0.7822,
+      "step": 3435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428245502729545e-05,
+      "loss": 0.788,
+      "step": 3436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427853411602668e-05,
+      "loss": 24.1601,
+      "step": 3437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427461194238709e-05,
+      "loss": 23.5427,
+      "step": 3438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427068850648852e-05,
+      "loss": 0.779,
+      "step": 3439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42667638084428e-05,
+      "loss": 23.3281,
+      "step": 3440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.426283784836182e-05,
+      "loss": 24.0767,
+      "step": 3441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425891062635753e-05,
+      "loss": 0.7631,
+      "step": 3442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425498214254189e-05,
+      "loss": 24.5962,
+      "step": 3443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425105239702688e-05,
+      "loss": 24.4627,
+      "step": 3444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424712138992458e-05,
+      "loss": 0.7764,
+      "step": 3445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4243189121347e-05,
+      "loss": 26.6332,
+      "step": 3446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423925559140629e-05,
+      "loss": 24.2978,
+      "step": 3447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42353208002146e-05,
+      "loss": 0.7816,
+      "step": 3448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423138474788409e-05,
+      "loss": 23.3197,
+      "step": 3449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422744743452699e-05,
+      "loss": 0.7035,
+      "step": 3450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422350886025552e-05,
+      "loss": 0.7494,
+      "step": 3451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421956902518201e-05,
+      "loss": 23.7333,
+      "step": 3452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421562792941875e-05,
+      "loss": 0.0527,
+      "step": 3453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421168557307814e-05,
+      "loss": 23.911,
+      "step": 3454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420774195627253e-05,
+      "loss": 0.7978,
+      "step": 3455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420379707911437e-05,
+      "loss": 23.3885,
+      "step": 3456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419985094171615e-05,
+      "loss": 22.9685,
+      "step": 3457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419590354419033e-05,
+      "loss": 25.5322,
+      "step": 3458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41919548866495e-05,
+      "loss": 22.7033,
+      "step": 3459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418800496920617e-05,
+      "loss": 0.8176,
+      "step": 3460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418405379197301e-05,
+      "loss": 22.1105,
+      "step": 3461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418010135506266e-05,
+      "loss": 23.4354,
+      "step": 3462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417614765858776e-05,
+      "loss": 0.7685,
+      "step": 3463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417219270266107e-05,
+      "loss": 21.6712,
+      "step": 3464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416823648739535e-05,
+      "loss": 22.2104,
+      "step": 3465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416427901290336e-05,
+      "loss": 0.7507,
+      "step": 3466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416032027929794e-05,
+      "loss": 24.5668,
+      "step": 3467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415636028669194e-05,
+      "loss": 24.5831,
+      "step": 3468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41523990351983e-05,
+      "loss": 0.7324,
+      "step": 3469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414843652492991e-05,
+      "loss": 22.9583,
+      "step": 3470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414447275599975e-05,
+      "loss": 24.335,
+      "step": 3471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414050772852084e-05,
+      "loss": 22.3772,
+      "step": 3472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413654144260621e-05,
+      "loss": 0.7216,
+      "step": 3473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413257389836896e-05,
+      "loss": 0.0711,
+      "step": 3474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412860509592216e-05,
+      "loss": 25.0208,
+      "step": 3475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4124635035379e-05,
+      "loss": 25.0239,
+      "step": 3476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412066371685266e-05,
+      "loss": 23.8856,
+      "step": 3477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411669114045634e-05,
+      "loss": 22.5954,
+      "step": 3478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411271730630331e-05,
+      "loss": 0.0718,
+      "step": 3479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410874221450687e-05,
+      "loss": 24.118,
+      "step": 3480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410476586518033e-05,
+      "loss": 26.9296,
+      "step": 3481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410078825843709e-05,
+      "loss": 20.8101,
+      "step": 3482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40968093943905e-05,
+      "loss": 23.3428,
+      "step": 3483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409282927315403e-05,
+      "loss": 0.0572,
+      "step": 3484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408884789484116e-05,
+      "loss": 0.088,
+      "step": 3485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408486525956537e-05,
+      "loss": 21.9098,
+      "step": 3486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408088136744023e-05,
+      "loss": 21.0769,
+      "step": 3487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40768962185793e-05,
+      "loss": 0.0716,
+      "step": 3488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407290981309622e-05,
+      "loss": 0.7763,
+      "step": 3489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406892215110462e-05,
+      "loss": 22.7055,
+      "step": 3490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406493323271817e-05,
+      "loss": 24.8859,
+      "step": 3491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406094305805063e-05,
+      "loss": 0.8274,
+      "step": 3492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405695162721574e-05,
+      "loss": 0.7112,
+      "step": 3493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40529589403273e-05,
+      "loss": 0.8099,
+      "step": 3494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404896499749915e-05,
+      "loss": 22.54,
+      "step": 3495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404496979884515e-05,
+      "loss": 24.9471,
+      "step": 3496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40409733444792e-05,
+      "loss": 22.2092,
+      "step": 3497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403697563451521e-05,
+      "loss": 0.7589,
+      "step": 3498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40329766690672e-05,
+      "loss": 21.6114,
+      "step": 3499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402897644824918e-05,
+      "loss": 0.8203,
+      "step": 3500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402497497217514e-05,
+      "loss": 23.6048,
+      "step": 3501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402097224095922e-05,
+      "loss": 22.5495,
+      "step": 3502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40169682547155e-05,
+      "loss": 24.1873,
+      "step": 3503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401296301355816e-05,
+      "loss": 25.9099,
+      "step": 3504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400895651760136e-05,
+      "loss": 21.3177,
+      "step": 3505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400494876695934e-05,
+      "loss": 22.3334,
+      "step": 3506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400093976174636e-05,
+      "loss": 24.5624,
+      "step": 3507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399692950207672e-05,
+      "loss": 0.7236,
+      "step": 3508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399291798806476e-05,
+      "loss": 0.7328,
+      "step": 3509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398890521982482e-05,
+      "loss": 0.7088,
+      "step": 3510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398489119747132e-05,
+      "loss": 27.3129,
+      "step": 3511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398087592111868e-05,
+      "loss": 25.4092,
+      "step": 3512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397685939088141e-05,
+      "loss": 23.3943,
+      "step": 3513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397284160687398e-05,
+      "loss": 24.5009,
+      "step": 3514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396882256921098e-05,
+      "loss": 0.0783,
+      "step": 3515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396480227800696e-05,
+      "loss": 0.7194,
+      "step": 3516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396078073337655e-05,
+      "loss": 23.4059,
+      "step": 3517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395675793543439e-05,
+      "loss": 25.0117,
+      "step": 3518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39527338842952e-05,
+      "loss": 0.101,
+      "step": 3519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394870858007366e-05,
+      "loss": 23.2222,
+      "step": 3520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394468202288457e-05,
+      "loss": 0.7255,
+      "step": 3521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394065421284271e-05,
+      "loss": 23.3698,
+      "step": 3522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393662515006289e-05,
+      "loss": 24.2951,
+      "step": 3523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393259483466001e-05,
+      "loss": 23.5618,
+      "step": 3524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392856326674898e-05,
+      "loss": 23.0342,
+      "step": 3525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392453044644471e-05,
+      "loss": 0.7016,
+      "step": 3526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392049637386219e-05,
+      "loss": 0.7972,
+      "step": 3527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391646104911642e-05,
+      "loss": 25.1551,
+      "step": 3528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391242447232246e-05,
+      "loss": 23.5398,
+      "step": 3529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390838664359538e-05,
+      "loss": 23.3427,
+      "step": 3530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390434756305031e-05,
+      "loss": 0.7589,
+      "step": 3531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39003072308024e-05,
+      "loss": 24.3342,
+      "step": 3532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.389626564696684e-05,
+      "loss": 0.7717,
+      "step": 3533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.389222281165883e-05,
+      "loss": 0.0604,
+      "step": 3534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388817872499366e-05,
+      "loss": 22.9753,
+      "step": 3535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388413338708661e-05,
+      "loss": 0.8171,
+      "step": 3536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388008679805303e-05,
+      "loss": 0.7888,
+      "step": 3537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387603895800827e-05,
+      "loss": 21.6553,
+      "step": 3538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387198986706773e-05,
+      "loss": 22.7893,
+      "step": 3539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386793952534687e-05,
+      "loss": 25.2789,
+      "step": 3540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386388793296115e-05,
+      "loss": 0.7362,
+      "step": 3541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385983509002607e-05,
+      "loss": 23.7423,
+      "step": 3542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38557809966572e-05,
+      "loss": 23.0255,
+      "step": 3543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385172565297009e-05,
+      "loss": 23.5819,
+      "step": 3544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38476690590804e-05,
+      "loss": 0.7665,
+      "step": 3545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384361121510373e-05,
+      "loss": 0.7446,
+      "step": 3546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.383955212115582e-05,
+      "loss": 0.0816,
+      "step": 3547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.383549177735236e-05,
+      "loss": 25.5089,
+      "step": 3548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.383143018380911e-05,
+      "loss": 24.1274,
+      "step": 3549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.382736734064188e-05,
+      "loss": 0.8153,
+      "step": 3550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.38233032479665e-05,
+      "loss": 23.6417,
+      "step": 3551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381923790589883e-05,
+      "loss": 22.67,
+      "step": 3552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381517131455477e-05,
+      "loss": 23.1279,
+      "step": 3553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381110347405026e-05,
+      "loss": 22.1309,
+      "step": 3554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380703438450127e-05,
+      "loss": 23.2188,
+      "step": 3555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380296404602382e-05,
+      "loss": 24.2259,
+      "step": 3556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379889245873395e-05,
+      "loss": 24.1022,
+      "step": 3557
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379481962274774e-05,
+      "loss": 23.3479,
+      "step": 3558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379074553818129e-05,
+      "loss": 0.7676,
+      "step": 3559
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378667020515078e-05,
+      "loss": 0.7681,
+      "step": 3560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378259362377236e-05,
+      "loss": 20.1562,
+      "step": 3561
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377851579416229e-05,
+      "loss": 23.2813,
+      "step": 3562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37744367164368e-05,
+      "loss": 21.7953,
+      "step": 3563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37703563907122e-05,
+      "loss": 24.0205,
+      "step": 3564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376627481710482e-05,
+      "loss": 24.5868,
+      "step": 3565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376219199573101e-05,
+      "loss": 0.8125,
+      "step": 3566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375810792670718e-05,
+      "loss": 22.1008,
+      "step": 3567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375402261014977e-05,
+      "loss": 0.0737,
+      "step": 3568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374993604617524e-05,
+      "loss": 24.8585,
+      "step": 3569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374584823490009e-05,
+      "loss": 24.4554,
+      "step": 3570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37417591764409e-05,
+      "loss": 0.7166,
+      "step": 3571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373766887091422e-05,
+      "loss": 23.4041,
+      "step": 3572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373357731843666e-05,
+      "loss": 22.6406,
+      "step": 3573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372948451912488e-05,
+      "loss": 25.9909,
+      "step": 3574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372539047309556e-05,
+      "loss": 22.2047,
+      "step": 3575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372129518046542e-05,
+      "loss": 24.315,
+      "step": 3576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371719864135122e-05,
+      "loss": 0.7388,
+      "step": 3577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371310085586975e-05,
+      "loss": 23.536,
+      "step": 3578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370900182413784e-05,
+      "loss": 0.6478,
+      "step": 3579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370490154627234e-05,
+      "loss": 25.206,
+      "step": 3580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370080002239017e-05,
+      "loss": 22.0838,
+      "step": 3581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369669725260824e-05,
+      "loss": 0.8283,
+      "step": 3582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369259323704351e-05,
+      "loss": 22.5329,
+      "step": 3583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368848797581304e-05,
+      "loss": 22.3321,
+      "step": 3584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36843814690338e-05,
+      "loss": 23.447,
+      "step": 3585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368027371682293e-05,
+      "loss": 21.5552,
+      "step": 3586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367616471929749e-05,
+      "loss": 0.1029,
+      "step": 3587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367205447657465e-05,
+      "loss": 23.9849,
+      "step": 3588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.366794298877159e-05,
+      "loss": 22.4645,
+      "step": 3589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.366383025600554e-05,
+      "loss": 24.1374,
+      "step": 3590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365971627839371e-05,
+      "loss": 23.2936,
+      "step": 3591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365560105605345e-05,
+      "loss": 0.7715,
+      "step": 3592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365148458910203e-05,
+      "loss": 22.2231,
+      "step": 3593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364736687765684e-05,
+      "loss": 23.6607,
+      "step": 3594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364324792183527e-05,
+      "loss": 25.9776,
+      "step": 3595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363912772175474e-05,
+      "loss": 23.2222,
+      "step": 3596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363500627753273e-05,
+      "loss": 0.0667,
+      "step": 3597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363088358928672e-05,
+      "loss": 23.3887,
+      "step": 3598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362675965713428e-05,
+      "loss": 0.7701,
+      "step": 3599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362263448119295e-05,
+      "loss": 0.786,
+      "step": 3600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361850806158035e-05,
+      "loss": 0.7635,
+      "step": 3601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361438039841413e-05,
+      "loss": 22.9683,
+      "step": 3602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361025149181197e-05,
+      "loss": 0.7342,
+      "step": 3603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360612134189157e-05,
+      "loss": 0.7688,
+      "step": 3604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360198994877069e-05,
+      "loss": 0.0815,
+      "step": 3605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.359785731256712e-05,
+      "loss": 22.1887,
+      "step": 3606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.359372343339866e-05,
+      "loss": 0.0873,
+      "step": 3607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358958831138319e-05,
+      "loss": 0.7738,
+      "step": 3608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358545194663858e-05,
+      "loss": 0.8063,
+      "step": 3609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358131433928276e-05,
+      "loss": 23.6467,
+      "step": 3610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357717548943372e-05,
+      "loss": 22.6066,
+      "step": 3611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357303539720944e-05,
+      "loss": 23.975,
+      "step": 3612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356889406272793e-05,
+      "loss": 21.2769,
+      "step": 3613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356475148610728e-05,
+      "loss": 0.7369,
+      "step": 3614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35606076674656e-05,
+      "loss": 21.4967,
+      "step": 3615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355646260692101e-05,
+      "loss": 24.4716,
+      "step": 3616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355231630459172e-05,
+      "loss": 22.2227,
+      "step": 3617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35481687605959e-05,
+      "loss": 0.77,
+      "step": 3618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354401997505182e-05,
+      "loss": 0.7512,
+      "step": 3619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353986994807775e-05,
+      "loss": 0.7648,
+      "step": 3620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3535718679792e-05,
+      "loss": 0.7548,
+      "step": 3621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353156617031294e-05,
+      "loss": 21.234,
+      "step": 3622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352741241975895e-05,
+      "loss": 0.7673,
+      "step": 3623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352325742824844e-05,
+      "loss": 0.7766,
+      "step": 3624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351910119589988e-05,
+      "loss": 0.7175,
+      "step": 3625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351494372283179e-05,
+      "loss": 21.0275,
+      "step": 3626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351078500916263e-05,
+      "loss": 26.0129,
+      "step": 3627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350662505501103e-05,
+      "loss": 0.0719,
+      "step": 3628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350246386049555e-05,
+      "loss": 24.9036,
+      "step": 3629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349830142573484e-05,
+      "loss": 0.7591,
+      "step": 3630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349413775084758e-05,
+      "loss": 23.2664,
+      "step": 3631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348997283595245e-05,
+      "loss": 23.5777,
+      "step": 3632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348580668116822e-05,
+      "loss": 24.236,
+      "step": 3633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348163928661363e-05,
+      "loss": 24.0055,
+      "step": 3634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347747065240752e-05,
+      "loss": 23.5968,
+      "step": 3635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347330077866874e-05,
+      "loss": 23.4482,
+      "step": 3636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.346912966551615e-05,
+      "loss": 21.6251,
+      "step": 3637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.346495731306869e-05,
+      "loss": 24.5137,
+      "step": 3638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.346078372144529e-05,
+      "loss": 0.7615,
+      "step": 3639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345660889076495e-05,
+      "loss": 0.7512,
+      "step": 3640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345243282114668e-05,
+      "loss": 23.562,
+      "step": 3641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344825551270957e-05,
+      "loss": 0.7926,
+      "step": 3642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34440769655727e-05,
+      "loss": 19.7881,
+      "step": 3643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34398971798552e-05,
+      "loss": 23.1212,
+      "step": 3644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343571615567621e-05,
+      "loss": 21.8208,
+      "step": 3645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343153389315495e-05,
+      "loss": 0.0945,
+      "step": 3646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342735039241067e-05,
+      "loss": 24.3505,
+      "step": 3647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342316565356262e-05,
+      "loss": 23.414,
+      "step": 3648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34189796767301e-05,
+      "loss": 22.7509,
+      "step": 3649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341479246203249e-05,
+      "loss": 25.8968,
+      "step": 3650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34106040095891e-05,
+      "loss": 22.8293,
+      "step": 3651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340641431951941e-05,
+      "loss": 24.2829,
+      "step": 3652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340222339194284e-05,
+      "loss": 22.6984,
+      "step": 3653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.339803122697886e-05,
+      "loss": 24.1403,
+      "step": 3654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3393837824747e-05,
+      "loss": 0.787,
+      "step": 3655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338964318536682e-05,
+      "loss": 0.7558,
+      "step": 3656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33854473089579e-05,
+      "loss": 23.191,
+      "step": 3657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338125019563986e-05,
+      "loss": 23.101,
+      "step": 3658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337705184553235e-05,
+      "loss": 22.5795,
+      "step": 3659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337285225875509e-05,
+      "loss": 23.2607,
+      "step": 3660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33686514354278e-05,
+      "loss": 0.6902,
+      "step": 3661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.336444937567022e-05,
+      "loss": 26.4021,
+      "step": 3662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.336024607960217e-05,
+      "loss": 24.7708,
+      "step": 3663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335604154734349e-05,
+      "loss": 0.7675,
+      "step": 3664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335183577901407e-05,
+      "loss": 22.527,
+      "step": 3665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334762877473375e-05,
+      "loss": 21.6963,
+      "step": 3666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334342053462254e-05,
+      "loss": 24.1416,
+      "step": 3667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333921105880038e-05,
+      "loss": 0.8445,
+      "step": 3668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333500034738728e-05,
+      "loss": 0.7637,
+      "step": 3669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33307884005033e-05,
+      "loss": 25.5562,
+      "step": 3670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332657521826852e-05,
+      "loss": 0.7628,
+      "step": 3671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332236080080305e-05,
+      "loss": 0.7512,
+      "step": 3672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331814514822703e-05,
+      "loss": 0.7586,
+      "step": 3673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331392826066067e-05,
+      "loss": 0.7749,
+      "step": 3674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330971013822421e-05,
+      "loss": 22.1512,
+      "step": 3675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330549078103785e-05,
+      "loss": 0.0831,
+      "step": 3676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330127018922194e-05,
+      "loss": 21.9748,
+      "step": 3677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329704836289676e-05,
+      "loss": 0.6766,
+      "step": 3678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329282530218271e-05,
+      "loss": 0.0644,
+      "step": 3679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328860100720019e-05,
+      "loss": 23.9774,
+      "step": 3680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32843754780696e-05,
+      "loss": 23.2269,
+      "step": 3681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328014871491141e-05,
+      "loss": 23.8205,
+      "step": 3682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327592071784619e-05,
+      "loss": 22.392,
+      "step": 3683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32716914869944e-05,
+      "loss": 25.534,
+      "step": 3684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326746102247664e-05,
+      "loss": 22.5842,
+      "step": 3685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326322932441354e-05,
+      "loss": 0.7968,
+      "step": 3686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325899639292572e-05,
+      "loss": 23.3327,
+      "step": 3687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325476222813388e-05,
+      "loss": 22.9311,
+      "step": 3688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325052683015871e-05,
+      "loss": 0.7408,
+      "step": 3689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3246290199121e-05,
+      "loss": 0.7905,
+      "step": 3690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324205233514148e-05,
+      "loss": 20.4458,
+      "step": 3691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323781323834102e-05,
+      "loss": 0.7399,
+      "step": 3692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323357290884045e-05,
+      "loss": 24.3519,
+      "step": 3693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322933134676067e-05,
+      "loss": 24.2591,
+      "step": 3694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32250885522226e-05,
+      "loss": 0.775,
+      "step": 3695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32208445253472e-05,
+      "loss": 22.3747,
+      "step": 3696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321659926625548e-05,
+      "loss": 0.7909,
+      "step": 3697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321235277506846e-05,
+      "loss": 0.7346,
+      "step": 3698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320810505190723e-05,
+      "loss": 21.2587,
+      "step": 3699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320385609689283e-05,
+      "loss": 23.1923,
+      "step": 3700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319960591014646e-05,
+      "loss": 0.8046,
+      "step": 3701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319535449178927e-05,
+      "loss": 0.7725,
+      "step": 3702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319110184194247e-05,
+      "loss": 0.0864,
+      "step": 3703
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318684796072727e-05,
+      "loss": 0.0681,
+      "step": 3704
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3182592848265e-05,
+      "loss": 0.7763,
+      "step": 3705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317833650467693e-05,
+      "loss": 24.4182,
+      "step": 3706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317407893008445e-05,
+      "loss": 0.0636,
+      "step": 3707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316982012460889e-05,
+      "loss": 23.1916,
+      "step": 3708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316556008837171e-05,
+      "loss": 0.7605,
+      "step": 3709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316129882149433e-05,
+      "loss": 0.7425,
+      "step": 3710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315703632409827e-05,
+      "loss": 25.599,
+      "step": 3711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315277259630504e-05,
+      "loss": 0.7982,
+      "step": 3712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31485076382362e-05,
+      "loss": 22.9605,
+      "step": 3713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314424145001333e-05,
+      "loss": 0.7089,
+      "step": 3714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313997403175807e-05,
+      "loss": 23.7143,
+      "step": 3715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313570538359208e-05,
+      "loss": 0.7879,
+      "step": 3716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313143550563706e-05,
+      "loss": 0.065,
+      "step": 3717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312716439801472e-05,
+      "loss": 21.8182,
+      "step": 3718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312289206084687e-05,
+      "loss": 22.2147,
+      "step": 3719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31186184942553e-05,
+      "loss": 0.7742,
+      "step": 3720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311434369836182e-05,
+      "loss": 0.7713,
+      "step": 3721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311006767328834e-05,
+      "loss": 0.7278,
+      "step": 3722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310579041915675e-05,
+      "loss": 22.9002,
+      "step": 3723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310151193608898e-05,
+      "loss": 0.742,
+      "step": 3724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309723222420705e-05,
+      "loss": 0.0857,
+      "step": 3725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309295128363293e-05,
+      "loss": 22.6471,
+      "step": 3726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30886691144887e-05,
+      "loss": 0.0756,
+      "step": 3727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308438571689644e-05,
+      "loss": 0.6978,
+      "step": 3728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308010109097825e-05,
+      "loss": 22.8204,
+      "step": 3729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307581523685628e-05,
+      "loss": 22.0431,
+      "step": 3730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307152815465275e-05,
+      "loss": 22.3844,
+      "step": 3731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.306723984448984e-05,
+      "loss": 23.143,
+      "step": 3732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.306295030648987e-05,
+      "loss": 24.8002,
+      "step": 3733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305865954077508e-05,
+      "loss": 24.2992,
+      "step": 3734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305436754746782e-05,
+      "loss": 0.7212,
+      "step": 3735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305007432669046e-05,
+      "loss": 23.9884,
+      "step": 3736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.304577987856537e-05,
+      "loss": 22.5943,
+      "step": 3737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.304148420321502e-05,
+      "loss": 20.5245,
+      "step": 3738
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.303718730076185e-05,
+      "loss": 20.3097,
+      "step": 3739
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.303288917132839e-05,
+      "loss": 23.4266,
+      "step": 3740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302858981503716e-05,
+      "loss": 0.7688,
+      "step": 3741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302428923201075e-05,
+      "loss": 21.3006,
+      "step": 3742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301998742237173e-05,
+      "loss": 19.8763,
+      "step": 3743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301568438624278e-05,
+      "loss": 0.7914,
+      "step": 3744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301138012374657e-05,
+      "loss": 23.7171,
+      "step": 3745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300707463500582e-05,
+      "loss": 0.7859,
+      "step": 3746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300276792014326e-05,
+      "loss": 23.7202,
+      "step": 3747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29984599792817e-05,
+      "loss": 0.119,
+      "step": 3748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299415081254393e-05,
+      "loss": 23.4435,
+      "step": 3749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29898404200528e-05,
+      "loss": 23.291,
+      "step": 3750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298552880193126e-05,
+      "loss": 0.7318,
+      "step": 3751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298121595830215e-05,
+      "loss": 0.786,
+      "step": 3752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297690188928847e-05,
+      "loss": 22.8519,
+      "step": 3753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297258659501322e-05,
+      "loss": 22.5946,
+      "step": 3754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296827007559941e-05,
+      "loss": 0.0782,
+      "step": 3755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296395233117012e-05,
+      "loss": 24.0507,
+      "step": 3756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295963336184843e-05,
+      "loss": 0.7302,
+      "step": 3757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295531316775748e-05,
+      "loss": 22.3512,
+      "step": 3758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295099174902044e-05,
+      "loss": 23.2317,
+      "step": 3759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294666910576052e-05,
+      "loss": 0.0814,
+      "step": 3760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294234523810094e-05,
+      "loss": 0.7249,
+      "step": 3761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293802014616498e-05,
+      "loss": 21.6942,
+      "step": 3762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293369383007595e-05,
+      "loss": 23.1407,
+      "step": 3763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29293662899572e-05,
+      "loss": 0.7572,
+      "step": 3764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292503752593209e-05,
+      "loss": 0.7664,
+      "step": 3765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292070753812403e-05,
+      "loss": 0.7991,
+      "step": 3766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29163763266565e-05,
+      "loss": 0.0694,
+      "step": 3767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291204389165295e-05,
+      "loss": 0.7802,
+      "step": 3768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29077102332369e-05,
+      "loss": 0.747,
+      "step": 3769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290337535153191e-05,
+      "loss": 0.781,
+      "step": 3770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289903924666155e-05,
+      "loss": 0.7632,
+      "step": 3771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289470191874948e-05,
+      "loss": 25.4792,
+      "step": 3772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289036336791932e-05,
+      "loss": 22.1352,
+      "step": 3773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288602359429476e-05,
+      "loss": 0.7889,
+      "step": 3774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288168259799956e-05,
+      "loss": 0.7677,
+      "step": 3775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287734037915745e-05,
+      "loss": 24.8844,
+      "step": 3776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287299693789223e-05,
+      "loss": 21.2861,
+      "step": 3777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286865227432775e-05,
+      "loss": 23.0109,
+      "step": 3778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286430638858785e-05,
+      "loss": 22.2597,
+      "step": 3779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285995928079644e-05,
+      "loss": 22.8339,
+      "step": 3780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285561095107745e-05,
+      "loss": 22.249,
+      "step": 3781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285126139955487e-05,
+      "loss": 0.7295,
+      "step": 3782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284691062635268e-05,
+      "loss": 21.7468,
+      "step": 3783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284255863159494e-05,
+      "loss": 21.5102,
+      "step": 3784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28382054154057e-05,
+      "loss": 20.2784,
+      "step": 3785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283385097790908e-05,
+      "loss": 24.0997,
+      "step": 3786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282949531922925e-05,
+      "loss": 24.801,
+      "step": 3787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282513843949036e-05,
+      "loss": 20.0237,
+      "step": 3788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282078033881663e-05,
+      "loss": 22.9771,
+      "step": 3789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28164210173323e-05,
+      "loss": 0.0812,
+      "step": 3790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281206047516167e-05,
+      "loss": 22.7347,
+      "step": 3791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280769871242904e-05,
+      "loss": 21.0721,
+      "step": 3792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280333572925879e-05,
+      "loss": 22.9754,
+      "step": 3793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279897152577528e-05,
+      "loss": 0.7645,
+      "step": 3794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279460610210293e-05,
+      "loss": 21.7077,
+      "step": 3795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279023945836622e-05,
+      "loss": 22.8654,
+      "step": 3796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278587159468964e-05,
+      "loss": 22.1122,
+      "step": 3797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278150251119771e-05,
+      "loss": 23.7711,
+      "step": 3798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277713220801499e-05,
+      "loss": 0.7496,
+      "step": 3799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277276068526606e-05,
+      "loss": 22.208,
+      "step": 3800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.276838794307559e-05,
+      "loss": 22.5678,
+      "step": 3801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27640139815682e-05,
+      "loss": 0.8225,
+      "step": 3802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275963880086862e-05,
+      "loss": 22.2499,
+      "step": 3803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27552624011016e-05,
+      "loss": 21.6143,
+      "step": 3804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275088478239186e-05,
+      "loss": 0.7202,
+      "step": 3805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274650594486426e-05,
+      "loss": 21.159,
+      "step": 3806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27421258886436e-05,
+      "loss": 23.8074,
+      "step": 3807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273774461385478e-05,
+      "loss": 24.2075,
+      "step": 3808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27333621206227e-05,
+      "loss": 24.5898,
+      "step": 3809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272897840907229e-05,
+      "loss": 0.0649,
+      "step": 3810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272459347932854e-05,
+      "loss": 22.6024,
+      "step": 3811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272020733151646e-05,
+      "loss": 0.6913,
+      "step": 3812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271581996576111e-05,
+      "loss": 0.7162,
+      "step": 3813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271143138218757e-05,
+      "loss": 0.7332,
+      "step": 3814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270704158092094e-05,
+      "loss": 0.7822,
+      "step": 3815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27026505620864e-05,
+      "loss": 0.097,
+      "step": 3816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269825832580912e-05,
+      "loss": 0.7734,
+      "step": 3817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26938648722143e-05,
+      "loss": 25.0848,
+      "step": 3818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268947020142724e-05,
+      "loss": 0.7316,
+      "step": 3819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268507431357321e-05,
+      "loss": 22.4909,
+      "step": 3820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268067720877753e-05,
+      "loss": 0.7804,
+      "step": 3821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267627888716556e-05,
+      "loss": 0.7635,
+      "step": 3822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267187934886271e-05,
+      "loss": 0.7177,
+      "step": 3823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26674785939944e-05,
+      "loss": 0.0579,
+      "step": 3824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266307662268609e-05,
+      "loss": 24.777,
+      "step": 3825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265867343506331e-05,
+      "loss": 22.7519,
+      "step": 3826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265426903125154e-05,
+      "loss": 22.1239,
+      "step": 3827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26498634113764e-05,
+      "loss": 24.6044,
+      "step": 3828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264545657556345e-05,
+      "loss": 0.7786,
+      "step": 3829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264104852393836e-05,
+      "loss": 23.6292,
+      "step": 3830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263663925662679e-05,
+      "loss": 21.7651,
+      "step": 3831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263222877375445e-05,
+      "loss": 0.0864,
+      "step": 3832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262781707544708e-05,
+      "loss": 23.3219,
+      "step": 3833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262340416183045e-05,
+      "loss": 24.458,
+      "step": 3834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261899003303038e-05,
+      "loss": 22.8841,
+      "step": 3835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261457468917272e-05,
+      "loss": 0.7594,
+      "step": 3836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261015813038334e-05,
+      "loss": 0.7749,
+      "step": 3837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260574035678816e-05,
+      "loss": 0.8225,
+      "step": 3838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260132136851314e-05,
+      "loss": 0.7637,
+      "step": 3839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259690116568424e-05,
+      "loss": 22.4345,
+      "step": 3840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259247974842749e-05,
+      "loss": 22.6243,
+      "step": 3841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.258805711686895e-05,
+      "loss": 0.7312,
+      "step": 3842
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25836332711347e-05,
+      "loss": 0.8012,
+      "step": 3843
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257920821135088e-05,
+      "loss": 0.7378,
+      "step": 3844
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257478193764362e-05,
+      "loss": 22.8407,
+      "step": 3845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257035445013912e-05,
+      "loss": 23.7503,
+      "step": 3846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256592574896363e-05,
+      "loss": 21.7598,
+      "step": 3847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256149583424341e-05,
+      "loss": 24.5096,
+      "step": 3848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25570647061047e-05,
+      "loss": 0.7751,
+      "step": 3849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25526323646739e-05,
+      "loss": 0.8152,
+      "step": 3850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254819881007734e-05,
+      "loss": 21.7894,
+      "step": 3851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254376404244141e-05,
+      "loss": 22.2754,
+      "step": 3852
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253932806189257e-05,
+      "loss": 0.7697,
+      "step": 3853
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253489086855729e-05,
+      "loss": 20.4985,
+      "step": 3854
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253045246256204e-05,
+      "loss": 23.0995,
+      "step": 3855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252601284403339e-05,
+      "loss": 22.3813,
+      "step": 3856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25215720130979e-05,
+      "loss": 0.7372,
+      "step": 3857
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251712996988218e-05,
+      "loss": 20.8073,
+      "step": 3858
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251268671451288e-05,
+      "loss": 21.6271,
+      "step": 3859
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.250824224711666e-05,
+      "loss": 24.179,
+      "step": 3860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.250379656782023e-05,
+      "loss": 21.7099,
+      "step": 3861
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249934967675035e-05,
+      "loss": 22.3233,
+      "step": 3862
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249490157403378e-05,
+      "loss": 23.151,
+      "step": 3863
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249045225979737e-05,
+      "loss": 20.4529,
+      "step": 3864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248600173416793e-05,
+      "loss": 22.6072,
+      "step": 3865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248154999727236e-05,
+      "loss": 0.0874,
+      "step": 3866
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24770970492376e-05,
+      "loss": 21.4445,
+      "step": 3867
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247264289019055e-05,
+      "loss": 22.5828,
+      "step": 3868
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.246818752025823e-05,
+      "loss": 22.397,
+      "step": 3869
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.246373093956769e-05,
+      "loss": 22.3505,
+      "step": 3870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245927314824591e-05,
+      "loss": 0.7664,
+      "step": 3871
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245481414642006e-05,
+      "loss": 0.7554,
+      "step": 3872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245035393421722e-05,
+      "loss": 0.7495,
+      "step": 3873
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244589251176455e-05,
+      "loss": 0.725,
+      "step": 3874
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244142987918927e-05,
+      "loss": 24.4821,
+      "step": 3875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.243696603661857e-05,
+      "loss": 26.3296,
+      "step": 3876
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.243250098417977e-05,
+      "loss": 24.4182,
+      "step": 3877
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24280347220001e-05,
+      "loss": 20.7435,
+      "step": 3878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.242356725020695e-05,
+      "loss": 22.6835,
+      "step": 3879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241909856892765e-05,
+      "loss": 22.0182,
+      "step": 3880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241462867828962e-05,
+      "loss": 0.7309,
+      "step": 3881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241015757842029e-05,
+      "loss": 23.0717,
+      "step": 3882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240568526944712e-05,
+      "loss": 0.8163,
+      "step": 3883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240121175149763e-05,
+      "loss": 21.5804,
+      "step": 3884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239673702469934e-05,
+      "loss": 24.2152,
+      "step": 3885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239226108917985e-05,
+      "loss": 22.3433,
+      "step": 3886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238778394506675e-05,
+      "loss": 21.7232,
+      "step": 3887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238330559248769e-05,
+      "loss": 21.8389,
+      "step": 3888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237882603157034e-05,
+      "loss": 0.7589,
+      "step": 3889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237434526244241e-05,
+      "loss": 21.9398,
+      "step": 3890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236986328523165e-05,
+      "loss": 23.2354,
+      "step": 3891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236538010006584e-05,
+      "loss": 0.7446,
+      "step": 3892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23608957070728e-05,
+      "loss": 20.5411,
+      "step": 3893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235641010638037e-05,
+      "loss": 23.3832,
+      "step": 3894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235192329811644e-05,
+      "loss": 21.0844,
+      "step": 3895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.234743528240893e-05,
+      "loss": 22.2719,
+      "step": 3896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.234294605938577e-05,
+      "loss": 21.9605,
+      "step": 3897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233845562917498e-05,
+      "loss": 22.4285,
+      "step": 3898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233396399190454e-05,
+      "loss": 20.981,
+      "step": 3899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.232947114770256e-05,
+      "loss": 0.7947,
+      "step": 3900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23249770966971e-05,
+      "loss": 23.6429,
+      "step": 3901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.232048183901627e-05,
+      "loss": 0.7676,
+      "step": 3902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231598537478825e-05,
+      "loss": 0.0862,
+      "step": 3903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231148770414122e-05,
+      "loss": 0.7728,
+      "step": 3904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230698882720341e-05,
+      "loss": 0.8086,
+      "step": 3905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23024887441031e-05,
+      "loss": 24.9093,
+      "step": 3906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229798745496857e-05,
+      "loss": 23.9482,
+      "step": 3907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229348495992815e-05,
+      "loss": 0.7585,
+      "step": 3908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22889812591102e-05,
+      "loss": 23.6585,
+      "step": 3909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228447635264315e-05,
+      "loss": 0.7393,
+      "step": 3910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22799702406554e-05,
+      "loss": 0.7874,
+      "step": 3911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.227546292327543e-05,
+      "loss": 24.236,
+      "step": 3912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.227095440063174e-05,
+      "loss": 23.276,
+      "step": 3913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226644467285287e-05,
+      "loss": 24.3473,
+      "step": 3914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226193374006739e-05,
+      "loss": 0.754,
+      "step": 3915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225742160240392e-05,
+      "loss": 21.5883,
+      "step": 3916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225290825999107e-05,
+      "loss": 19.7613,
+      "step": 3917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224839371295754e-05,
+      "loss": 22.5724,
+      "step": 3918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224387796143202e-05,
+      "loss": 0.7705,
+      "step": 3919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223936100554327e-05,
+      "loss": 0.7818,
+      "step": 3920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223484284542007e-05,
+      "loss": 0.7936,
+      "step": 3921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22303234811912e-05,
+      "loss": 25.1067,
+      "step": 3922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222580291298553e-05,
+      "loss": 21.9181,
+      "step": 3923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222128114093196e-05,
+      "loss": 0.2556,
+      "step": 3924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221675816515937e-05,
+      "loss": 21.8613,
+      "step": 3925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221223398579671e-05,
+      "loss": 0.7618,
+      "step": 3926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220770860297299e-05,
+      "loss": 0.7294,
+      "step": 3927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220318201681721e-05,
+      "loss": 0.7476,
+      "step": 3928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219865422745841e-05,
+      "loss": 23.3271,
+      "step": 3929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219412523502571e-05,
+      "loss": 0.7607,
+      "step": 3930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.218959503964822e-05,
+      "loss": 22.2411,
+      "step": 3931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.218506364145506e-05,
+      "loss": 21.3343,
+      "step": 3932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.218053104057546e-05,
+      "loss": 19.4436,
+      "step": 3933
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217599723713863e-05,
+      "loss": 0.749,
+      "step": 3934
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217146223127381e-05,
+      "loss": 21.6289,
+      "step": 3935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216692602311032e-05,
+      "loss": 0.7172,
+      "step": 3936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216238861277749e-05,
+      "loss": 23.6848,
+      "step": 3937
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.215785000040465e-05,
+      "loss": 0.1825,
+      "step": 3938
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.215331018612118e-05,
+      "loss": 22.8486,
+      "step": 3939
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21487691700566e-05,
+      "loss": 20.4742,
+      "step": 3940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214422695234027e-05,
+      "loss": 23.2929,
+      "step": 3941
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213968353310174e-05,
+      "loss": 0.7505,
+      "step": 3942
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213513891247051e-05,
+      "loss": 0.8319,
+      "step": 3943
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213059309057622e-05,
+      "loss": 0.6826,
+      "step": 3944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212604606754838e-05,
+      "loss": 22.2575,
+      "step": 3945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212149784351666e-05,
+      "loss": 23.9254,
+      "step": 3946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.211694841861073e-05,
+      "loss": 0.1156,
+      "step": 3947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.211239779296033e-05,
+      "loss": 0.7146,
+      "step": 3948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210784596669512e-05,
+      "loss": 21.3633,
+      "step": 3949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210329293994495e-05,
+      "loss": 22.3866,
+      "step": 3950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209873871283955e-05,
+      "loss": 24.6425,
+      "step": 3951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209418328550883e-05,
+      "loss": 22.6124,
+      "step": 3952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208962665808262e-05,
+      "loss": 22.787,
+      "step": 3953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208506883069086e-05,
+      "loss": 0.7651,
+      "step": 3954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208050980346345e-05,
+      "loss": 22.0309,
+      "step": 3955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20759495765304e-05,
+      "loss": 22.1669,
+      "step": 3956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207138815002172e-05,
+      "loss": 0.7575,
+      "step": 3957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206682552406745e-05,
+      "loss": 21.1116,
+      "step": 3958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206226169879764e-05,
+      "loss": 22.377,
+      "step": 3959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205769667434246e-05,
+      "loss": 0.7267,
+      "step": 3960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205313045083202e-05,
+      "loss": 0.086,
+      "step": 3961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.204856302839651e-05,
+      "loss": 0.7618,
+      "step": 3962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.204399440716613e-05,
+      "loss": 24.9936,
+      "step": 3963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203942458727116e-05,
+      "loss": 24.8067,
+      "step": 3964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203485356884188e-05,
+      "loss": 0.0855,
+      "step": 3965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203028135200857e-05,
+      "loss": 22.8397,
+      "step": 3966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202570793690163e-05,
+      "loss": 0.7577,
+      "step": 3967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202113332365145e-05,
+      "loss": 0.7237,
+      "step": 3968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.201655751238841e-05,
+      "loss": 23.1856,
+      "step": 3969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.2011980503243e-05,
+      "loss": 23.5095,
+      "step": 3970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200740229634568e-05,
+      "loss": 21.4584,
+      "step": 3971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200282289182699e-05,
+      "loss": 0.0761,
+      "step": 3972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199824228981749e-05,
+      "loss": 21.3954,
+      "step": 3973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199366049044777e-05,
+      "loss": 24.9424,
+      "step": 3974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198907749384846e-05,
+      "loss": 0.7185,
+      "step": 3975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19844933001502e-05,
+      "loss": 22.8009,
+      "step": 3976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197990790948373e-05,
+      "loss": 0.097,
+      "step": 3977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197532132197972e-05,
+      "loss": 23.4753,
+      "step": 3978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197073353776897e-05,
+      "loss": 24.1204,
+      "step": 3979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196614455698228e-05,
+      "loss": 0.6895,
+      "step": 3980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196155437975045e-05,
+      "loss": 23.378,
+      "step": 3981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195696300620436e-05,
+      "loss": 0.7252,
+      "step": 3982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195237043647493e-05,
+      "loss": 0.0783,
+      "step": 3983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194777667069306e-05,
+      "loss": 0.7431,
+      "step": 3984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194318170898972e-05,
+      "loss": 20.9019,
+      "step": 3985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193858555149595e-05,
+      "loss": 0.077,
+      "step": 3986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193398819834275e-05,
+      "loss": 0.7093,
+      "step": 3987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19293896496612e-05,
+      "loss": 23.2816,
+      "step": 3988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19247899055824e-05,
+      "loss": 0.7383,
+      "step": 3989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.192018896623749e-05,
+      "loss": 0.0678,
+      "step": 3990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191558683175764e-05,
+      "loss": 22.9694,
+      "step": 3991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191098350227405e-05,
+      "loss": 21.7949,
+      "step": 3992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.190637897791797e-05,
+      "loss": 21.2732,
+      "step": 3993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.190177325882068e-05,
+      "loss": 0.0744,
+      "step": 3994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189716634511347e-05,
+      "loss": 23.4869,
+      "step": 3995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18925582369277e-05,
+      "loss": 23.5022,
+      "step": 3996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188794893439474e-05,
+      "loss": 21.6774,
+      "step": 3997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188333843764598e-05,
+      "loss": 20.5033,
+      "step": 3998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.187872674681289e-05,
+      "loss": 0.0817,
+      "step": 3999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.187411386202694e-05,
+      "loss": 25.5257,
+      "step": 4000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186949978341964e-05,
+      "loss": 0.781,
+      "step": 4001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186488451112253e-05,
+      "loss": 0.7296,
+      "step": 4002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186026804526722e-05,
+      "loss": 23.9054,
+      "step": 4003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18556503859853e-05,
+      "loss": 23.6933,
+      "step": 4004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185103153340841e-05,
+      "loss": 21.3206,
+      "step": 4005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184641148766825e-05,
+      "loss": 22.6049,
+      "step": 4006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184179024889652e-05,
+      "loss": 0.7276,
+      "step": 4007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.1837167817225e-05,
+      "loss": 21.4883,
+      "step": 4008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183254419278544e-05,
+      "loss": 21.3584,
+      "step": 4009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182791937570966e-05,
+      "loss": 22.9038,
+      "step": 4010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182329336612956e-05,
+      "loss": 21.7819,
+      "step": 4011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181866616417698e-05,
+      "loss": 24.7171,
+      "step": 4012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181403776998384e-05,
+      "loss": 20.4193,
+      "step": 4013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.180940818368211e-05,
+      "loss": 0.7818,
+      "step": 4014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.180477740540378e-05,
+      "loss": 19.8408,
+      "step": 4015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.180014543528088e-05,
+      "loss": 0.7053,
+      "step": 4016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.179551227344545e-05,
+      "loss": 24.516,
+      "step": 4017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.179087792002959e-05,
+      "loss": 0.8162,
+      "step": 4018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17862423751654e-05,
+      "loss": 23.0761,
+      "step": 4019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.178160563898506e-05,
+      "loss": 24.1793,
+      "step": 4020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177696771162077e-05,
+      "loss": 24.7422,
+      "step": 4021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177232859320474e-05,
+      "loss": 23.4693,
+      "step": 4022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176768828386923e-05,
+      "loss": 22.5082,
+      "step": 4023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176304678374656e-05,
+      "loss": 23.8859,
+      "step": 4024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.175840409296903e-05,
+      "loss": 0.8188,
+      "step": 4025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.175376021166902e-05,
+      "loss": 0.0932,
+      "step": 4026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17491151399789e-05,
+      "loss": 22.2774,
+      "step": 4027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174446887803114e-05,
+      "loss": 0.0702,
+      "step": 4028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.173982142595816e-05,
+      "loss": 21.9481,
+      "step": 4029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.173517278389247e-05,
+      "loss": 23.7318,
+      "step": 4030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.173052295196662e-05,
+      "loss": 0.0764,
+      "step": 4031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172587193031318e-05,
+      "loss": 0.08,
+      "step": 4032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172121971906473e-05,
+      "loss": 0.0829,
+      "step": 4033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171656631835391e-05,
+      "loss": 0.7253,
+      "step": 4034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171191172831337e-05,
+      "loss": 0.7673,
+      "step": 4035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.170725594907584e-05,
+      "loss": 20.693,
+      "step": 4036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.170259898077402e-05,
+      "loss": 0.7519,
+      "step": 4037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169794082354072e-05,
+      "loss": 0.7595,
+      "step": 4038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169328147750875e-05,
+      "loss": 0.7799,
+      "step": 4039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168862094281088e-05,
+      "loss": 24.5067,
+      "step": 4040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168395921958004e-05,
+      "loss": 21.9042,
+      "step": 4041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.167929630794911e-05,
+      "loss": 0.0976,
+      "step": 4042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.167463220805106e-05,
+      "loss": 21.1068,
+      "step": 4043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166996692001882e-05,
+      "loss": 23.0873,
+      "step": 4044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16653004439854e-05,
+      "loss": 21.4551,
+      "step": 4045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166063278008388e-05,
+      "loss": 0.7199,
+      "step": 4046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165596392844729e-05,
+      "loss": 22.7771,
+      "step": 4047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165129388920877e-05,
+      "loss": 0.0513,
+      "step": 4048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164662266250143e-05,
+      "loss": 19.9479,
+      "step": 4049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164195024845849e-05,
+      "loss": 22.3983,
+      "step": 4050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163727664721312e-05,
+      "loss": 0.6883,
+      "step": 4051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163260185889857e-05,
+      "loss": 20.9633,
+      "step": 4052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162792588364812e-05,
+      "loss": 0.0732,
+      "step": 4053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16232487215951e-05,
+      "loss": 23.1785,
+      "step": 4054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161857037287282e-05,
+      "loss": 22.4355,
+      "step": 4055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16138908376147e-05,
+      "loss": 0.8008,
+      "step": 4056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160921011595412e-05,
+      "loss": 20.7228,
+      "step": 4057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160452820802455e-05,
+      "loss": 21.5286,
+      "step": 4058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159984511395942e-05,
+      "loss": 20.7691,
+      "step": 4059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159516083389233e-05,
+      "loss": 0.7783,
+      "step": 4060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159047536795675e-05,
+      "loss": 23.3364,
+      "step": 4061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.158578871628629e-05,
+      "loss": 25.5892,
+      "step": 4062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.158110087901456e-05,
+      "loss": 22.4282,
+      "step": 4063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157641185627522e-05,
+      "loss": 22.5933,
+      "step": 4064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157172164820195e-05,
+      "loss": 0.8264,
+      "step": 4065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156703025492847e-05,
+      "loss": 22.9673,
+      "step": 4066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156233767658853e-05,
+      "loss": 0.6778,
+      "step": 4067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155764391331589e-05,
+      "loss": 0.7784,
+      "step": 4068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15529489652444e-05,
+      "loss": 24.9848,
+      "step": 4069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15482528325079e-05,
+      "loss": 22.3441,
+      "step": 4070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.154355551524028e-05,
+      "loss": 0.7955,
+      "step": 4071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.153885701357543e-05,
+      "loss": 23.0627,
+      "step": 4072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.153415732764735e-05,
+      "loss": 0.7785,
+      "step": 4073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152945645758998e-05,
+      "loss": 20.8818,
+      "step": 4074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15247544035374e-05,
+      "loss": 0.7749,
+      "step": 4075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152005116562361e-05,
+      "loss": 0.7905,
+      "step": 4076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151534674398272e-05,
+      "loss": 0.7024,
+      "step": 4077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151064113874886e-05,
+      "loss": 0.7649,
+      "step": 4078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150593435005616e-05,
+      "loss": 0.7648,
+      "step": 4079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150122637803882e-05,
+      "loss": 24.1422,
+      "step": 4080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149651722283109e-05,
+      "loss": 0.7507,
+      "step": 4081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149180688456719e-05,
+      "loss": 0.0667,
+      "step": 4082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148709536338143e-05,
+      "loss": 21.8168,
+      "step": 4083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148238265940812e-05,
+      "loss": 0.7022,
+      "step": 4084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.147766877278165e-05,
+      "loss": 21.4836,
+      "step": 4085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.147295370363638e-05,
+      "loss": 0.7308,
+      "step": 4086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.146823745210674e-05,
+      "loss": 22.0897,
+      "step": 4087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14635200183272e-05,
+      "loss": 23.4188,
+      "step": 4088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145880140243225e-05,
+      "loss": 0.7592,
+      "step": 4089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145408160455641e-05,
+      "loss": 23.3749,
+      "step": 4090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144936062483425e-05,
+      "loss": 22.9738,
+      "step": 4091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144463846340035e-05,
+      "loss": 0.6984,
+      "step": 4092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143991512038937e-05,
+      "loss": 0.7599,
+      "step": 4093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143519059593592e-05,
+      "loss": 22.4424,
+      "step": 4094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143046489017473e-05,
+      "loss": 23.9467,
+      "step": 4095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142573800324052e-05,
+      "loss": 21.965,
+      "step": 4096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142100993526808e-05,
+      "loss": 0.7102,
+      "step": 4097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.141628068639215e-05,
+      "loss": 22.5227,
+      "step": 4098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.141155025674759e-05,
+      "loss": 21.2932,
+      "step": 4099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140681864646928e-05,
+      "loss": 0.7079,
+      "step": 4100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140208585569209e-05,
+      "loss": 22.9801,
+      "step": 4101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139735188455099e-05,
+      "loss": 22.6144,
+      "step": 4102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139261673318089e-05,
+      "loss": 0.7701,
+      "step": 4103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13878804017168e-05,
+      "loss": 23.3615,
+      "step": 4104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13831428902938e-05,
+      "loss": 0.7419,
+      "step": 4105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13784041990469e-05,
+      "loss": 25.6887,
+      "step": 4106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.137366432811123e-05,
+      "loss": 24.9366,
+      "step": 4107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13689232776219e-05,
+      "loss": 0.7076,
+      "step": 4108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136418104771411e-05,
+      "loss": 0.747,
+      "step": 4109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135943763852301e-05,
+      "loss": 21.9523,
+      "step": 4110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13546930501839e-05,
+      "loss": 20.0985,
+      "step": 4111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134994728283196e-05,
+      "loss": 0.7122,
+      "step": 4112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134520033660259e-05,
+      "loss": 20.4041,
+      "step": 4113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134045221163106e-05,
+      "loss": 23.8157,
+      "step": 4114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133570290805273e-05,
+      "loss": 0.7214,
+      "step": 4115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133095242600305e-05,
+      "loss": 21.0714,
+      "step": 4116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132620076561741e-05,
+      "loss": 20.1645,
+      "step": 4117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132144792703132e-05,
+      "loss": 0.7478,
+      "step": 4118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131669391038025e-05,
+      "loss": 20.0842,
+      "step": 4119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131193871579975e-05,
+      "loss": 22.2062,
+      "step": 4120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130718234342538e-05,
+      "loss": 19.8654,
+      "step": 4121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130242479339275e-05,
+      "loss": 19.4912,
+      "step": 4122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.12976660658375e-05,
+      "loss": 0.8672,
+      "step": 4123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.12929061608953e-05,
+      "loss": 0.7934,
+      "step": 4124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128814507870183e-05,
+      "loss": 0.7931,
+      "step": 4125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128338281939286e-05,
+      "loss": 24.4501,
+      "step": 4126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.127861938310416e-05,
+      "loss": 0.7333,
+      "step": 4127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12738547699715e-05,
+      "loss": 22.2911,
+      "step": 4128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.126908898013076e-05,
+      "loss": 22.0059,
+      "step": 4129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.126432201371777e-05,
+      "loss": 0.0534,
+      "step": 4130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125955387086847e-05,
+      "loss": 22.3854,
+      "step": 4131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125478455171879e-05,
+      "loss": 0.077,
+      "step": 4132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12500140564047e-05,
+      "loss": 21.5024,
+      "step": 4133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124524238506219e-05,
+      "loss": 0.7921,
+      "step": 4134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124046953782732e-05,
+      "loss": 0.0658,
+      "step": 4135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123569551483614e-05,
+      "loss": 22.1266,
+      "step": 4136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12309203162248e-05,
+      "loss": 0.0811,
+      "step": 4137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12261439421294e-05,
+      "loss": 21.7082,
+      "step": 4138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122136639268613e-05,
+      "loss": 0.763,
+      "step": 4139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.121658766803118e-05,
+      "loss": 0.7316,
+      "step": 4140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.121180776830082e-05,
+      "loss": 21.0536,
+      "step": 4141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120702669363131e-05,
+      "loss": 0.0797,
+      "step": 4142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120224444415892e-05,
+      "loss": 0.0687,
+      "step": 4143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119746102002005e-05,
+      "loss": 0.8251,
+      "step": 4144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119267642135104e-05,
+      "loss": 0.0785,
+      "step": 4145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118789064828831e-05,
+      "loss": 0.7949,
+      "step": 4146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11831037009683e-05,
+      "loss": 0.7277,
+      "step": 4147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.117831557952748e-05,
+      "loss": 0.7746,
+      "step": 4148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.117352628410236e-05,
+      "loss": 22.66,
+      "step": 4149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116873581482948e-05,
+      "loss": 0.0844,
+      "step": 4150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116394417184541e-05,
+      "loss": 22.3727,
+      "step": 4151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.115915135528678e-05,
+      "loss": 0.6898,
+      "step": 4152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.115435736529022e-05,
+      "loss": 0.7059,
+      "step": 4153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11495622019924e-05,
+      "loss": 0.7418,
+      "step": 4154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114476586553002e-05,
+      "loss": 0.7545,
+      "step": 4155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.113996835603987e-05,
+      "loss": 22.7647,
+      "step": 4156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.113516967365866e-05,
+      "loss": 0.0935,
+      "step": 4157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.113036981852323e-05,
+      "loss": 0.0746,
+      "step": 4158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.112556879077043e-05,
+      "loss": 0.7847,
+      "step": 4159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.112076659053714e-05,
+      "loss": 0.0771,
+      "step": 4160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111596321796025e-05,
+      "loss": 0.7545,
+      "step": 4161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111115867317672e-05,
+      "loss": 22.9048,
+      "step": 4162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11063529563235e-05,
+      "loss": 0.7603,
+      "step": 4163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110154606753764e-05,
+      "loss": 0.7483,
+      "step": 4164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.109673800695615e-05,
+      "loss": 0.794,
+      "step": 4165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.109192877471612e-05,
+      "loss": 0.0686,
+      "step": 4166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108711837095467e-05,
+      "loss": 24.7616,
+      "step": 4167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108230679580891e-05,
+      "loss": 0.0696,
+      "step": 4168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.107749404941605e-05,
+      "loss": 0.7974,
+      "step": 4169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10726801319133e-05,
+      "loss": 20.8912,
+      "step": 4170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10678650434379e-05,
+      "loss": 19.3729,
+      "step": 4171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.106304878412711e-05,
+      "loss": 0.741,
+      "step": 4172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105823135411823e-05,
+      "loss": 0.0756,
+      "step": 4173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105341275354864e-05,
+      "loss": 0.0862,
+      "step": 4174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10485929825557e-05,
+      "loss": 0.7591,
+      "step": 4175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.104377204127683e-05,
+      "loss": 0.0664,
+      "step": 4176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.103894992984945e-05,
+      "loss": 0.7758,
+      "step": 4177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.103412664841106e-05,
+      "loss": 0.7924,
+      "step": 4178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102930219709917e-05,
+      "loss": 0.7626,
+      "step": 4179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102447657605132e-05,
+      "loss": 21.7769,
+      "step": 4180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101964978540508e-05,
+      "loss": 22.9396,
+      "step": 4181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101482182529806e-05,
+      "loss": 22.8515,
+      "step": 4182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100999269586791e-05,
+      "loss": 23.5243,
+      "step": 4183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100516239725233e-05,
+      "loss": 19.3405,
+      "step": 4184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100033092958899e-05,
+      "loss": 22.9396,
+      "step": 4185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099549829301566e-05,
+      "loss": 0.7382,
+      "step": 4186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099066448767011e-05,
+      "loss": 0.7529,
+      "step": 4187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.098582951369013e-05,
+      "loss": 21.2437,
+      "step": 4188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09809933712136e-05,
+      "loss": 0.783,
+      "step": 4189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09761560603784e-05,
+      "loss": 0.8044,
+      "step": 4190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.097131758132239e-05,
+      "loss": 23.1776,
+      "step": 4191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.096647793418356e-05,
+      "loss": 0.0605,
+      "step": 4192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.096163711909987e-05,
+      "loss": 21.2474,
+      "step": 4193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095679513620934e-05,
+      "loss": 0.7584,
+      "step": 4194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095195198565e-05,
+      "loss": 21.4693,
+      "step": 4195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094710766755993e-05,
+      "loss": 22.9184,
+      "step": 4196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094226218207725e-05,
+      "loss": 0.0701,
+      "step": 4197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09374155293401e-05,
+      "loss": 0.7344,
+      "step": 4198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.093256770948666e-05,
+      "loss": 21.8194,
+      "step": 4199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092771872265514e-05,
+      "loss": 0.7748,
+      "step": 4200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092286856898378e-05,
+      "loss": 22.864,
+      "step": 4201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091801724861084e-05,
+      "loss": 21.738,
+      "step": 4202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091316476167467e-05,
+      "loss": 21.97,
+      "step": 4203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.090831110831357e-05,
+      "loss": 21.0225,
+      "step": 4204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.090345628866594e-05,
+      "loss": 0.7708,
+      "step": 4205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089860030287018e-05,
+      "loss": 21.724,
+      "step": 4206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089374315106476e-05,
+      "loss": 0.8657,
+      "step": 4207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08888848333881e-05,
+      "loss": 0.07,
+      "step": 4208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.088402534997877e-05,
+      "loss": 23.163,
+      "step": 4209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087916470097529e-05,
+      "loss": 0.7242,
+      "step": 4210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087430288651624e-05,
+      "loss": 0.7479,
+      "step": 4211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08694399067402e-05,
+      "loss": 0.7859,
+      "step": 4212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086457576178585e-05,
+      "loss": 0.7393,
+      "step": 4213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085971045179184e-05,
+      "loss": 0.7299,
+      "step": 4214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08548439768969e-05,
+      "loss": 23.3352,
+      "step": 4215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.084997633723976e-05,
+      "loss": 0.6846,
+      "step": 4216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08451075329592e-05,
+      "loss": 22.3855,
+      "step": 4217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.084023756419402e-05,
+      "loss": 22.4125,
+      "step": 4218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.083536643108306e-05,
+      "loss": 0.7146,
+      "step": 4219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08304941337652e-05,
+      "loss": 22.4912,
+      "step": 4220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.082562067237936e-05,
+      "loss": 21.3088,
+      "step": 4221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.082074604706447e-05,
+      "loss": 0.0901,
+      "step": 4222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081587025795952e-05,
+      "loss": 22.5223,
+      "step": 4223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081099330520349e-05,
+      "loss": 0.0715,
+      "step": 4224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080611518893544e-05,
+      "loss": 22.2062,
+      "step": 4225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080123590929443e-05,
+      "loss": 0.0569,
+      "step": 4226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.079635546641958e-05,
+      "loss": 21.9597,
+      "step": 4227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.079147386045002e-05,
+      "loss": 23.1362,
+      "step": 4228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.078659109152494e-05,
+      "loss": 22.8511,
+      "step": 4229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.078170715978353e-05,
+      "loss": 0.7708,
+      "step": 4230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077682206536503e-05,
+      "loss": 22.2185,
+      "step": 4231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077193580840873e-05,
+      "loss": 22.4725,
+      "step": 4232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07670483890539e-05,
+      "loss": 0.7239,
+      "step": 4233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.076215980743994e-05,
+      "loss": 20.173,
+      "step": 4234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075727006370615e-05,
+      "loss": 21.0628,
+      "step": 4235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075237915799199e-05,
+      "loss": 0.7643,
+      "step": 4236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.074748709043689e-05,
+      "loss": 21.2311,
+      "step": 4237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.074259386118029e-05,
+      "loss": 0.767,
+      "step": 4238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.073769947036174e-05,
+      "loss": 0.7732,
+      "step": 4239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.073280391812074e-05,
+      "loss": 0.0734,
+      "step": 4240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072790720459687e-05,
+      "loss": 0.8263,
+      "step": 4241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072300932992976e-05,
+      "loss": 20.691,
+      "step": 4242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0718110294259e-05,
+      "loss": 0.0725,
+      "step": 4243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071321009772432e-05,
+      "loss": 21.4928,
+      "step": 4244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.070830874046536e-05,
+      "loss": 0.7682,
+      "step": 4245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.070340622262191e-05,
+      "loss": 0.7561,
+      "step": 4246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.069850254433371e-05,
+      "loss": 23.2806,
+      "step": 4247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.069359770574057e-05,
+      "loss": 0.735,
+      "step": 4248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068869170698234e-05,
+      "loss": 0.0911,
+      "step": 4249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068378454819885e-05,
+      "loss": 21.6795,
+      "step": 4250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.067887622953003e-05,
+      "loss": 0.7399,
+      "step": 4251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.067396675111581e-05,
+      "loss": 0.7965,
+      "step": 4252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066905611309618e-05,
+      "loss": 0.0614,
+      "step": 4253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06641443156111e-05,
+      "loss": 0.0698,
+      "step": 4254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065923135880063e-05,
+      "loss": 0.7563,
+      "step": 4255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065431724280484e-05,
+      "loss": 22.1644,
+      "step": 4256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064940196776382e-05,
+      "loss": 22.4222,
+      "step": 4257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064448553381771e-05,
+      "loss": 0.0479,
+      "step": 4258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063956794110666e-05,
+      "loss": 0.7034,
+      "step": 4259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06346491897709e-05,
+      "loss": 0.7431,
+      "step": 4260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062972927995064e-05,
+      "loss": 21.7716,
+      "step": 4261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062480821178614e-05,
+      "loss": 22.262,
+      "step": 4262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06198859854177e-05,
+      "loss": 0.7534,
+      "step": 4263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061496260098568e-05,
+      "loss": 20.5731,
+      "step": 4264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061003805863041e-05,
+      "loss": 0.7653,
+      "step": 4265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.060511235849232e-05,
+      "loss": 20.8107,
+      "step": 4266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06001855007118e-05,
+      "loss": 0.0649,
+      "step": 4267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.059525748542936e-05,
+      "loss": 0.7319,
+      "step": 4268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.059032831278546e-05,
+      "loss": 23.4442,
+      "step": 4269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058539798292064e-05,
+      "loss": 0.6979,
+      "step": 4270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058046649597546e-05,
+      "loss": 20.2587,
+      "step": 4271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057553385209054e-05,
+      "loss": 22.8088,
+      "step": 4272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057060005140647e-05,
+      "loss": 23.7805,
+      "step": 4273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056566509406394e-05,
+      "loss": 19.2618,
+      "step": 4274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056072898020365e-05,
+      "loss": 20.5422,
+      "step": 4275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055579170996629e-05,
+      "loss": 0.0796,
+      "step": 4276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055085328349264e-05,
+      "loss": 20.7189,
+      "step": 4277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054591370092351e-05,
+      "loss": 22.7805,
+      "step": 4278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054097296239972e-05,
+      "loss": 0.7182,
+      "step": 4279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05360310680621e-05,
+      "loss": 0.6917,
+      "step": 4280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053108801805158e-05,
+      "loss": 21.7603,
+      "step": 4281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.052614381250907e-05,
+      "loss": 23.0195,
+      "step": 4282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.052119845157552e-05,
+      "loss": 23.674,
+      "step": 4283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.051625193539193e-05,
+      "loss": 0.7557,
+      "step": 4284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.051130426409932e-05,
+      "loss": 22.6838,
+      "step": 4285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050635543783877e-05,
+      "loss": 20.5692,
+      "step": 4286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050140545675132e-05,
+      "loss": 0.6912,
+      "step": 4287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049645432097815e-05,
+      "loss": 23.4001,
+      "step": 4288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049150203066037e-05,
+      "loss": 22.7096,
+      "step": 4289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04865485859392e-05,
+      "loss": 22.5151,
+      "step": 4290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048159398695583e-05,
+      "loss": 21.54,
+      "step": 4291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047663823385154e-05,
+      "loss": 19.0797,
+      "step": 4292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047168132676762e-05,
+      "loss": 23.2412,
+      "step": 4293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.046672326584538e-05,
+      "loss": 21.1262,
+      "step": 4294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.046176405122616e-05,
+      "loss": 22.2624,
+      "step": 4295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045680368305137e-05,
+      "loss": 0.0708,
+      "step": 4296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045184216146243e-05,
+      "loss": 22.2996,
+      "step": 4297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044687948660076e-05,
+      "loss": 21.2409,
+      "step": 4298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044191565860787e-05,
+      "loss": 0.0777,
+      "step": 4299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.043695067762527e-05,
+      "loss": 0.7374,
+      "step": 4300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.043198454379453e-05,
+      "loss": 0.7489,
+      "step": 4301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042701725725721e-05,
+      "loss": 0.7459,
+      "step": 4302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042204881815493e-05,
+      "loss": 0.7176,
+      "step": 4303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.041707922662934e-05,
+      "loss": 0.8095,
+      "step": 4304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.041210848282213e-05,
+      "loss": 22.7444,
+      "step": 4305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0407136586875e-05,
+      "loss": 0.7748,
+      "step": 4306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040216353892972e-05,
+      "loss": 24.2327,
+      "step": 4307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039718933912804e-05,
+      "loss": 22.8887,
+      "step": 4308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.03922139876118e-05,
+      "loss": 23.3409,
+      "step": 4309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.038723748452284e-05,
+      "loss": 0.7576,
+      "step": 4310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.038225983000303e-05,
+      "loss": 0.8239,
+      "step": 4311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037728102419428e-05,
+      "loss": 22.1146,
+      "step": 4312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037230106723856e-05,
+      "loss": 22.2545,
+      "step": 4313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.03673199592778e-05,
+      "loss": 23.5782,
+      "step": 4314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.036233770045405e-05,
+      "loss": 0.7285,
+      "step": 4315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.035735429090937e-05,
+      "loss": 0.7798,
+      "step": 4316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.035236973078577e-05,
+      "loss": 21.0734,
+      "step": 4317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034738402022542e-05,
+      "loss": 21.9656,
+      "step": 4318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034239715937044e-05,
+      "loss": 21.086,
+      "step": 4319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033740914836299e-05,
+      "loss": 23.3398,
+      "step": 4320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033241998734529e-05,
+      "loss": 0.0804,
+      "step": 4321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.032742967645959e-05,
+      "loss": 0.6924,
+      "step": 4322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.032243821584816e-05,
+      "loss": 22.8707,
+      "step": 4323
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031744560565327e-05,
+      "loss": 0.7503,
+      "step": 4324
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03124518460173e-05,
+      "loss": 23.3205,
+      "step": 4325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.030745693708261e-05,
+      "loss": 21.3461,
+      "step": 4326
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03024608789916e-05,
+      "loss": 0.7872,
+      "step": 4327
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029746367188671e-05,
+      "loss": 0.8001,
+      "step": 4328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029246531591039e-05,
+      "loss": 0.7973,
+      "step": 4329
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028746581120517e-05,
+      "loss": 0.7383,
+      "step": 4330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028246515791357e-05,
+      "loss": 0.0776,
+      "step": 4331
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027746335617816e-05,
+      "loss": 0.7302,
+      "step": 4332
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027246040614155e-05,
+      "loss": 25.8679,
+      "step": 4333
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.026745630794634e-05,
+      "loss": 26.4455,
+      "step": 4334
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.026245106173523e-05,
+      "loss": 22.2929,
+      "step": 4335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025744466765091e-05,
+      "loss": 0.082,
+      "step": 4336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025243712583611e-05,
+      "loss": 24.7355,
+      "step": 4337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024742843643359e-05,
+      "loss": 22.1182,
+      "step": 4338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024241859958615e-05,
+      "loss": 0.7003,
+      "step": 4339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.023740761543663e-05,
+      "loss": 0.7437,
+      "step": 4340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.023239548412785e-05,
+      "loss": 0.7733,
+      "step": 4341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.022738220580276e-05,
+      "loss": 0.7309,
+      "step": 4342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.022236778060426e-05,
+      "loss": 0.7597,
+      "step": 4343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02173522086753e-05,
+      "loss": 0.7733,
+      "step": 4344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021233549015891e-05,
+      "loss": 25.342,
+      "step": 4345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020731762519807e-05,
+      "loss": 23.8856,
+      "step": 4346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020229861393585e-05,
+      "loss": 0.8008,
+      "step": 4347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.019727845651538e-05,
+      "loss": 0.1062,
+      "step": 4348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.019225715307973e-05,
+      "loss": 21.6649,
+      "step": 4349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.018723470377208e-05,
+      "loss": 23.2761,
+      "step": 4350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.018221110873563e-05,
+      "loss": 21.5701,
+      "step": 4351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017718636811358e-05,
+      "loss": 0.7864,
+      "step": 4352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01721604820492e-05,
+      "loss": 0.0891,
+      "step": 4353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016713345068577e-05,
+      "loss": 21.6539,
+      "step": 4354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016210527416662e-05,
+      "loss": 20.4764,
+      "step": 4355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015707595263508e-05,
+      "loss": 0.7451,
+      "step": 4356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015204548623454e-05,
+      "loss": 21.1299,
+      "step": 4357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.014701387510845e-05,
+      "loss": 21.7666,
+      "step": 4358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.014198111940022e-05,
+      "loss": 0.7731,
+      "step": 4359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013694721925334e-05,
+      "loss": 21.2978,
+      "step": 4360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013191217481133e-05,
+      "loss": 0.0954,
+      "step": 4361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.012687598621776e-05,
+      "loss": 0.7804,
+      "step": 4362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.012183865361618e-05,
+      "loss": 0.7224,
+      "step": 4363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011680017715022e-05,
+      "loss": 0.735,
+      "step": 4364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011176055696354e-05,
+      "loss": 0.7899,
+      "step": 4365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.010671979319977e-05,
+      "loss": 24.1073,
+      "step": 4366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.010167788600266e-05,
+      "loss": 21.6173,
+      "step": 4367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.009663483551594e-05,
+      "loss": 22.133,
+      "step": 4368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00915906418834e-05,
+      "loss": 23.1881,
+      "step": 4369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008654530524882e-05,
+      "loss": 0.7764,
+      "step": 4370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008149882575607e-05,
+      "loss": 0.7915,
+      "step": 4371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007645120354902e-05,
+      "loss": 20.0692,
+      "step": 4372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007140243877157e-05,
+      "loss": 0.0742,
+      "step": 4373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.006635253156766e-05,
+      "loss": 0.0864,
+      "step": 4374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.006130148208125e-05,
+      "loss": 20.2451,
+      "step": 4375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005624929045635e-05,
+      "loss": 23.3879,
+      "step": 4376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005119595683702e-05,
+      "loss": 21.3294,
+      "step": 4377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004614148136731e-05,
+      "loss": 0.8202,
+      "step": 4378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004108586419131e-05,
+      "loss": 23.1852,
+      "step": 4379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003602910545319e-05,
+      "loss": 0.7445,
+      "step": 4380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003097120529708e-05,
+      "loss": 23.6859,
+      "step": 4381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00259121638672e-05,
+      "loss": 21.2526,
+      "step": 4382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002085198130778e-05,
+      "loss": 0.7791,
+      "step": 4383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.001579065776307e-05,
+      "loss": 22.9228,
+      "step": 4384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00107281933774e-05,
+      "loss": 0.0837,
+      "step": 4385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000566458829504e-05,
+      "loss": 22.4482,
+      "step": 4386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000059984266043e-05,
+      "loss": 22.7163,
+      "step": 4387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99955339566179e-05,
+      "loss": 0.7478,
+      "step": 4388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999046693031192e-05,
+      "loss": 0.7424,
+      "step": 4389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998539876388691e-05,
+      "loss": 20.1044,
+      "step": 4390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998032945748742e-05,
+      "loss": 22.0397,
+      "step": 4391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.997525901125792e-05,
+      "loss": 23.8294,
+      "step": 4392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.997018742534298e-05,
+      "loss": 20.5781,
+      "step": 4393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99651146998872e-05,
+      "loss": 0.7108,
+      "step": 4394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99600408350352e-05,
+      "loss": 0.7622,
+      "step": 4395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.995496583093164e-05,
+      "loss": 20.3477,
+      "step": 4396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.994988968772119e-05,
+      "loss": 0.6935,
+      "step": 4397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.994481240554859e-05,
+      "loss": 20.935,
+      "step": 4398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993973398455858e-05,
+      "loss": 0.7676,
+      "step": 4399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993465442489593e-05,
+      "loss": 0.6972,
+      "step": 4400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.992957372670549e-05,
+      "loss": 0.7818,
+      "step": 4401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99244918901321e-05,
+      "loss": 0.7699,
+      "step": 4402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.991940891532062e-05,
+      "loss": 0.7233,
+      "step": 4403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.9914324802416e-05,
+      "loss": 0.7143,
+      "step": 4404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990923955156314e-05,
+      "loss": 23.2245,
+      "step": 4405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990415316290707e-05,
+      "loss": 0.8096,
+      "step": 4406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989906563659276e-05,
+      "loss": 21.5639,
+      "step": 4407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989397697276529e-05,
+      "loss": 24.5157,
+      "step": 4408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988888717156972e-05,
+      "loss": 0.7526,
+      "step": 4409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988379623315114e-05,
+      "loss": 21.7068,
+      "step": 4410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987870415765472e-05,
+      "loss": 0.7703,
+      "step": 4411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987361094522561e-05,
+      "loss": 22.3241,
+      "step": 4412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.986851659600904e-05,
+      "loss": 21.1199,
+      "step": 4413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.986342111015025e-05,
+      "loss": 0.7582,
+      "step": 4414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.985832448779448e-05,
+      "loss": 0.7312,
+      "step": 4415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.985322672908707e-05,
+      "loss": 0.7299,
+      "step": 4416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984812783417333e-05,
+      "loss": 0.7065,
+      "step": 4417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984302780319863e-05,
+      "loss": 19.3668,
+      "step": 4418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983792663630839e-05,
+      "loss": 21.7777,
+      "step": 4419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983282433364803e-05,
+      "loss": 21.6437,
+      "step": 4420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.9827720895363e-05,
+      "loss": 21.9552,
+      "step": 4421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.982261632159883e-05,
+      "loss": 0.795,
+      "step": 4422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.981751061250104e-05,
+      "loss": 21.1961,
+      "step": 4423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.981240376821518e-05,
+      "loss": 0.7216,
+      "step": 4424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980729578888685e-05,
+      "loss": 20.7637,
+      "step": 4425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980218667466167e-05,
+      "loss": 21.8693,
+      "step": 4426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979707642568533e-05,
+      "loss": 0.7495,
+      "step": 4427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979196504210349e-05,
+      "loss": 22.5146,
+      "step": 4428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97868525240619e-05,
+      "loss": 21.5669,
+      "step": 4429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.978173887170628e-05,
+      "loss": 0.7876,
+      "step": 4430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977662408518245e-05,
+      "loss": 0.6574,
+      "step": 4431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977150816463622e-05,
+      "loss": 19.3835,
+      "step": 4432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.976639111021345e-05,
+      "loss": 22.6844,
+      "step": 4433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.976127292206003e-05,
+      "loss": 24.2296,
+      "step": 4434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975615360032186e-05,
+      "loss": 0.7505,
+      "step": 4435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97510331451449e-05,
+      "loss": 23.3558,
+      "step": 4436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.974591155667513e-05,
+      "loss": 23.4315,
+      "step": 4437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.974078883505859e-05,
+      "loss": 0.6999,
+      "step": 4438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97356649804413e-05,
+      "loss": 24.3028,
+      "step": 4439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.973053999296934e-05,
+      "loss": 0.0748,
+      "step": 4440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.972541387278884e-05,
+      "loss": 23.4301,
+      "step": 4441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.972028662004591e-05,
+      "loss": 23.4693,
+      "step": 4442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971515823488676e-05,
+      "loss": 0.0723,
+      "step": 4443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97100287174576e-05,
+      "loss": 21.874,
+      "step": 4444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.970489806790466e-05,
+      "loss": 20.929,
+      "step": 4445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969976628637422e-05,
+      "loss": 21.082,
+      "step": 4446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969463337301256e-05,
+      "loss": 0.6897,
+      "step": 4447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968949932796606e-05,
+      "loss": 0.713,
+      "step": 4448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968436415138106e-05,
+      "loss": 0.6795,
+      "step": 4449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.9679227843404e-05,
+      "loss": 22.0114,
+      "step": 4450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967409040418125e-05,
+      "loss": 0.7415,
+      "step": 4451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966895183385933e-05,
+      "loss": 23.3523,
+      "step": 4452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966381213258472e-05,
+      "loss": 23.8904,
+      "step": 4453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965867130050396e-05,
+      "loss": 21.3155,
+      "step": 4454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96535293377636e-05,
+      "loss": 21.8083,
+      "step": 4455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.964838624451025e-05,
+      "loss": 23.7,
+      "step": 4456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.964324202089053e-05,
+      "loss": 0.7851,
+      "step": 4457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.963809666705112e-05,
+      "loss": 0.7087,
+      "step": 4458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.963295018313868e-05,
+      "loss": 0.7618,
+      "step": 4459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962780256929996e-05,
+      "loss": 21.9639,
+      "step": 4460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962265382568171e-05,
+      "loss": 22.8028,
+      "step": 4461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961750395243071e-05,
+      "loss": 23.322,
+      "step": 4462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961235294969381e-05,
+      "loss": 0.0807,
+      "step": 4463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960720081761784e-05,
+      "loss": 0.0817,
+      "step": 4464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960204755634967e-05,
+      "loss": 22.794,
+      "step": 4465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.959689316603625e-05,
+      "loss": 21.4496,
+      "step": 4466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.959173764682454e-05,
+      "loss": 0.0695,
+      "step": 4467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958658099886148e-05,
+      "loss": 21.8421,
+      "step": 4468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958142322229412e-05,
+      "loss": 0.7588,
+      "step": 4469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957626431726949e-05,
+      "loss": 0.7664,
+      "step": 4470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957110428393466e-05,
+      "loss": 24.5975,
+      "step": 4471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956594312243678e-05,
+      "loss": 21.2192,
+      "step": 4472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956078083292297e-05,
+      "loss": 22.3329,
+      "step": 4473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955561741554037e-05,
+      "loss": 0.7258,
+      "step": 4474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955045287043624e-05,
+      "loss": 22.1979,
+      "step": 4475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.954528719775781e-05,
+      "loss": 21.6855,
+      "step": 4476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.954012039765234e-05,
+      "loss": 0.0925,
+      "step": 4477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.953495247026714e-05,
+      "loss": 21.9013,
+      "step": 4478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952978341574954e-05,
+      "loss": 0.7962,
+      "step": 4479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952461323424693e-05,
+      "loss": 22.2349,
+      "step": 4480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.951944192590667e-05,
+      "loss": 22.7453,
+      "step": 4481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.951426949087623e-05,
+      "loss": 21.5861,
+      "step": 4482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.950909592930305e-05,
+      "loss": 0.0626,
+      "step": 4483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.950392124133464e-05,
+      "loss": 20.0537,
+      "step": 4484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949874542711854e-05,
+      "loss": 0.7779,
+      "step": 4485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949356848680229e-05,
+      "loss": 20.4819,
+      "step": 4486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94883904205335e-05,
+      "loss": 20.8663,
+      "step": 4487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.948321122845977e-05,
+      "loss": 0.0659,
+      "step": 4488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94780309107288e-05,
+      "loss": 0.7463,
+      "step": 4489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947284946748824e-05,
+      "loss": 0.7404,
+      "step": 4490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946766689888584e-05,
+      "loss": 22.8268,
+      "step": 4491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946248320506933e-05,
+      "loss": 22.9344,
+      "step": 4492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.945729838618652e-05,
+      "loss": 0.8066,
+      "step": 4493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.945211244238519e-05,
+      "loss": 23.3397,
+      "step": 4494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944692537381325e-05,
+      "loss": 0.767,
+      "step": 4495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944173718061852e-05,
+      "loss": 0.7621,
+      "step": 4496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.943654786294895e-05,
+      "loss": 0.8013,
+      "step": 4497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.943135742095247e-05,
+      "loss": 0.7876,
+      "step": 4498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942616585477708e-05,
+      "loss": 22.245,
+      "step": 4499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942097316457076e-05,
+      "loss": 21.6906,
+      "step": 4500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941577935048158e-05,
+      "loss": 22.5679,
+      "step": 4501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941058441265759e-05,
+      "loss": 0.7732,
+      "step": 4502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940538835124693e-05,
+      "loss": 22.819,
+      "step": 4503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940019116639771e-05,
+      "loss": 0.0777,
+      "step": 4504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.93949928582581e-05,
+      "loss": 22.5271,
+      "step": 4505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.938979342697631e-05,
+      "loss": 22.1631,
+      "step": 4506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.93845928727006e-05,
+      "loss": 23.9458,
+      "step": 4507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.937939119557918e-05,
+      "loss": 18.8074,
+      "step": 4508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.937418839576039e-05,
+      "loss": 0.7754,
+      "step": 4509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.936898447339257e-05,
+      "loss": 0.071,
+      "step": 4510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.936377942862404e-05,
+      "loss": 20.5547,
+      "step": 4511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.935857326160323e-05,
+      "loss": 0.7717,
+      "step": 4512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.935336597247857e-05,
+      "loss": 23.0757,
+      "step": 4513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934815756139848e-05,
+      "loss": 22.4476,
+      "step": 4514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934294802851149e-05,
+      "loss": 0.0751,
+      "step": 4515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93377373739661e-05,
+      "loss": 21.0409,
+      "step": 4516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.933252559791087e-05,
+      "loss": 22.8967,
+      "step": 4517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93273127004944e-05,
+      "loss": 20.5965,
+      "step": 4518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932209868186527e-05,
+      "loss": 0.7556,
+      "step": 4519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.931688354217219e-05,
+      "loss": 20.856,
+      "step": 4520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93116672815638e-05,
+      "loss": 21.1793,
+      "step": 4521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930644990018881e-05,
+      "loss": 21.5758,
+      "step": 4522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930123139819599e-05,
+      "loss": 0.8102,
+      "step": 4523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.929601177573412e-05,
+      "loss": 22.0066,
+      "step": 4524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.929079103295198e-05,
+      "loss": 0.7073,
+      "step": 4525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.928556916999844e-05,
+      "loss": 0.7154,
+      "step": 4526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.928034618702237e-05,
+      "loss": 20.12,
+      "step": 4527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927512208417266e-05,
+      "loss": 22.2599,
+      "step": 4528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926989686159827e-05,
+      "loss": 20.9473,
+      "step": 4529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926467051944816e-05,
+      "loss": 0.7472,
+      "step": 4530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.925944305787132e-05,
+      "loss": 21.7437,
+      "step": 4531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92542144770168e-05,
+      "loss": 22.3511,
+      "step": 4532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.924898477703367e-05,
+      "loss": 22.0377,
+      "step": 4533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.9243753958071e-05,
+      "loss": 0.0535,
+      "step": 4534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923852202027793e-05,
+      "loss": 0.7529,
+      "step": 4535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923328896380365e-05,
+      "loss": 22.7036,
+      "step": 4536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.922805478879734e-05,
+      "loss": 0.0591,
+      "step": 4537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92228194954082e-05,
+      "loss": 22.028,
+      "step": 4538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92175830837855e-05,
+      "loss": 0.6954,
+      "step": 4539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.921234555407854e-05,
+      "loss": 0.7876,
+      "step": 4540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920710690643664e-05,
+      "loss": 24.386,
+      "step": 4541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920186714100915e-05,
+      "loss": 0.0672,
+      "step": 4542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.919662625794544e-05,
+      "loss": 0.7807,
+      "step": 4543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.919138425739493e-05,
+      "loss": 0.7523,
+      "step": 4544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91861411395071e-05,
+      "loss": 20.4728,
+      "step": 4545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.918089690443138e-05,
+      "loss": 0.6987,
+      "step": 4546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91756515523173e-05,
+      "loss": 0.7135,
+      "step": 4547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917040508331444e-05,
+      "loss": 0.0742,
+      "step": 4548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.916515749757233e-05,
+      "loss": 0.7547,
+      "step": 4549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91599087952406e-05,
+      "loss": 0.0633,
+      "step": 4550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.915465897646889e-05,
+      "loss": 23.1668,
+      "step": 4551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.914940804140685e-05,
+      "loss": 21.6648,
+      "step": 4552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.914415599020422e-05,
+      "loss": 21.3158,
+      "step": 4553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91389028230107e-05,
+      "loss": 22.0844,
+      "step": 4554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.913364853997605e-05,
+      "loss": 22.0966,
+      "step": 4555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912839314125012e-05,
+      "loss": 0.7715,
+      "step": 4556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912313662698268e-05,
+      "loss": 22.2099,
+      "step": 4557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911787899732364e-05,
+      "loss": 19.6245,
+      "step": 4558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911262025242286e-05,
+      "loss": 21.125,
+      "step": 4559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.910736039243028e-05,
+      "loss": 23.9364,
+      "step": 4560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.910209941749587e-05,
+      "loss": 22.1202,
+      "step": 4561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90968373277696e-05,
+      "loss": 0.7517,
+      "step": 4562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90915741234015e-05,
+      "loss": 0.7201,
+      "step": 4563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90863098045416e-05,
+      "loss": 20.4919,
+      "step": 4564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.908104437134003e-05,
+      "loss": 23.0411,
+      "step": 4565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907577782394688e-05,
+      "loss": 0.8175,
+      "step": 4566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907051016251226e-05,
+      "loss": 0.7855,
+      "step": 4567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.906524138718643e-05,
+      "loss": 0.7331,
+      "step": 4568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.905997149811953e-05,
+      "loss": 20.9317,
+      "step": 4569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.905470049546185e-05,
+      "loss": 23.2739,
+      "step": 4570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904942837936366e-05,
+      "loss": 0.0652,
+      "step": 4571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904415514997523e-05,
+      "loss": 19.4917,
+      "step": 4572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903888080744693e-05,
+      "loss": 0.7256,
+      "step": 4573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903360535192913e-05,
+      "loss": 0.7714,
+      "step": 4574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902832878357223e-05,
+      "loss": 23.6304,
+      "step": 4575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902305110252665e-05,
+      "loss": 25.1626,
+      "step": 4576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901777230894286e-05,
+      "loss": 20.7332,
+      "step": 4577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901249240297136e-05,
+      "loss": 21.8846,
+      "step": 4578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90072113847627e-05,
+      "loss": 20.3888,
+      "step": 4579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.900192925446741e-05,
+      "loss": 0.7706,
+      "step": 4580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89966460122361e-05,
+      "loss": 0.7522,
+      "step": 4581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.899136165821938e-05,
+      "loss": 0.0618,
+      "step": 4582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.898607619256791e-05,
+      "loss": 0.7506,
+      "step": 4583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89807896154324e-05,
+      "loss": 21.3905,
+      "step": 4584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.897550192696353e-05,
+      "loss": 0.0633,
+      "step": 4585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.897021312731209e-05,
+      "loss": 0.7298,
+      "step": 4586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.896492321662883e-05,
+      "loss": 0.0886,
+      "step": 4587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895963219506458e-05,
+      "loss": 22.4621,
+      "step": 4588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89543400627702e-05,
+      "loss": 22.9498,
+      "step": 4589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.894904681989652e-05,
+      "loss": 20.2133,
+      "step": 4590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89437524665945e-05,
+      "loss": 21.6051,
+      "step": 4591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.893845700301506e-05,
+      "loss": 21.0666,
+      "step": 4592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.893316042930916e-05,
+      "loss": 0.7753,
+      "step": 4593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.892786274562784e-05,
+      "loss": 0.7533,
+      "step": 4594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.892256395212209e-05,
+      "loss": 23.4631,
+      "step": 4595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8917264048943e-05,
+      "loss": 0.0867,
+      "step": 4596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89119630362417e-05,
+      "loss": 22.4578,
+      "step": 4597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.890666091416926e-05,
+      "loss": 21.5464,
+      "step": 4598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.890135768287688e-05,
+      "loss": 23.0344,
+      "step": 4599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889605334251575e-05,
+      "loss": 21.9635,
+      "step": 4600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88907478932371e-05,
+      "loss": 0.0828,
+      "step": 4601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.888544133519216e-05,
+      "loss": 0.7114,
+      "step": 4602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.888013366853225e-05,
+      "loss": 21.0154,
+      "step": 4603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.887482489340868e-05,
+      "loss": 0.7626,
+      "step": 4604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886951500997278e-05,
+      "loss": 21.218,
+      "step": 4605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886420401837599e-05,
+      "loss": 22.5709,
+      "step": 4606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.885889191876966e-05,
+      "loss": 21.1289,
+      "step": 4607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.885357871130527e-05,
+      "loss": 0.0699,
+      "step": 4608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88482643961343e-05,
+      "loss": 19.7826,
+      "step": 4609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.884294897340825e-05,
+      "loss": 0.7612,
+      "step": 4610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883763244327867e-05,
+      "loss": 19.8482,
+      "step": 4611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883231480589714e-05,
+      "loss": 0.0969,
+      "step": 4612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.882699606141523e-05,
+      "loss": 22.7756,
+      "step": 4613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.882167620998461e-05,
+      "loss": 21.9518,
+      "step": 4614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881635525175694e-05,
+      "loss": 0.7649,
+      "step": 4615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881103318688388e-05,
+      "loss": 20.4854,
+      "step": 4616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.880571001551725e-05,
+      "loss": 21.0507,
+      "step": 4617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.880038573780874e-05,
+      "loss": 21.802,
+      "step": 4618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879506035391015e-05,
+      "loss": 19.8114,
+      "step": 4619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.878973386397333e-05,
+      "loss": 21.4957,
+      "step": 4620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.878440626815013e-05,
+      "loss": 0.7643,
+      "step": 4621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87790775665924e-05,
+      "loss": 19.4557,
+      "step": 4622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877374775945213e-05,
+      "loss": 0.7248,
+      "step": 4623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.876841684688125e-05,
+      "loss": 0.0644,
+      "step": 4624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.876308482903169e-05,
+      "loss": 22.4344,
+      "step": 4625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875775170605552e-05,
+      "loss": 21.4035,
+      "step": 4626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875241747810477e-05,
+      "loss": 20.6757,
+      "step": 4627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.874708214533152e-05,
+      "loss": 22.803,
+      "step": 4628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.874174570788786e-05,
+      "loss": 21.7548,
+      "step": 4629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873640816592598e-05,
+      "loss": 0.7504,
+      "step": 4630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873106951959799e-05,
+      "loss": 20.8812,
+      "step": 4631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.872572976905614e-05,
+      "loss": 23.362,
+      "step": 4632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.872038891445265e-05,
+      "loss": 20.633,
+      "step": 4633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.871504695593977e-05,
+      "loss": 0.8263,
+      "step": 4634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870970389366981e-05,
+      "loss": 21.9166,
+      "step": 4635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870435972779511e-05,
+      "loss": 22.4996,
+      "step": 4636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869901445846802e-05,
+      "loss": 22.0837,
+      "step": 4637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869366808584095e-05,
+      "loss": 20.2904,
+      "step": 4638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868832061006629e-05,
+      "loss": 0.729,
+      "step": 4639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868297203129652e-05,
+      "loss": 0.7717,
+      "step": 4640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.867762234968412e-05,
+      "loss": 0.7291,
+      "step": 4641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86722715653816e-05,
+      "loss": 23.0014,
+      "step": 4642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.866691967854155e-05,
+      "loss": 22.6831,
+      "step": 4643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86615666893165e-05,
+      "loss": 20.3211,
+      "step": 4644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865621259785908e-05,
+      "loss": 21.7147,
+      "step": 4645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865085740432193e-05,
+      "loss": 24.2289,
+      "step": 4646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864550110885774e-05,
+      "loss": 0.7478,
+      "step": 4647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86401437116192e-05,
+      "loss": 22.5322,
+      "step": 4648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.863478521275908e-05,
+      "loss": 0.7759,
+      "step": 4649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86294256124301e-05,
+      "loss": 22.4302,
+      "step": 4650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86240649107851e-05,
+      "loss": 0.7636,
+      "step": 4651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86187031079769e-05,
+      "loss": 23.267,
+      "step": 4652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.861334020415836e-05,
+      "loss": 0.7368,
+      "step": 4653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.860797619948238e-05,
+      "loss": 21.0353,
+      "step": 4654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.860261109410188e-05,
+      "loss": 23.0487,
+      "step": 4655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.859724488816983e-05,
+      "loss": 0.7377,
+      "step": 4656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85918775818392e-05,
+      "loss": 21.6513,
+      "step": 4657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858650917526304e-05,
+      "loss": 22.3601,
+      "step": 4658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858113966859438e-05,
+      "loss": 20.2627,
+      "step": 4659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85757690619863e-05,
+      "loss": 21.7567,
+      "step": 4660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.857039735559193e-05,
+      "loss": 21.3769,
+      "step": 4661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.856502454956441e-05,
+      "loss": 0.6756,
+      "step": 4662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855965064405692e-05,
+      "loss": 20.9791,
+      "step": 4663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855427563922265e-05,
+      "loss": 0.7368,
+      "step": 4664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.854889953521486e-05,
+      "loss": 22.3884,
+      "step": 4665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.854352233218682e-05,
+      "loss": 0.7894,
+      "step": 4666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.853814403029184e-05,
+      "loss": 21.2905,
+      "step": 4667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.853276462968323e-05,
+      "loss": 21.646,
+      "step": 4668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852738413051437e-05,
+      "loss": 0.775,
+      "step": 4669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852200253293867e-05,
+      "loss": 21.5019,
+      "step": 4670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851661983710953e-05,
+      "loss": 22.0949,
+      "step": 4671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851123604318044e-05,
+      "loss": 22.1891,
+      "step": 4672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.850585115130485e-05,
+      "loss": 22.9159,
+      "step": 4673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.850046516163632e-05,
+      "loss": 0.7794,
+      "step": 4674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.849507807432839e-05,
+      "loss": 19.7572,
+      "step": 4675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848968988953465e-05,
+      "loss": 0.7475,
+      "step": 4676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84843006074087e-05,
+      "loss": 0.0828,
+      "step": 4677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.847891022810421e-05,
+      "loss": 21.3478,
+      "step": 4678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.847351875177483e-05,
+      "loss": 0.8159,
+      "step": 4679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84681261785743e-05,
+      "loss": 0.7969,
+      "step": 4680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846273250865633e-05,
+      "loss": 0.7262,
+      "step": 4681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.845733774217473e-05,
+      "loss": 21.6509,
+      "step": 4682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.845194187928328e-05,
+      "loss": 0.7412,
+      "step": 4683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84465449201358e-05,
+      "loss": 23.5729,
+      "step": 4684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.844114686488618e-05,
+      "loss": 22.5482,
+      "step": 4685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.843574771368832e-05,
+      "loss": 0.761,
+      "step": 4686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.843034746669613e-05,
+      "loss": 19.891,
+      "step": 4687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.842494612406358e-05,
+      "loss": 21.4563,
+      "step": 4688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841954368594468e-05,
+      "loss": 0.7435,
+      "step": 4689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841414015249341e-05,
+      "loss": 21.4353,
+      "step": 4690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.840873552386385e-05,
+      "loss": 19.6736,
+      "step": 4691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.840332980021008e-05,
+      "loss": 22.3819,
+      "step": 4692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.839792298168621e-05,
+      "loss": 21.9436,
+      "step": 4693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.839251506844641e-05,
+      "loss": 0.0717,
+      "step": 4694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.838710606064484e-05,
+      "loss": 0.7057,
+      "step": 4695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.83816959584357e-05,
+      "loss": 0.7069,
+      "step": 4696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837628476197326e-05,
+      "loss": 22.3181,
+      "step": 4697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837087247141176e-05,
+      "loss": 0.7241,
+      "step": 4698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.836545908690552e-05,
+      "loss": 0.7679,
+      "step": 4699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.836004460860887e-05,
+      "loss": 21.0755,
+      "step": 4700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.835462903667618e-05,
+      "loss": 23.4701,
+      "step": 4701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.834921237126186e-05,
+      "loss": 20.5249,
+      "step": 4702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.834379461252029e-05,
+      "loss": 21.6049,
+      "step": 4703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.833837576060599e-05,
+      "loss": 22.0191,
+      "step": 4704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83329558156734e-05,
+      "loss": 0.8101,
+      "step": 4705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832753477787708e-05,
+      "loss": 20.6251,
+      "step": 4706
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832211264737156e-05,
+      "loss": 22.0675,
+      "step": 4707
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.831668942431142e-05,
+      "loss": 0.725,
+      "step": 4708
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.831126510885131e-05,
+      "loss": 20.4378,
+      "step": 4709
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83058397011458e-05,
+      "loss": 0.7159,
+      "step": 4710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.830041320134966e-05,
+      "loss": 22.2213,
+      "step": 4711
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.829498560961755e-05,
+      "loss": 21.8305,
+      "step": 4712
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.828955692610421e-05,
+      "loss": 21.2813,
+      "step": 4713
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.828412715096441e-05,
+      "loss": 20.5909,
+      "step": 4714
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827869628435295e-05,
+      "loss": 0.7187,
+      "step": 4715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827326432642468e-05,
+      "loss": 20.0774,
+      "step": 4716
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.826783127733446e-05,
+      "loss": 0.7472,
+      "step": 4717
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.826239713723716e-05,
+      "loss": 0.7897,
+      "step": 4718
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825696190628774e-05,
+      "loss": 21.7628,
+      "step": 4719
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825152558464114e-05,
+      "loss": 20.5608,
+      "step": 4720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.824608817245234e-05,
+      "loss": 23.4346,
+      "step": 4721
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.824064966987639e-05,
+      "loss": 0.7736,
+      "step": 4722
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82352100770683e-05,
+      "loss": 21.4312,
+      "step": 4723
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822976939418317e-05,
+      "loss": 22.3873,
+      "step": 4724
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822432762137613e-05,
+      "loss": 23.1704,
+      "step": 4725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.821888475880229e-05,
+      "loss": 21.8195,
+      "step": 4726
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.821344080661685e-05,
+      "loss": 21.6902,
+      "step": 4727
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820799576497501e-05,
+      "loss": 21.1259,
+      "step": 4728
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.8202549634032e-05,
+      "loss": 21.5344,
+      "step": 4729
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81971024139431e-05,
+      "loss": 0.0627,
+      "step": 4730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.819165410486361e-05,
+      "loss": 21.6399,
+      "step": 4731
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818620470694883e-05,
+      "loss": 22.706,
+      "step": 4732
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818075422035415e-05,
+      "loss": 22.3021,
+      "step": 4733
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817530264523497e-05,
+      "loss": 0.6599,
+      "step": 4734
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.816984998174669e-05,
+      "loss": 0.7591,
+      "step": 4735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.816439623004476e-05,
+      "loss": 24.6026,
+      "step": 4736
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815894139028469e-05,
+      "loss": 0.7527,
+      "step": 4737
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815348546262199e-05,
+      "loss": 24.082,
+      "step": 4738
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81480284472122e-05,
+      "loss": 21.7696,
+      "step": 4739
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.814257034421088e-05,
+      "loss": 21.6874,
+      "step": 4740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.813711115377367e-05,
+      "loss": 20.9636,
+      "step": 4741
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81316508760562e-05,
+      "loss": 0.0641,
+      "step": 4742
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.812618951121414e-05,
+      "loss": 0.7663,
+      "step": 4743
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.812072705940319e-05,
+      "loss": 20.3621,
+      "step": 4744
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81152635207791e-05,
+      "loss": 23.1586,
+      "step": 4745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81097988954976e-05,
+      "loss": 0.7613,
+      "step": 4746
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.810433318371451e-05,
+      "loss": 22.1232,
+      "step": 4747
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.809886638558567e-05,
+      "loss": 0.7735,
+      "step": 4748
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80933985012669e-05,
+      "loss": 0.8159,
+      "step": 4749
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80879295309141e-05,
+      "loss": 0.7481,
+      "step": 4750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808245947468321e-05,
+      "loss": 0.7525,
+      "step": 4751
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.807698833273017e-05,
+      "loss": 0.7627,
+      "step": 4752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.807151610521094e-05,
+      "loss": 19.8963,
+      "step": 4753
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.806604279228157e-05,
+      "loss": 21.1927,
+      "step": 4754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.806056839409806e-05,
+      "loss": 22.8636,
+      "step": 4755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80550929108165e-05,
+      "loss": 0.7354,
+      "step": 4756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.804961634259303e-05,
+      "loss": 20.1092,
+      "step": 4757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.804413868958374e-05,
+      "loss": 20.5678,
+      "step": 4758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.803865995194482e-05,
+      "loss": 0.7737,
+      "step": 4759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.803318012983245e-05,
+      "loss": 0.7477,
+      "step": 4760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802769922340287e-05,
+      "loss": 0.755,
+      "step": 4761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802221723281234e-05,
+      "loss": 20.2052,
+      "step": 4762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801673415821714e-05,
+      "loss": 23.048,
+      "step": 4763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801124999977362e-05,
+      "loss": 0.7524,
+      "step": 4764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800576475763808e-05,
+      "loss": 0.1115,
+      "step": 4765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800027843196696e-05,
+      "loss": 0.6933,
+      "step": 4766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.799479102291665e-05,
+      "loss": 22.7818,
+      "step": 4767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798930253064359e-05,
+      "loss": 20.2852,
+      "step": 4768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798381295530424e-05,
+      "loss": 22.3836,
+      "step": 4769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.797832229705515e-05,
+      "loss": 21.1238,
+      "step": 4770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.797283055605282e-05,
+      "loss": 0.0998,
+      "step": 4771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.796733773245384e-05,
+      "loss": 21.9159,
+      "step": 4772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79618438264148e-05,
+      "loss": 0.8077,
+      "step": 4773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.795634883809232e-05,
+      "loss": 0.7689,
+      "step": 4774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.795085276764306e-05,
+      "loss": 24.2074,
+      "step": 4775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.794535561522375e-05,
+      "loss": 0.0765,
+      "step": 4776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.793985738099106e-05,
+      "loss": 0.7679,
+      "step": 4777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.793435806510178e-05,
+      "loss": 20.5047,
+      "step": 4778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.792885766771268e-05,
+      "loss": 0.0906,
+      "step": 4779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79233561889806e-05,
+      "loss": 0.7631,
+      "step": 4780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791785362906233e-05,
+      "loss": 0.0738,
+      "step": 4781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79123499881148e-05,
+      "loss": 0.7298,
+      "step": 4782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.790684526629489e-05,
+      "loss": 0.7157,
+      "step": 4783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.790133946375956e-05,
+      "loss": 22.3336,
+      "step": 4784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.789583258066576e-05,
+      "loss": 25.6023,
+      "step": 4785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.789032461717048e-05,
+      "loss": 0.0733,
+      "step": 4786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788481557343079e-05,
+      "loss": 22.2386,
+      "step": 4787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.787930544960371e-05,
+      "loss": 0.0895,
+      "step": 4788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.787379424584637e-05,
+      "loss": 0.8196,
+      "step": 4789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.786828196231585e-05,
+      "loss": 0.0587,
+      "step": 4790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.786276859916933e-05,
+      "loss": 19.9967,
+      "step": 4791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.785725415656401e-05,
+      "loss": 0.72,
+      "step": 4792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.785173863465708e-05,
+      "loss": 0.7583,
+      "step": 4793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784622203360579e-05,
+      "loss": 0.7768,
+      "step": 4794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784070435356741e-05,
+      "loss": 0.7649,
+      "step": 4795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.783518559469928e-05,
+      "loss": 21.6772,
+      "step": 4796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78296657571587e-05,
+      "loss": 23.3947,
+      "step": 4797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.782414484110306e-05,
+      "loss": 0.7649,
+      "step": 4798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781862284668977e-05,
+      "loss": 23.202,
+      "step": 4799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781309977407623e-05,
+      "loss": 0.8216,
+      "step": 4800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.780757562341994e-05,
+      "loss": 19.6927,
+      "step": 4801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.780205039487836e-05,
+      "loss": 0.0707,
+      "step": 4802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.779652408860902e-05,
+      "loss": 22.663,
+      "step": 4803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.779099670476947e-05,
+      "loss": 0.7269,
+      "step": 4804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778546824351732e-05,
+      "loss": 21.88,
+      "step": 4805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.777993870501015e-05,
+      "loss": 19.3439,
+      "step": 4806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.777440808940564e-05,
+      "loss": 0.7533,
+      "step": 4807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.776887639686144e-05,
+      "loss": 0.8102,
+      "step": 4808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.776334362753526e-05,
+      "loss": 21.227,
+      "step": 4809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775780978158487e-05,
+      "loss": 0.0692,
+      "step": 4810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775227485916799e-05,
+      "loss": 0.7068,
+      "step": 4811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774673886044246e-05,
+      "loss": 22.2375,
+      "step": 4812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774120178556608e-05,
+      "loss": 20.506,
+      "step": 4813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.773566363469673e-05,
+      "loss": 23.3685,
+      "step": 4814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.773012440799229e-05,
+      "loss": 0.7725,
+      "step": 4815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77245841056107e-05,
+      "loss": 0.7692,
+      "step": 4816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.771904272770988e-05,
+      "loss": 0.7761,
+      "step": 4817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.771350027444786e-05,
+      "loss": 20.0042,
+      "step": 4818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.770795674598262e-05,
+      "loss": 22.2704,
+      "step": 4819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.770241214247219e-05,
+      "loss": 0.7502,
+      "step": 4820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769686646407469e-05,
+      "loss": 20.0808,
+      "step": 4821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76913197109482e-05,
+      "loss": 23.2297,
+      "step": 4822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768577188325085e-05,
+      "loss": 20.7457,
+      "step": 4823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768022298114082e-05,
+      "loss": 21.8845,
+      "step": 4824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.767467300477633e-05,
+      "loss": 22.958,
+      "step": 4825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766912195431556e-05,
+      "loss": 0.8159,
+      "step": 4826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766356982991679e-05,
+      "loss": 22.1645,
+      "step": 4827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765801663173832e-05,
+      "loss": 0.829,
+      "step": 4828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765246235993847e-05,
+      "loss": 20.322,
+      "step": 4829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.764690701467557e-05,
+      "loss": 0.7156,
+      "step": 4830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.764135059610802e-05,
+      "loss": 19.7813,
+      "step": 4831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.763579310439425e-05,
+      "loss": 0.6977,
+      "step": 4832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.763023453969265e-05,
+      "loss": 0.7777,
+      "step": 4833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.762467490216176e-05,
+      "loss": 21.3154,
+      "step": 4834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761911419196002e-05,
+      "loss": 0.0841,
+      "step": 4835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761355240924601e-05,
+      "loss": 22.8746,
+      "step": 4836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.760798955417827e-05,
+      "loss": 0.7932,
+      "step": 4837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76024256269154e-05,
+      "loss": 0.7491,
+      "step": 4838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759686062761603e-05,
+      "loss": 0.7311,
+      "step": 4839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759129455643883e-05,
+      "loss": 0.6969,
+      "step": 4840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.758572741354247e-05,
+      "loss": 21.2833,
+      "step": 4841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.758015919908566e-05,
+      "loss": 23.4422,
+      "step": 4842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.757458991322718e-05,
+      "loss": 23.6797,
+      "step": 4843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.756901955612578e-05,
+      "loss": 0.7532,
+      "step": 4844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.756344812794028e-05,
+      "loss": 21.0438,
+      "step": 4845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755787562882951e-05,
+      "loss": 0.6945,
+      "step": 4846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755230205895236e-05,
+      "loss": 0.0539,
+      "step": 4847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754672741846772e-05,
+      "loss": 0.7482,
+      "step": 4848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754115170753452e-05,
+      "loss": 0.7214,
+      "step": 4849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753557492631174e-05,
+      "loss": 0.7368,
+      "step": 4850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.752999707495835e-05,
+      "loss": 0.8208,
+      "step": 4851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.752441815363337e-05,
+      "loss": 0.7197,
+      "step": 4852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.751883816249588e-05,
+      "loss": 0.7415,
+      "step": 4853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.751325710170494e-05,
+      "loss": 0.7153,
+      "step": 4854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75076749714197e-05,
+      "loss": 0.7673,
+      "step": 4855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.750209177179924e-05,
+      "loss": 0.0905,
+      "step": 4856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74965075030028e-05,
+      "loss": 0.8203,
+      "step": 4857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.749092216518956e-05,
+      "loss": 0.8097,
+      "step": 4858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.748533575851876e-05,
+      "loss": 24.0025,
+      "step": 4859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.747974828314967e-05,
+      "loss": 23.1042,
+      "step": 4860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.747415973924155e-05,
+      "loss": 21.6901,
+      "step": 4861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74685701269538e-05,
+      "loss": 0.7023,
+      "step": 4862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746297944644573e-05,
+      "loss": 22.25,
+      "step": 4863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745738769787672e-05,
+      "loss": 0.7238,
+      "step": 4864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745179488140622e-05,
+      "loss": 21.1293,
+      "step": 4865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.744620099719368e-05,
+      "loss": 0.7178,
+      "step": 4866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.744060604539855e-05,
+      "loss": 0.8056,
+      "step": 4867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.743501002618037e-05,
+      "loss": 0.7581,
+      "step": 4868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.742941293969868e-05,
+      "loss": 0.7544,
+      "step": 4869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.742381478611303e-05,
+      "loss": 22.9203,
+      "step": 4870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.741821556558303e-05,
+      "loss": 0.8042,
+      "step": 4871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.741261527826831e-05,
+      "loss": 22.1374,
+      "step": 4872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740701392432855e-05,
+      "loss": 20.9989,
+      "step": 4873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740141150392345e-05,
+      "loss": 0.762,
+      "step": 4874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73958080172127e-05,
+      "loss": 0.0666,
+      "step": 4875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739020346435607e-05,
+      "loss": 24.7317,
+      "step": 4876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.738459784551335e-05,
+      "loss": 0.7081,
+      "step": 4877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.737899116084433e-05,
+      "loss": 0.7441,
+      "step": 4878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73733834105089e-05,
+      "loss": 0.7492,
+      "step": 4879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.736777459466691e-05,
+      "loss": 22.8061,
+      "step": 4880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.736216471347827e-05,
+      "loss": 22.5228,
+      "step": 4881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73565537671029e-05,
+      "loss": 20.7231,
+      "step": 4882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.735094175570078e-05,
+      "loss": 0.7025,
+      "step": 4883
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.734532867943191e-05,
+      "loss": 24.1891,
+      "step": 4884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733971453845632e-05,
+      "loss": 0.7616,
+      "step": 4885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733409933293406e-05,
+      "loss": 21.1509,
+      "step": 4886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.732848306302522e-05,
+      "loss": 0.7168,
+      "step": 4887
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.732286572888992e-05,
+      "loss": 21.345,
+      "step": 4888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.731724733068831e-05,
+      "loss": 0.7033,
+      "step": 4889
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.731162786858056e-05,
+      "loss": 0.7572,
+      "step": 4890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.730600734272691e-05,
+      "loss": 21.191,
+      "step": 4891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.730038575328756e-05,
+      "loss": 19.7089,
+      "step": 4892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72947631004228e-05,
+      "loss": 0.0808,
+      "step": 4893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728913938429292e-05,
+      "loss": 22.4039,
+      "step": 4894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728351460505828e-05,
+      "loss": 22.4609,
+      "step": 4895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.727788876287921e-05,
+      "loss": 0.7729,
+      "step": 4896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72722618579161e-05,
+      "loss": 20.9871,
+      "step": 4897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72666338903294e-05,
+      "loss": 0.8114,
+      "step": 4898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726100486027954e-05,
+      "loss": 21.7029,
+      "step": 4899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.7255374767927e-05,
+      "loss": 0.0712,
+      "step": 4900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.724974361343231e-05,
+      "loss": 0.7657,
+      "step": 4901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.724411139695601e-05,
+      "loss": 21.3443,
+      "step": 4902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.723847811865864e-05,
+      "loss": 0.7549,
+      "step": 4903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.723284377870085e-05,
+      "loss": 0.061,
+      "step": 4904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.722720837724325e-05,
+      "loss": 0.7612,
+      "step": 4905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72215719144465e-05,
+      "loss": 0.7669,
+      "step": 4906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.721593439047133e-05,
+      "loss": 20.6211,
+      "step": 4907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72102958054784e-05,
+      "loss": 21.6209,
+      "step": 4908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720465615962852e-05,
+      "loss": 20.3472,
+      "step": 4909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719901545308246e-05,
+      "loss": 20.8873,
+      "step": 4910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719337368600102e-05,
+      "loss": 0.0643,
+      "step": 4911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.718773085854506e-05,
+      "loss": 21.5606,
+      "step": 4912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.718208697087545e-05,
+      "loss": 0.7569,
+      "step": 4913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.717644202315311e-05,
+      "loss": 21.3539,
+      "step": 4914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.717079601553896e-05,
+      "loss": 0.7047,
+      "step": 4915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.716514894819396e-05,
+      "loss": 19.0725,
+      "step": 4916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.715950082127914e-05,
+      "loss": 0.0706,
+      "step": 4917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.715385163495548e-05,
+      "loss": 18.4951,
+      "step": 4918
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714820138938408e-05,
+      "loss": 0.6984,
+      "step": 4919
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.7142550084726e-05,
+      "loss": 0.7522,
+      "step": 4920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.713689772114238e-05,
+      "loss": 0.7534,
+      "step": 4921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.713124429879433e-05,
+      "loss": 0.7251,
+      "step": 4922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.712558981784306e-05,
+      "loss": 21.6639,
+      "step": 4923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711993427844977e-05,
+      "loss": 0.7339,
+      "step": 4924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711427768077572e-05,
+      "loss": 18.936,
+      "step": 4925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710862002498213e-05,
+      "loss": 0.6988,
+      "step": 4926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710296131123032e-05,
+      "loss": 21.3606,
+      "step": 4927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709730153968165e-05,
+      "loss": 21.0916,
+      "step": 4928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709164071049743e-05,
+      "loss": 22.4556,
+      "step": 4929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.708597882383908e-05,
+      "loss": 20.345,
+      "step": 4930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.7080315879868e-05,
+      "loss": 0.7666,
+      "step": 4931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.707465187874566e-05,
+      "loss": 21.3503,
+      "step": 4932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706898682063351e-05,
+      "loss": 0.06,
+      "step": 4933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70633207056931e-05,
+      "loss": 0.7407,
+      "step": 4934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.705765353408594e-05,
+      "loss": 0.7599,
+      "step": 4935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70519853059736e-05,
+      "loss": 19.9315,
+      "step": 4936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.704631602151769e-05,
+      "loss": 21.7496,
+      "step": 4937
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.704064568087984e-05,
+      "loss": 0.7613,
+      "step": 4938
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.703497428422172e-05,
+      "loss": 20.9415,
+      "step": 4939
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.702930183170498e-05,
+      "loss": 21.504,
+      "step": 4940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.702362832349139e-05,
+      "loss": 0.7467,
+      "step": 4941
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.701795375974268e-05,
+      "loss": 19.6754,
+      "step": 4942
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70122781406206e-05,
+      "loss": 20.9462,
+      "step": 4943
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700660146628701e-05,
+      "loss": 0.751,
+      "step": 4944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700092373690374e-05,
+      "loss": 0.1064,
+      "step": 4945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.699524495263264e-05,
+      "loss": 20.6534,
+      "step": 4946
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.698956511363563e-05,
+      "loss": 23.3247,
+      "step": 4947
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.698388422007462e-05,
+      "loss": 20.0777,
+      "step": 4948
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697820227211158e-05,
+      "loss": 0.7269,
+      "step": 4949
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697251926990852e-05,
+      "loss": 20.6585,
+      "step": 4950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696683521362743e-05,
+      "loss": 22.7443,
+      "step": 4951
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696115010343039e-05,
+      "loss": 0.0641,
+      "step": 4952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695546393947947e-05,
+      "loss": 19.5806,
+      "step": 4953
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.694977672193677e-05,
+      "loss": 0.7968,
+      "step": 4954
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.694408845096442e-05,
+      "loss": 19.6355,
+      "step": 4955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.693839912672464e-05,
+      "loss": 20.8658,
+      "step": 4956
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69327087493796e-05,
+      "loss": 20.9209,
+      "step": 4957
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.692701731909153e-05,
+      "loss": 21.891,
+      "step": 4958
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69213248360227e-05,
+      "loss": 20.832,
+      "step": 4959
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.691563130033539e-05,
+      "loss": 0.7529,
+      "step": 4960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.690993671219193e-05,
+      "loss": 0.7184,
+      "step": 4961
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.690424107175467e-05,
+      "loss": 0.0742,
+      "step": 4962
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.689854437918598e-05,
+      "loss": 0.7246,
+      "step": 4963
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68928466346483e-05,
+      "loss": 22.1246,
+      "step": 4964
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.688714783830405e-05,
+      "loss": 0.7058,
+      "step": 4965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.688144799031571e-05,
+      "loss": 0.7673,
+      "step": 4966
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.687574709084577e-05,
+      "loss": 0.7362,
+      "step": 4967
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.687004514005677e-05,
+      "loss": 0.7481,
+      "step": 4968
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686434213811127e-05,
+      "loss": 23.2328,
+      "step": 4969
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.685863808517186e-05,
+      "loss": 20.5393,
+      "step": 4970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.685293298140116e-05,
+      "loss": 22.4764,
+      "step": 4971
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.684722682696183e-05,
+      "loss": 24.427,
+      "step": 4972
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.684151962201655e-05,
+      "loss": 0.7736,
+      "step": 4973
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6835811366728e-05,
+      "loss": 21.7317,
+      "step": 4974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683010206125896e-05,
+      "loss": 20.8491,
+      "step": 4975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68243917057722e-05,
+      "loss": 0.7385,
+      "step": 4976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681868030043049e-05,
+      "loss": 0.0621,
+      "step": 4977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681296784539668e-05,
+      "loss": 20.2678,
+      "step": 4978
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.680725434083363e-05,
+      "loss": 0.7746,
+      "step": 4979
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.680153978690425e-05,
+      "loss": 0.7199,
+      "step": 4980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.679582418377143e-05,
+      "loss": 0.7219,
+      "step": 4981
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.679010753159812e-05,
+      "loss": 21.3885,
+      "step": 4982
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.678438983054732e-05,
+      "loss": 0.7401,
+      "step": 4983
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.677867108078204e-05,
+      "loss": 22.8815,
+      "step": 4984
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.677295128246529e-05,
+      "loss": 22.8979,
+      "step": 4985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676723043576019e-05,
+      "loss": 21.634,
+      "step": 4986
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67615085408298e-05,
+      "loss": 21.7357,
+      "step": 4987
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675578559783726e-05,
+      "loss": 0.7432,
+      "step": 4988
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675006160694575e-05,
+      "loss": 22.9732,
+      "step": 4989
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.674433656831842e-05,
+      "loss": 23.0411,
+      "step": 4990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673861048211853e-05,
+      "loss": 0.7679,
+      "step": 4991
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673288334850931e-05,
+      "loss": 0.061,
+      "step": 4992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.672715516765403e-05,
+      "loss": 21.0348,
+      "step": 4993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.672142593971602e-05,
+      "loss": 21.0372,
+      "step": 4994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67156956648586e-05,
+      "loss": 0.7602,
+      "step": 4995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.670996434324514e-05,
+      "loss": 0.0861,
+      "step": 4996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.670423197503906e-05,
+      "loss": 0.7605,
+      "step": 4997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669849856040378e-05,
+      "loss": 20.2746,
+      "step": 4998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669276409950274e-05,
+      "loss": 20.2886,
+      "step": 4999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.668702859249944e-05,
+      "loss": 21.8783,
+      "step": 5000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.668129203955739e-05,
+      "loss": 0.7721,
+      "step": 5001
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.667555444084017e-05,
+      "loss": 0.7694,
+      "step": 5002
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666981579651132e-05,
+      "loss": 0.0762,
+      "step": 5003
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666407610673448e-05,
+      "loss": 22.8679,
+      "step": 5004
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.665833537167325e-05,
+      "loss": 21.0592,
+      "step": 5005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.665259359149132e-05,
+      "loss": 0.75,
+      "step": 5006
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664685076635238e-05,
+      "loss": 22.0616,
+      "step": 5007
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664110689642018e-05,
+      "loss": 21.5868,
+      "step": 5008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.663536198185843e-05,
+      "loss": 21.625,
+      "step": 5009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662961602283094e-05,
+      "loss": 0.7108,
+      "step": 5010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662386901950156e-05,
+      "loss": 24.5686,
+      "step": 5011
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.661812097203408e-05,
+      "loss": 21.3932,
+      "step": 5012
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.661237188059242e-05,
+      "loss": 0.7758,
+      "step": 5013
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660662174534044e-05,
+      "loss": 19.5888,
+      "step": 5014
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66008705664421e-05,
+      "loss": 0.743,
+      "step": 5015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659511834406138e-05,
+      "loss": 21.8327,
+      "step": 5016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.658936507836226e-05,
+      "loss": 22.6244,
+      "step": 5017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.658361076950873e-05,
+      "loss": 21.5995,
+      "step": 5018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657785541766492e-05,
+      "loss": 20.0339,
+      "step": 5019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657209902299485e-05,
+      "loss": 21.0421,
+      "step": 5020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.656634158566265e-05,
+      "loss": 20.9957,
+      "step": 5021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.656058310583248e-05,
+      "loss": 19.6482,
+      "step": 5022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65548235836685e-05,
+      "loss": 22.3925,
+      "step": 5023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654906301933492e-05,
+      "loss": 0.7641,
+      "step": 5024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654330141299596e-05,
+      "loss": 22.4862,
+      "step": 5025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.653753876481587e-05,
+      "loss": 0.7153,
+      "step": 5026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.653177507495898e-05,
+      "loss": 0.0869,
+      "step": 5027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65260103435896e-05,
+      "loss": 0.0811,
+      "step": 5028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.652024457087205e-05,
+      "loss": 19.6348,
+      "step": 5029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.651447775697074e-05,
+      "loss": 19.9967,
+      "step": 5030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.650870990205007e-05,
+      "loss": 21.5433,
+      "step": 5031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65029410062745e-05,
+      "loss": 20.2824,
+      "step": 5032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.649717106980846e-05,
+      "loss": 20.8938,
+      "step": 5033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.649140009281648e-05,
+      "loss": 0.7697,
+      "step": 5034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64856280754631e-05,
+      "loss": 19.9429,
+      "step": 5035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647985501791284e-05,
+      "loss": 21.6441,
+      "step": 5036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647408092033033e-05,
+      "loss": 0.7549,
+      "step": 5037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646830578288015e-05,
+      "loss": 18.9984,
+      "step": 5038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646252960572699e-05,
+      "loss": 20.248,
+      "step": 5039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645675238903547e-05,
+      "loss": 22.9562,
+      "step": 5040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645097413297035e-05,
+      "loss": 20.7277,
+      "step": 5041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.644519483769635e-05,
+      "loss": 0.0804,
+      "step": 5042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.643941450337823e-05,
+      "loss": 21.5359,
+      "step": 5043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.643363313018077e-05,
+      "loss": 20.7293,
+      "step": 5044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642785071826884e-05,
+      "loss": 0.744,
+      "step": 5045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642206726780728e-05,
+      "loss": 0.7069,
+      "step": 5046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.641628277896094e-05,
+      "loss": 0.7167,
+      "step": 5047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.641049725189478e-05,
+      "loss": 0.7772,
+      "step": 5048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.640471068677371e-05,
+      "loss": 20.8692,
+      "step": 5049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639892308376272e-05,
+      "loss": 0.751,
+      "step": 5050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639313444302682e-05,
+      "loss": 20.5561,
+      "step": 5051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.638734476473102e-05,
+      "loss": 19.94,
+      "step": 5052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.638155404904041e-05,
+      "loss": 21.0185,
+      "step": 5053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.637576229612004e-05,
+      "loss": 0.7295,
+      "step": 5054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636996950613508e-05,
+      "loss": 21.2726,
+      "step": 5055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636417567925066e-05,
+      "loss": 0.6815,
+      "step": 5056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.635838081563197e-05,
+      "loss": 20.882,
+      "step": 5057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.635258491544421e-05,
+      "loss": 20.7509,
+      "step": 5058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.634678797885261e-05,
+      "loss": 22.623,
+      "step": 5059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.634099000602245e-05,
+      "loss": 21.7538,
+      "step": 5060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.633519099711905e-05,
+      "loss": 21.3855,
+      "step": 5061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632939095230768e-05,
+      "loss": 0.0747,
+      "step": 5062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632358987175376e-05,
+      "loss": 19.7488,
+      "step": 5063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.631778775562266e-05,
+      "loss": 20.4609,
+      "step": 5064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.631198460407978e-05,
+      "loss": 22.575,
+      "step": 5065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.630618041729057e-05,
+      "loss": 0.7777,
+      "step": 5066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.630037519542052e-05,
+      "loss": 22.8336,
+      "step": 5067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.629456893863513e-05,
+      "loss": 19.7925,
+      "step": 5068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.628876164709993e-05,
+      "loss": 0.7382,
+      "step": 5069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62829533209805e-05,
+      "loss": 0.7937,
+      "step": 5070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.627714396044242e-05,
+      "loss": 0.7885,
+      "step": 5071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62713335656513e-05,
+      "loss": 23.373,
+      "step": 5072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626552213677281e-05,
+      "loss": 20.5558,
+      "step": 5073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.625970967397265e-05,
+      "loss": 23.0554,
+      "step": 5074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62538961774165e-05,
+      "loss": 0.7081,
+      "step": 5075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.624808164727011e-05,
+      "loss": 0.0774,
+      "step": 5076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.624226608369927e-05,
+      "loss": 0.75,
+      "step": 5077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.623644948686975e-05,
+      "loss": 21.0128,
+      "step": 5078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62306318569474e-05,
+      "loss": 21.4712,
+      "step": 5079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.622481319409807e-05,
+      "loss": 23.3745,
+      "step": 5080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.621899349848768e-05,
+      "loss": 20.9725,
+      "step": 5081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.621317277028209e-05,
+      "loss": 0.85,
+      "step": 5082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62073510096473e-05,
+      "loss": 0.7399,
+      "step": 5083
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.620152821674925e-05,
+      "loss": 20.0468,
+      "step": 5084
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.619570439175399e-05,
+      "loss": 0.7417,
+      "step": 5085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.618987953482751e-05,
+      "loss": 20.8285,
+      "step": 5086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.618405364613589e-05,
+      "loss": 22.0082,
+      "step": 5087
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617822672584524e-05,
+      "loss": 19.1539,
+      "step": 5088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617239877412166e-05,
+      "loss": 0.7439,
+      "step": 5089
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.616656979113133e-05,
+      "loss": 22.2731,
+      "step": 5090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.616073977704041e-05,
+      "loss": 0.7656,
+      "step": 5091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.615490873201512e-05,
+      "loss": 22.1183,
+      "step": 5092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614907665622171e-05,
+      "loss": 21.8709,
+      "step": 5093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614324354982644e-05,
+      "loss": 22.4369,
+      "step": 5094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.613740941299563e-05,
+      "loss": 0.7048,
+      "step": 5095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.613157424589559e-05,
+      "loss": 21.1351,
+      "step": 5096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.612573804869268e-05,
+      "loss": 21.4043,
+      "step": 5097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.611990082155329e-05,
+      "loss": 0.7799,
+      "step": 5098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.611406256464385e-05,
+      "loss": 21.7875,
+      "step": 5099
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.610822327813079e-05,
+      "loss": 21.8748,
+      "step": 5100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61023829621806e-05,
+      "loss": 0.0656,
+      "step": 5101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.609654161695977e-05,
+      "loss": 0.0801,
+      "step": 5102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.609069924263486e-05,
+      "loss": 20.976,
+      "step": 5103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.608485583937242e-05,
+      "loss": 0.0595,
+      "step": 5104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.607901140733905e-05,
+      "loss": 0.708,
+      "step": 5105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.607316594670137e-05,
+      "loss": 21.6976,
+      "step": 5106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.606731945762602e-05,
+      "loss": 22.6241,
+      "step": 5107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.606147194027969e-05,
+      "loss": 0.7158,
+      "step": 5108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.605562339482911e-05,
+      "loss": 20.5047,
+      "step": 5109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604977382144101e-05,
+      "loss": 20.838,
+      "step": 5110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604392322028214e-05,
+      "loss": 0.0737,
+      "step": 5111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.603807159151931e-05,
+      "loss": 21.0483,
+      "step": 5112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.603221893531937e-05,
+      "loss": 21.3673,
+      "step": 5113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602636525184914e-05,
+      "loss": 0.0693,
+      "step": 5114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602051054127556e-05,
+      "loss": 20.7776,
+      "step": 5115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.601465480376551e-05,
+      "loss": 21.5312,
+      "step": 5116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.600879803948591e-05,
+      "loss": 0.075,
+      "step": 5117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.60029402486038e-05,
+      "loss": 21.1933,
+      "step": 5118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599708143128614e-05,
+      "loss": 22.1222,
+      "step": 5119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599122158769997e-05,
+      "loss": 18.9705,
+      "step": 5120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.598536071801237e-05,
+      "loss": 18.7633,
+      "step": 5121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59794988223904e-05,
+      "loss": 20.8764,
+      "step": 5122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59736359010012e-05,
+      "loss": 21.7236,
+      "step": 5123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.596777195401193e-05,
+      "loss": 20.0995,
+      "step": 5124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.596190698158975e-05,
+      "loss": 0.1132,
+      "step": 5125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59560409839019e-05,
+      "loss": 0.754,
+      "step": 5126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.595017396111557e-05,
+      "loss": 0.0772,
+      "step": 5127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.594430591339808e-05,
+      "loss": 21.9264,
+      "step": 5128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.593843684091669e-05,
+      "loss": 0.7533,
+      "step": 5129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.593256674383874e-05,
+      "loss": 0.785,
+      "step": 5130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.592669562233158e-05,
+      "loss": 19.7028,
+      "step": 5131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59208234765626e-05,
+      "loss": 21.7336,
+      "step": 5132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.591495030669922e-05,
+      "loss": 20.3419,
+      "step": 5133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.590907611290886e-05,
+      "loss": 0.0826,
+      "step": 5134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.590320089535899e-05,
+      "loss": 0.0517,
+      "step": 5135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.589732465421716e-05,
+      "loss": 0.073,
+      "step": 5136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.589144738965085e-05,
+      "loss": 0.7182,
+      "step": 5137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.588556910182763e-05,
+      "loss": 21.9581,
+      "step": 5138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58796897909151e-05,
+      "loss": 22.7461,
+      "step": 5139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.587380945708088e-05,
+      "loss": 0.6752,
+      "step": 5140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58679281004926e-05,
+      "loss": 19.3026,
+      "step": 5141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.586204572131795e-05,
+      "loss": 23.056,
+      "step": 5142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585616231972462e-05,
+      "loss": 19.9072,
+      "step": 5143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585027789588037e-05,
+      "loss": 21.9058,
+      "step": 5144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.584439244995295e-05,
+      "loss": 0.7551,
+      "step": 5145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.583850598211013e-05,
+      "loss": 22.4656,
+      "step": 5146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.583261849251977e-05,
+      "loss": 0.7659,
+      "step": 5147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58267299813497e-05,
+      "loss": 0.6969,
+      "step": 5148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582084044876782e-05,
+      "loss": 22.3502,
+      "step": 5149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5814949894942e-05,
+      "loss": 20.2628,
+      "step": 5150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.580905832004022e-05,
+      "loss": 0.7742,
+      "step": 5151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.580316572423042e-05,
+      "loss": 0.0678,
+      "step": 5152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.579727210768062e-05,
+      "loss": 21.6147,
+      "step": 5153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57913774705588e-05,
+      "loss": 20.6379,
+      "step": 5154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.578548181303308e-05,
+      "loss": 19.7218,
+      "step": 5155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577958513527149e-05,
+      "loss": 19.7686,
+      "step": 5156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577368743744218e-05,
+      "loss": 22.0287,
+      "step": 5157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.576778871971326e-05,
+      "loss": 23.869,
+      "step": 5158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.576188898225293e-05,
+      "loss": 22.9964,
+      "step": 5159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.575598822522936e-05,
+      "loss": 0.0593,
+      "step": 5160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57500864488108e-05,
+      "loss": 20.9273,
+      "step": 5161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574418365316551e-05,
+      "loss": 0.7846,
+      "step": 5162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.573827983846177e-05,
+      "loss": 24.1877,
+      "step": 5163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.573237500486788e-05,
+      "loss": 21.936,
+      "step": 5164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.572646915255223e-05,
+      "loss": 0.0781,
+      "step": 5165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.572056228168315e-05,
+      "loss": 22.2747,
+      "step": 5166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571465439242908e-05,
+      "loss": 19.5844,
+      "step": 5167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.570874548495842e-05,
+      "loss": 0.7566,
+      "step": 5168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.570283555943965e-05,
+      "loss": 22.7144,
+      "step": 5169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.569692461604124e-05,
+      "loss": 0.6883,
+      "step": 5170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.569101265493175e-05,
+      "loss": 0.0628,
+      "step": 5171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.568509967627969e-05,
+      "loss": 0.6813,
+      "step": 5172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567918568025366e-05,
+      "loss": 21.0429,
+      "step": 5173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567327066702225e-05,
+      "loss": 21.49,
+      "step": 5174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56673546367541e-05,
+      "loss": 0.7435,
+      "step": 5175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56614375896179e-05,
+      "loss": 20.2516,
+      "step": 5176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.565551952578229e-05,
+      "loss": 20.1995,
+      "step": 5177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.564960044541605e-05,
+      "loss": 22.1692,
+      "step": 5178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56436803486879e-05,
+      "loss": 20.1879,
+      "step": 5179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.563775923576663e-05,
+      "loss": 0.7854,
+      "step": 5180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.563183710682103e-05,
+      "loss": 0.09,
+      "step": 5181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.562591396201998e-05,
+      "loss": 0.7657,
+      "step": 5182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56199898015323e-05,
+      "loss": 21.3715,
+      "step": 5183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561406462552692e-05,
+      "loss": 20.058,
+      "step": 5184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.560813843417277e-05,
+      "loss": 0.0717,
+      "step": 5185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.560221122763877e-05,
+      "loss": 20.7789,
+      "step": 5186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559628300609393e-05,
+      "loss": 19.5119,
+      "step": 5187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559035376970725e-05,
+      "loss": 0.7654,
+      "step": 5188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.558442351864778e-05,
+      "loss": 0.0971,
+      "step": 5189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557849225308458e-05,
+      "loss": 20.0395,
+      "step": 5190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557255997318676e-05,
+      "loss": 21.6613,
+      "step": 5191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556662667912346e-05,
+      "loss": 0.0564,
+      "step": 5192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556069237106379e-05,
+      "loss": 21.3527,
+      "step": 5193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.555475704917698e-05,
+      "loss": 0.0631,
+      "step": 5194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.554882071363223e-05,
+      "loss": 0.7325,
+      "step": 5195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.554288336459878e-05,
+      "loss": 0.0804,
+      "step": 5196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553694500224592e-05,
+      "loss": 20.5878,
+      "step": 5197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553100562674292e-05,
+      "loss": 0.7337,
+      "step": 5198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.552506523825912e-05,
+      "loss": 22.4024,
+      "step": 5199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.55191238369639e-05,
+      "loss": 20.8378,
+      "step": 5200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.551318142302665e-05,
+      "loss": 21.4664,
+      "step": 5201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.550723799661675e-05,
+      "loss": 18.9498,
+      "step": 5202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.550129355790367e-05,
+      "loss": 0.7864,
+      "step": 5203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.549534810705689e-05,
+      "loss": 20.8609,
+      "step": 5204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.548940164424589e-05,
+      "loss": 20.6003,
+      "step": 5205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.548345416964023e-05,
+      "loss": 0.087,
+      "step": 5206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.547750568340946e-05,
+      "loss": 0.7939,
+      "step": 5207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.547155618572316e-05,
+      "loss": 20.6318,
+      "step": 5208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.546560567675098e-05,
+      "loss": 0.0709,
+      "step": 5209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545965415666252e-05,
+      "loss": 21.8316,
+      "step": 5210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54537016256275e-05,
+      "loss": 19.7635,
+      "step": 5211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54477480838156e-05,
+      "loss": 0.7331,
+      "step": 5212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.544179353139654e-05,
+      "loss": 20.0035,
+      "step": 5213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543583796854014e-05,
+      "loss": 0.784,
+      "step": 5214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.542988139541612e-05,
+      "loss": 20.6893,
+      "step": 5215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.542392381219437e-05,
+      "loss": 0.7655,
+      "step": 5216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541796521904468e-05,
+      "loss": 0.7502,
+      "step": 5217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541200561613697e-05,
+      "loss": 19.9455,
+      "step": 5218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.540604500364113e-05,
+      "loss": 20.6169,
+      "step": 5219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.540008338172709e-05,
+      "loss": 21.3489,
+      "step": 5220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.539412075056482e-05,
+      "loss": 19.8073,
+      "step": 5221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.53881571103243e-05,
+      "loss": 0.7397,
+      "step": 5222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.538219246117561e-05,
+      "loss": 0.7359,
+      "step": 5223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.537622680328872e-05,
+      "loss": 0.7462,
+      "step": 5224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.537026013683376e-05,
+      "loss": 19.6102,
+      "step": 5225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.536429246198081e-05,
+      "loss": 20.6364,
+      "step": 5226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.535832377890003e-05,
+      "loss": 0.7301,
+      "step": 5227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.535235408776159e-05,
+      "loss": 22.628,
+      "step": 5228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534638338873565e-05,
+      "loss": 0.7207,
+      "step": 5229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534041168199247e-05,
+      "loss": 20.045,
+      "step": 5230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.533443896770228e-05,
+      "loss": 22.0742,
+      "step": 5231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532846524603538e-05,
+      "loss": 20.656,
+      "step": 5232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532249051716205e-05,
+      "loss": 21.4578,
+      "step": 5233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531651478125265e-05,
+      "loss": 19.1749,
+      "step": 5234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531053803847755e-05,
+      "loss": 19.5867,
+      "step": 5235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.530456028900713e-05,
+      "loss": 0.0704,
+      "step": 5236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.529858153301183e-05,
+      "loss": 19.3891,
+      "step": 5237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.529260177066209e-05,
+      "loss": 20.4318,
+      "step": 5238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52866210021284e-05,
+      "loss": 21.3163,
+      "step": 5239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.528063922758128e-05,
+      "loss": 0.7144,
+      "step": 5240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.527465644719124e-05,
+      "loss": 0.7212,
+      "step": 5241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.526867266112886e-05,
+      "loss": 0.7708,
+      "step": 5242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.526268786956475e-05,
+      "loss": 21.4584,
+      "step": 5243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.525670207266952e-05,
+      "loss": 22.9359,
+      "step": 5244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.525071527061384e-05,
+      "loss": 0.0756,
+      "step": 5245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.524472746356836e-05,
+      "loss": 20.4421,
+      "step": 5246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523873865170383e-05,
+      "loss": 21.8872,
+      "step": 5247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523274883519097e-05,
+      "loss": 0.7621,
+      "step": 5248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.522675801420053e-05,
+      "loss": 0.7564,
+      "step": 5249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.522076618890335e-05,
+      "loss": 19.1136,
+      "step": 5250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.521477335947023e-05,
+      "loss": 19.4046,
+      "step": 5251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.520877952607203e-05,
+      "loss": 20.0005,
+      "step": 5252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.520278468887962e-05,
+      "loss": 19.762,
+      "step": 5253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51967888480639e-05,
+      "loss": 21.4659,
+      "step": 5254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.519079200379587e-05,
+      "loss": 21.6832,
+      "step": 5255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.518479415624644e-05,
+      "loss": 0.725,
+      "step": 5256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.517879530558664e-05,
+      "loss": 0.7547,
+      "step": 5257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.517279545198746e-05,
+      "loss": 0.0964,
+      "step": 5258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516679459562e-05,
+      "loss": 0.7236,
+      "step": 5259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516079273665531e-05,
+      "loss": 0.7834,
+      "step": 5260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.515478987526451e-05,
+      "loss": 22.6382,
+      "step": 5261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.514878601161875e-05,
+      "loss": 20.7829,
+      "step": 5262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.514278114588919e-05,
+      "loss": 20.4964,
+      "step": 5263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.513677527824703e-05,
+      "loss": 21.692,
+      "step": 5264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51307684088635e-05,
+      "loss": 22.2449,
+      "step": 5265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.512476053790982e-05,
+      "loss": 21.2435,
+      "step": 5266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.511875166555733e-05,
+      "loss": 22.298,
+      "step": 5267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51127417919773e-05,
+      "loss": 22.1244,
+      "step": 5268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.510673091734109e-05,
+      "loss": 19.8341,
+      "step": 5269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.510071904182006e-05,
+      "loss": 22.4906,
+      "step": 5270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.509470616558561e-05,
+      "loss": 20.324,
+      "step": 5271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.508869228880916e-05,
+      "loss": 0.6897,
+      "step": 5272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.508267741166216e-05,
+      "loss": 20.6097,
+      "step": 5273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.507666153431613e-05,
+      "loss": 21.7304,
+      "step": 5274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.507064465694252e-05,
+      "loss": 24.261,
+      "step": 5275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.506462677971293e-05,
+      "loss": 0.7415,
+      "step": 5276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.50586079027989e-05,
+      "loss": 20.0825,
+      "step": 5277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.5052588026372e-05,
+      "loss": 0.7786,
+      "step": 5278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.50465671506039e-05,
+      "loss": 0.7318,
+      "step": 5279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.504054527566625e-05,
+      "loss": 0.7189,
+      "step": 5280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.50345224017307e-05,
+      "loss": 21.2339,
+      "step": 5281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.5028498528969e-05,
+      "loss": 20.4493,
+      "step": 5282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.502247365755286e-05,
+      "loss": 0.7635,
+      "step": 5283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501644778765406e-05,
+      "loss": 0.6878,
+      "step": 5284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501042091944439e-05,
+      "loss": 0.7316,
+      "step": 5285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.500439305309567e-05,
+      "loss": 0.7434,
+      "step": 5286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.499836418877978e-05,
+      "loss": 0.7502,
+      "step": 5287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.499233432666858e-05,
+      "loss": 0.0831,
+      "step": 5288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.498630346693397e-05,
+      "loss": 19.978,
+      "step": 5289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.498027160974788e-05,
+      "loss": 0.765,
+      "step": 5290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.497423875528234e-05,
+      "loss": 20.281,
+      "step": 5291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.496820490370928e-05,
+      "loss": 0.739,
+      "step": 5292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.496217005520075e-05,
+      "loss": 0.723,
+      "step": 5293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49561342099288e-05,
+      "loss": 19.5834,
+      "step": 5294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.495009736806552e-05,
+      "loss": 20.0204,
+      "step": 5295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.494405952978299e-05,
+      "loss": 22.0651,
+      "step": 5296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.493802069525338e-05,
+      "loss": 20.5404,
+      "step": 5297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.493198086464884e-05,
+      "loss": 0.7062,
+      "step": 5298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.492594003814156e-05,
+      "loss": 0.7917,
+      "step": 5299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.491989821590379e-05,
+      "loss": 20.5816,
+      "step": 5300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.491385539810773e-05,
+      "loss": 0.7187,
+      "step": 5301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49078115849257e-05,
+      "loss": 22.5776,
+      "step": 5302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490176677652998e-05,
+      "loss": 0.0621,
+      "step": 5303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.489572097309295e-05,
+      "loss": 0.7039,
+      "step": 5304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488967417478693e-05,
+      "loss": 20.8913,
+      "step": 5305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488362638178433e-05,
+      "loss": 0.7192,
+      "step": 5306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.487757759425759e-05,
+      "loss": 20.8485,
+      "step": 5307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.487152781237912e-05,
+      "loss": 19.7242,
+      "step": 5308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486547703632142e-05,
+      "loss": 0.7661,
+      "step": 5309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4859425266257e-05,
+      "loss": 0.7466,
+      "step": 5310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.485337250235838e-05,
+      "loss": 18.0179,
+      "step": 5311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.484731874479814e-05,
+      "loss": 0.7247,
+      "step": 5312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.484126399374888e-05,
+      "loss": 0.0719,
+      "step": 5313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.483520824938318e-05,
+      "loss": 0.7445,
+      "step": 5314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.482915151187372e-05,
+      "loss": 0.7508,
+      "step": 5315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.482309378139317e-05,
+      "loss": 19.5798,
+      "step": 5316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.481703505811422e-05,
+      "loss": 21.5298,
+      "step": 5317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.481097534220962e-05,
+      "loss": 22.5304,
+      "step": 5318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.480491463385212e-05,
+      "loss": 0.6936,
+      "step": 5319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479885293321452e-05,
+      "loss": 0.7539,
+      "step": 5320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479279024046964e-05,
+      "loss": 21.5014,
+      "step": 5321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.478672655579031e-05,
+      "loss": 0.739,
+      "step": 5322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.478066187934941e-05,
+      "loss": 0.7907,
+      "step": 5323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477459621131984e-05,
+      "loss": 21.8871,
+      "step": 5324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.476852955187456e-05,
+      "loss": 20.7601,
+      "step": 5325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.476246190118649e-05,
+      "loss": 0.7872,
+      "step": 5326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.475639325942863e-05,
+      "loss": 21.2463,
+      "step": 5327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4750323626774e-05,
+      "loss": 0.071,
+      "step": 5328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.474425300339566e-05,
+      "loss": 19.7962,
+      "step": 5329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473818138946664e-05,
+      "loss": 22.1367,
+      "step": 5330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473210878516007e-05,
+      "loss": 20.951,
+      "step": 5331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.472603519064908e-05,
+      "loss": 20.5629,
+      "step": 5332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471996060610681e-05,
+      "loss": 20.0265,
+      "step": 5333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471388503170645e-05,
+      "loss": 0.7665,
+      "step": 5334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.470780846762123e-05,
+      "loss": 0.7817,
+      "step": 5335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.470173091402437e-05,
+      "loss": 0.6826,
+      "step": 5336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469565237108916e-05,
+      "loss": 0.7078,
+      "step": 5337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.468957283898888e-05,
+      "loss": 19.8891,
+      "step": 5338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.468349231789687e-05,
+      "loss": 19.6237,
+      "step": 5339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.467741080798647e-05,
+      "loss": 0.7394,
+      "step": 5340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.467132830943108e-05,
+      "loss": 0.0618,
+      "step": 5341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466524482240408e-05,
+      "loss": 0.8132,
+      "step": 5342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.465916034707893e-05,
+      "loss": 20.3566,
+      "step": 5343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.465307488362913e-05,
+      "loss": 0.7069,
+      "step": 5344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.464698843222809e-05,
+      "loss": 0.8242,
+      "step": 5345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.464090099304942e-05,
+      "loss": 21.7042,
+      "step": 5346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.463481256626664e-05,
+      "loss": 0.8285,
+      "step": 5347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46287231520533e-05,
+      "loss": 21.5012,
+      "step": 5348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.462263275058304e-05,
+      "loss": 0.7867,
+      "step": 5349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46165413620295e-05,
+      "loss": 21.5342,
+      "step": 5350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.461044898656634e-05,
+      "loss": 0.7617,
+      "step": 5351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.460435562436724e-05,
+      "loss": 22.5012,
+      "step": 5352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.45982612756059e-05,
+      "loss": 22.5165,
+      "step": 5353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.459216594045612e-05,
+      "loss": 0.7213,
+      "step": 5354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.458606961909165e-05,
+      "loss": 23.6796,
+      "step": 5355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.45799723116863e-05,
+      "loss": 0.7382,
+      "step": 5356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.457387401841388e-05,
+      "loss": 20.8984,
+      "step": 5357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456777473944829e-05,
+      "loss": 0.7216,
+      "step": 5358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456167447496338e-05,
+      "loss": 0.7458,
+      "step": 5359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.45555732251331e-05,
+      "loss": 19.8255,
+      "step": 5360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454947099013137e-05,
+      "loss": 0.7146,
+      "step": 5361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454336777013219e-05,
+      "loss": 21.2548,
+      "step": 5362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.453726356530953e-05,
+      "loss": 0.0567,
+      "step": 5363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.453115837583744e-05,
+      "loss": 0.8042,
+      "step": 5364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.452505220188998e-05,
+      "loss": 0.0734,
+      "step": 5365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.451894504364122e-05,
+      "loss": 19.5852,
+      "step": 5366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.451283690126528e-05,
+      "loss": 22.8013,
+      "step": 5367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.450672777493631e-05,
+      "loss": 19.9132,
+      "step": 5368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.450061766482846e-05,
+      "loss": 0.0801,
+      "step": 5369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.449450657111595e-05,
+      "loss": 20.024,
+      "step": 5370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4488394493973e-05,
+      "loss": 20.5229,
+      "step": 5371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.448228143357386e-05,
+      "loss": 0.6968,
+      "step": 5372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.447616739009283e-05,
+      "loss": 22.1371,
+      "step": 5373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.447005236370418e-05,
+      "loss": 19.3378,
+      "step": 5374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.446393635458227e-05,
+      "loss": 0.0523,
+      "step": 5375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44578193629015e-05,
+      "loss": 19.2769,
+      "step": 5376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.445170138883619e-05,
+      "loss": 22.5056,
+      "step": 5377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.444558243256082e-05,
+      "loss": 20.2973,
+      "step": 5378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.443946249424982e-05,
+      "loss": 0.7588,
+      "step": 5379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.443334157407768e-05,
+      "loss": 0.7666,
+      "step": 5380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442721967221889e-05,
+      "loss": 0.7696,
+      "step": 5381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442109678884798e-05,
+      "loss": 20.1604,
+      "step": 5382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.441497292413955e-05,
+      "loss": 20.8218,
+      "step": 5383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.440884807826814e-05,
+      "loss": 0.7529,
+      "step": 5384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44027222514084e-05,
+      "loss": 19.3278,
+      "step": 5385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.439659544373498e-05,
+      "loss": 0.7107,
+      "step": 5386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.439046765542252e-05,
+      "loss": 21.2744,
+      "step": 5387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.438433888664574e-05,
+      "loss": 21.0949,
+      "step": 5388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.437820913757939e-05,
+      "loss": 0.7026,
+      "step": 5389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.437207840839821e-05,
+      "loss": 21.2647,
+      "step": 5390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.436594669927698e-05,
+      "loss": 20.0489,
+      "step": 5391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.435981401039052e-05,
+      "loss": 22.7043,
+      "step": 5392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.435368034191366e-05,
+      "loss": 22.9077,
+      "step": 5393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.434754569402129e-05,
+      "loss": 0.7663,
+      "step": 5394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43414100668883e-05,
+      "loss": 0.7103,
+      "step": 5395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.433527346068961e-05,
+      "loss": 0.724,
+      "step": 5396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.432913587560018e-05,
+      "loss": 21.6701,
+      "step": 5397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.432299731179498e-05,
+      "loss": 20.9804,
+      "step": 5398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.431685776944904e-05,
+      "loss": 20.8506,
+      "step": 5399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.431071724873738e-05,
+      "loss": 21.5471,
+      "step": 5400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.430457574983505e-05,
+      "loss": 0.7762,
+      "step": 5401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42984332729172e-05,
+      "loss": 21.6845,
+      "step": 5402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.429228981815888e-05,
+      "loss": 20.9869,
+      "step": 5403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.428614538573528e-05,
+      "loss": 0.0584,
+      "step": 5404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.427999997582157e-05,
+      "loss": 20.7544,
+      "step": 5405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.427385358859293e-05,
+      "loss": 21.249,
+      "step": 5406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.426770622422465e-05,
+      "loss": 21.7003,
+      "step": 5407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.426155788289191e-05,
+      "loss": 20.2152,
+      "step": 5408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.425540856477008e-05,
+      "loss": 0.7296,
+      "step": 5409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42492582700344e-05,
+      "loss": 20.1605,
+      "step": 5410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.424310699886028e-05,
+      "loss": 21.0289,
+      "step": 5411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.423695475142305e-05,
+      "loss": 0.7514,
+      "step": 5412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.423080152789812e-05,
+      "loss": 20.9292,
+      "step": 5413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422464732846092e-05,
+      "loss": 0.6507,
+      "step": 5414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42184921532869e-05,
+      "loss": 0.7155,
+      "step": 5415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.421233600255153e-05,
+      "loss": 0.7132,
+      "step": 5416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.420617887643036e-05,
+      "loss": 0.7714,
+      "step": 5417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42000207750989e-05,
+      "loss": 20.7301,
+      "step": 5418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.419386169873271e-05,
+      "loss": 0.7613,
+      "step": 5419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41877016475074e-05,
+      "loss": 22.0046,
+      "step": 5420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418154062159858e-05,
+      "loss": 0.7249,
+      "step": 5421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41753786211819e-05,
+      "loss": 0.7254,
+      "step": 5422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.416921564643306e-05,
+      "loss": 0.7658,
+      "step": 5423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.416305169752772e-05,
+      "loss": 0.07,
+      "step": 5424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.415688677464166e-05,
+      "loss": 0.761,
+      "step": 5425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.415072087795062e-05,
+      "loss": 21.7287,
+      "step": 5426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.414455400763037e-05,
+      "loss": 0.735,
+      "step": 5427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.413838616385676e-05,
+      "loss": 21.5035,
+      "step": 5428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41322173468056e-05,
+      "loss": 20.8447,
+      "step": 5429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41260475566528e-05,
+      "loss": 0.7493,
+      "step": 5430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.411987679357423e-05,
+      "loss": 19.9447,
+      "step": 5431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.411370505774583e-05,
+      "loss": 0.7579,
+      "step": 5432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410753234934355e-05,
+      "loss": 19.2636,
+      "step": 5433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410135866854338e-05,
+      "loss": 20.786,
+      "step": 5434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40951840155213e-05,
+      "loss": 20.0898,
+      "step": 5435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40890083904534e-05,
+      "loss": 21.9295,
+      "step": 5436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40828317935157e-05,
+      "loss": 19.8952,
+      "step": 5437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.407665422488431e-05,
+      "loss": 21.5284,
+      "step": 5438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.407047568473536e-05,
+      "loss": 22.0209,
+      "step": 5439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406429617324499e-05,
+      "loss": 0.691,
+      "step": 5440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.405811569058937e-05,
+      "loss": 0.7368,
+      "step": 5441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.405193423694471e-05,
+      "loss": 19.6816,
+      "step": 5442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.404575181248724e-05,
+      "loss": 21.5206,
+      "step": 5443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.403956841739323e-05,
+      "loss": 0.0691,
+      "step": 5444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.403338405183896e-05,
+      "loss": 0.7359,
+      "step": 5445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.402719871600074e-05,
+      "loss": 20.2205,
+      "step": 5446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40210124100549e-05,
+      "loss": 21.3266,
+      "step": 5447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.401482513417786e-05,
+      "loss": 18.9826,
+      "step": 5448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.400863688854597e-05,
+      "loss": 20.3991,
+      "step": 5449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.400244767333567e-05,
+      "loss": 21.5577,
+      "step": 5450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.399625748872342e-05,
+      "loss": 0.6995,
+      "step": 5451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.399006633488569e-05,
+      "loss": 22.31,
+      "step": 5452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.398387421199899e-05,
+      "loss": 0.7684,
+      "step": 5453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.397768112023985e-05,
+      "loss": 20.5574,
+      "step": 5454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.397148705978485e-05,
+      "loss": 22.001,
+      "step": 5455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.396529203081056e-05,
+      "loss": 21.3337,
+      "step": 5456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.395909603349363e-05,
+      "loss": 19.8881,
+      "step": 5457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39528990680107e-05,
+      "loss": 0.0789,
+      "step": 5458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39467011345384e-05,
+      "loss": 0.7485,
+      "step": 5459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.394050223325348e-05,
+      "loss": 0.7533,
+      "step": 5460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.393430236433265e-05,
+      "loss": 0.7507,
+      "step": 5461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.392810152795267e-05,
+      "loss": 0.0511,
+      "step": 5462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.392189972429034e-05,
+      "loss": 0.7457,
+      "step": 5463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.391569695352244e-05,
+      "loss": 0.7158,
+      "step": 5464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.390949321582583e-05,
+      "loss": 20.1599,
+      "step": 5465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.390328851137736e-05,
+      "loss": 21.4621,
+      "step": 5466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.389708284035397e-05,
+      "loss": 0.7238,
+      "step": 5467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.389087620293253e-05,
+      "loss": 0.7569,
+      "step": 5468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.388466859929001e-05,
+      "loss": 20.0519,
+      "step": 5469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.387846002960337e-05,
+      "loss": 0.1063,
+      "step": 5470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.387225049404967e-05,
+      "loss": 0.7123,
+      "step": 5471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.386603999280589e-05,
+      "loss": 0.7848,
+      "step": 5472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38598285260491e-05,
+      "loss": 0.0696,
+      "step": 5473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38536160939564e-05,
+      "loss": 0.0819,
+      "step": 5474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.384740269670488e-05,
+      "loss": 0.7739,
+      "step": 5475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38411883344717e-05,
+      "loss": 0.6745,
+      "step": 5476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.383497300743405e-05,
+      "loss": 21.033,
+      "step": 5477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.382875671576909e-05,
+      "loss": 0.7378,
+      "step": 5478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.382253945965408e-05,
+      "loss": 20.6538,
+      "step": 5479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.381632123926624e-05,
+      "loss": 20.524,
+      "step": 5480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.381010205478286e-05,
+      "loss": 0.0742,
+      "step": 5481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.380388190638127e-05,
+      "loss": 20.6564,
+      "step": 5482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.379766079423878e-05,
+      "loss": 0.6865,
+      "step": 5483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.379143871853277e-05,
+      "loss": 0.0523,
+      "step": 5484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37852156794406e-05,
+      "loss": 0.0863,
+      "step": 5485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377899167713972e-05,
+      "loss": 21.2298,
+      "step": 5486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377276671180756e-05,
+      "loss": 0.6972,
+      "step": 5487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.376654078362159e-05,
+      "loss": 0.7582,
+      "step": 5488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.376031389275932e-05,
+      "loss": 21.0747,
+      "step": 5489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.375408603939827e-05,
+      "loss": 22.0353,
+      "step": 5490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.3747857223716e-05,
+      "loss": 21.6684,
+      "step": 5491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374162744589007e-05,
+      "loss": 0.0812,
+      "step": 5492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.373539670609812e-05,
+      "loss": 22.92,
+      "step": 5493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372916500451778e-05,
+      "loss": 21.5966,
+      "step": 5494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372293234132671e-05,
+      "loss": 0.7507,
+      "step": 5495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37166987167026e-05,
+      "loss": 0.8026,
+      "step": 5496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.371046413082317e-05,
+      "loss": 0.7292,
+      "step": 5497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.370422858386615e-05,
+      "loss": 22.8161,
+      "step": 5498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.369799207600935e-05,
+      "loss": 22.5446,
+      "step": 5499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.369175460743053e-05,
+      "loss": 0.7733,
+      "step": 5500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.368551617830757e-05,
+      "loss": 0.6945,
+      "step": 5501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.367927678881829e-05,
+      "loss": 0.7514,
+      "step": 5502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.367303643914057e-05,
+      "loss": 20.7817,
+      "step": 5503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.366679512945233e-05,
+      "loss": 0.8041,
+      "step": 5504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.366055285993152e-05,
+      "loss": 20.9167,
+      "step": 5505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.365430963075609e-05,
+      "loss": 20.3984,
+      "step": 5506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.364806544210404e-05,
+      "loss": 0.7726,
+      "step": 5507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36418202941534e-05,
+      "loss": 0.7483,
+      "step": 5508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36355741870822e-05,
+      "loss": 0.7961,
+      "step": 5509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.362932712106851e-05,
+      "loss": 22.7087,
+      "step": 5510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.362307909629045e-05,
+      "loss": 19.2038,
+      "step": 5511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.361683011292615e-05,
+      "loss": 19.2243,
+      "step": 5512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.361058017115375e-05,
+      "loss": 21.4423,
+      "step": 5513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.360432927115145e-05,
+      "loss": 0.738,
+      "step": 5514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.359807741309747e-05,
+      "loss": 20.0016,
+      "step": 5515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.359182459717003e-05,
+      "loss": 20.8889,
+      "step": 5516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.358557082354739e-05,
+      "loss": 21.4636,
+      "step": 5517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.357931609240787e-05,
+      "loss": 22.0091,
+      "step": 5518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.357306040392976e-05,
+      "loss": 0.0815,
+      "step": 5519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.356680375829144e-05,
+      "loss": 21.7246,
+      "step": 5520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.356054615567127e-05,
+      "loss": 0.7611,
+      "step": 5521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.355428759624765e-05,
+      "loss": 20.6169,
+      "step": 5522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354802808019902e-05,
+      "loss": 20.7907,
+      "step": 5523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354176760770382e-05,
+      "loss": 22.05,
+      "step": 5524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.353550617894057e-05,
+      "loss": 0.7915,
+      "step": 5525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.352924379408773e-05,
+      "loss": 20.3725,
+      "step": 5526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.352298045332388e-05,
+      "loss": 0.7361,
+      "step": 5527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.351671615682757e-05,
+      "loss": 21.7947,
+      "step": 5528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.351045090477739e-05,
+      "loss": 20.0586,
+      "step": 5529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.350418469735197e-05,
+      "loss": 0.7901,
+      "step": 5530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.349791753472997e-05,
+      "loss": 0.7145,
+      "step": 5531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.349164941709004e-05,
+      "loss": 0.7347,
+      "step": 5532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.348538034461091e-05,
+      "loss": 19.9435,
+      "step": 5533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.347911031747127e-05,
+      "loss": 0.7459,
+      "step": 5534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34728393358499e-05,
+      "loss": 22.9187,
+      "step": 5535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34665673999256e-05,
+      "loss": 19.4267,
+      "step": 5536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.346029450987717e-05,
+      "loss": 21.5054,
+      "step": 5537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.345402066588344e-05,
+      "loss": 21.3548,
+      "step": 5538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.344774586812328e-05,
+      "loss": 20.4503,
+      "step": 5539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.344147011677556e-05,
+      "loss": 0.7852,
+      "step": 5540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.343519341201926e-05,
+      "loss": 21.4962,
+      "step": 5541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.342891575403327e-05,
+      "loss": 0.7483,
+      "step": 5542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.342263714299661e-05,
+      "loss": 20.5368,
+      "step": 5543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.341635757908824e-05,
+      "loss": 20.7914,
+      "step": 5544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34100770624872e-05,
+      "loss": 20.1301,
+      "step": 5545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.340379559337256e-05,
+      "loss": 19.6719,
+      "step": 5546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.339751317192339e-05,
+      "loss": 0.0722,
+      "step": 5547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.339122979831882e-05,
+      "loss": 19.6517,
+      "step": 5548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.338494547273797e-05,
+      "loss": 21.1844,
+      "step": 5549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.337866019536001e-05,
+      "loss": 0.7354,
+      "step": 5550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.337237396636413e-05,
+      "loss": 0.7506,
+      "step": 5551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.336608678592955e-05,
+      "loss": 22.4136,
+      "step": 5552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33597986542355e-05,
+      "loss": 0.0653,
+      "step": 5553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335350957146128e-05,
+      "loss": 22.3358,
+      "step": 5554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33472195377862e-05,
+      "loss": 18.9231,
+      "step": 5555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.334092855338953e-05,
+      "loss": 0.733,
+      "step": 5556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.333463661845069e-05,
+      "loss": 0.7632,
+      "step": 5557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.332834373314901e-05,
+      "loss": 0.7259,
+      "step": 5558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.332204989766394e-05,
+      "loss": 20.3942,
+      "step": 5559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.331575511217488e-05,
+      "loss": 0.0746,
+      "step": 5560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.330945937686132e-05,
+      "loss": 20.2092,
+      "step": 5561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.330316269190274e-05,
+      "loss": 0.7055,
+      "step": 5562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.329686505747865e-05,
+      "loss": 21.457,
+      "step": 5563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32905664737686e-05,
+      "loss": 0.0533,
+      "step": 5564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.328426694095217e-05,
+      "loss": 20.8468,
+      "step": 5565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.327796645920895e-05,
+      "loss": 0.7068,
+      "step": 5566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.327166502871857e-05,
+      "loss": 18.334,
+      "step": 5567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.326536264966068e-05,
+      "loss": 20.4849,
+      "step": 5568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.325905932221494e-05,
+      "loss": 0.7402,
+      "step": 5569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.325275504656109e-05,
+      "loss": 0.7894,
+      "step": 5570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.324644982287886e-05,
+      "loss": 22.7751,
+      "step": 5571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.324014365134797e-05,
+      "loss": 19.1176,
+      "step": 5572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323383653214825e-05,
+      "loss": 0.0847,
+      "step": 5573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32275284654595e-05,
+      "loss": 22.1223,
+      "step": 5574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.322121945146155e-05,
+      "loss": 20.3868,
+      "step": 5575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321490949033427e-05,
+      "loss": 0.6804,
+      "step": 5576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32085985822576e-05,
+      "loss": 21.3166,
+      "step": 5577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320228672741142e-05,
+      "loss": 19.2902,
+      "step": 5578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.319597392597567e-05,
+      "loss": 22.0235,
+      "step": 5579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.318966017813035e-05,
+      "loss": 0.7385,
+      "step": 5580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.318334548405546e-05,
+      "loss": 20.7585,
+      "step": 5581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.317702984393103e-05,
+      "loss": 19.7734,
+      "step": 5582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31707132579371e-05,
+      "loss": 20.1068,
+      "step": 5583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.316439572625379e-05,
+      "loss": 20.1721,
+      "step": 5584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31580772490612e-05,
+      "loss": 0.6678,
+      "step": 5585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315175782653943e-05,
+      "loss": 0.0643,
+      "step": 5586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31454374588687e-05,
+      "loss": 21.2081,
+      "step": 5587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313911614622915e-05,
+      "loss": 19.9477,
+      "step": 5588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313279388880104e-05,
+      "loss": 0.7736,
+      "step": 5589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.312647068676462e-05,
+      "loss": 21.6097,
+      "step": 5590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.312014654030012e-05,
+      "loss": 21.3593,
+      "step": 5591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.311382144958786e-05,
+      "loss": 20.516,
+      "step": 5592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31074954148082e-05,
+      "loss": 0.0804,
+      "step": 5593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.310116843614145e-05,
+      "loss": 0.753,
+      "step": 5594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.309484051376802e-05,
+      "loss": 0.7661,
+      "step": 5595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.308851164786828e-05,
+      "loss": 18.8434,
+      "step": 5596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.308218183862272e-05,
+      "loss": 20.659,
+      "step": 5597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.307585108621175e-05,
+      "loss": 0.091,
+      "step": 5598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.306951939081589e-05,
+      "loss": 0.7375,
+      "step": 5599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.306318675261562e-05,
+      "loss": 21.0649,
+      "step": 5600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.305685317179153e-05,
+      "loss": 19.2155,
+      "step": 5601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.305051864852415e-05,
+      "loss": 17.5116,
+      "step": 5602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304418318299411e-05,
+      "loss": 0.084,
+      "step": 5603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.3037846775382e-05,
+      "loss": 0.7187,
+      "step": 5604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.303150942586848e-05,
+      "loss": 0.7459,
+      "step": 5605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.302517113463424e-05,
+      "loss": 23.1168,
+      "step": 5606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.301883190185996e-05,
+      "loss": 0.0785,
+      "step": 5607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.301249172772638e-05,
+      "loss": 0.7262,
+      "step": 5608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300615061241428e-05,
+      "loss": 0.7311,
+      "step": 5609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29998085561044e-05,
+      "loss": 0.7179,
+      "step": 5610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.299346555897758e-05,
+      "loss": 20.7495,
+      "step": 5611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.298712162121466e-05,
+      "loss": 21.2859,
+      "step": 5612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29807767429965e-05,
+      "loss": 18.9257,
+      "step": 5613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.297443092450396e-05,
+      "loss": 0.6982,
+      "step": 5614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.2968084165918e-05,
+      "loss": 19.1892,
+      "step": 5615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.296173646741954e-05,
+      "loss": 19.3491,
+      "step": 5616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.295538782918957e-05,
+      "loss": 0.6614,
+      "step": 5617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.294903825140909e-05,
+      "loss": 19.4183,
+      "step": 5618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.294268773425909e-05,
+      "loss": 18.102,
+      "step": 5619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.293633627792065e-05,
+      "loss": 0.7599,
+      "step": 5620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.292998388257486e-05,
+      "loss": 21.5371,
+      "step": 5621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29236305484028e-05,
+      "loss": 0.0651,
+      "step": 5622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.291727627558561e-05,
+      "loss": 18.9738,
+      "step": 5623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.291092106430445e-05,
+      "loss": 23.3212,
+      "step": 5624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29045649147405e-05,
+      "loss": 0.7813,
+      "step": 5625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.289820782707499e-05,
+      "loss": 19.7004,
+      "step": 5626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.289184980148914e-05,
+      "loss": 0.6586,
+      "step": 5627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.288549083816421e-05,
+      "loss": 0.7998,
+      "step": 5628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.287913093728153e-05,
+      "loss": 19.5333,
+      "step": 5629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.287277009902237e-05,
+      "loss": 0.6855,
+      "step": 5630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.286640832356812e-05,
+      "loss": 0.7515,
+      "step": 5631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.286004561110012e-05,
+      "loss": 0.7457,
+      "step": 5632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.285368196179979e-05,
+      "loss": 0.7564,
+      "step": 5633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.284731737584854e-05,
+      "loss": 0.7259,
+      "step": 5634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.284095185342784e-05,
+      "loss": 19.6378,
+      "step": 5635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.283458539471918e-05,
+      "loss": 18.1905,
+      "step": 5636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.282821799990402e-05,
+      "loss": 0.752,
+      "step": 5637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.282184966916394e-05,
+      "loss": 19.3051,
+      "step": 5638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.281548040268046e-05,
+      "loss": 19.9574,
+      "step": 5639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28091102006352e-05,
+      "loss": 0.7262,
+      "step": 5640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280273906320977e-05,
+      "loss": 0.0876,
+      "step": 5641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.279636699058577e-05,
+      "loss": 0.7646,
+      "step": 5642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.278999398294491e-05,
+      "loss": 20.531,
+      "step": 5643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.278362004046889e-05,
+      "loss": 18.8212,
+      "step": 5644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.277724516333937e-05,
+      "loss": 0.7682,
+      "step": 5645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.277086935173816e-05,
+      "loss": 20.5135,
+      "step": 5646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.2764492605847e-05,
+      "loss": 19.8478,
+      "step": 5647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.275811492584772e-05,
+      "loss": 20.0824,
+      "step": 5648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27517363119221e-05,
+      "loss": 0.7439,
+      "step": 5649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.274535676425202e-05,
+      "loss": 18.1798,
+      "step": 5650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.273897628301936e-05,
+      "loss": 21.0761,
+      "step": 5651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.2732594868406e-05,
+      "loss": 0.7728,
+      "step": 5652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272621252059392e-05,
+      "loss": 21.5533,
+      "step": 5653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.271982923976505e-05,
+      "loss": 18.7069,
+      "step": 5654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.271344502610137e-05,
+      "loss": 20.3632,
+      "step": 5655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270705987978491e-05,
+      "loss": 0.6438,
+      "step": 5656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270067380099768e-05,
+      "loss": 20.466,
+      "step": 5657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.26942867899218e-05,
+      "loss": 20.0865,
+      "step": 5658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268789884673931e-05,
+      "loss": 24.9154,
+      "step": 5659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268150997163235e-05,
+      "loss": 0.694,
+      "step": 5660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.267512016478306e-05,
+      "loss": 0.8101,
+      "step": 5661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.266872942637362e-05,
+      "loss": 20.1034,
+      "step": 5662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.266233775658622e-05,
+      "loss": 0.8283,
+      "step": 5663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.265594515560309e-05,
+      "loss": 0.7034,
+      "step": 5664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.264955162360648e-05,
+      "loss": 20.605,
+      "step": 5665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.264315716077867e-05,
+      "loss": 21.2795,
+      "step": 5666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.263676176730197e-05,
+      "loss": 19.7322,
+      "step": 5667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.263036544335869e-05,
+      "loss": 0.7121,
+      "step": 5668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.26239681891312e-05,
+      "loss": 0.7401,
+      "step": 5669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.26175700048019e-05,
+      "loss": 0.749,
+      "step": 5670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.261117089055319e-05,
+      "loss": 19.9464,
+      "step": 5671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.260477084656748e-05,
+      "loss": 0.7415,
+      "step": 5672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.259836987302726e-05,
+      "loss": 22.6396,
+      "step": 5673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.259196797011504e-05,
+      "loss": 0.092,
+      "step": 5674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.258556513801331e-05,
+      "loss": 20.7672,
+      "step": 5675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25791613769046e-05,
+      "loss": 0.7016,
+      "step": 5676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.257275668697153e-05,
+      "loss": 21.7705,
+      "step": 5677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.256635106839663e-05,
+      "loss": 0.7006,
+      "step": 5678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.255994452136258e-05,
+      "loss": 20.2807,
+      "step": 5679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2553537046052e-05,
+      "loss": 20.4054,
+      "step": 5680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.254712864264755e-05,
+      "loss": 0.0737,
+      "step": 5681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.254071931133198e-05,
+      "loss": 20.4255,
+      "step": 5682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.253430905228799e-05,
+      "loss": 22.3356,
+      "step": 5683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.252789786569834e-05,
+      "loss": 0.6583,
+      "step": 5684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.252148575174578e-05,
+      "loss": 19.7388,
+      "step": 5685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.251507271061318e-05,
+      "loss": 18.5096,
+      "step": 5686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.250865874248332e-05,
+      "loss": 0.0499,
+      "step": 5687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25022438475391e-05,
+      "loss": 20.1212,
+      "step": 5688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.249582802596339e-05,
+      "loss": 21.6481,
+      "step": 5689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.248941127793908e-05,
+      "loss": 18.9097,
+      "step": 5690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.248299360364913e-05,
+      "loss": 18.5504,
+      "step": 5691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.247657500327653e-05,
+      "loss": 21.7796,
+      "step": 5692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.247015547700425e-05,
+      "loss": 0.7834,
+      "step": 5693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.246373502501531e-05,
+      "loss": 0.7365,
+      "step": 5694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245731364749274e-05,
+      "loss": 21.4704,
+      "step": 5695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245089134461965e-05,
+      "loss": 0.7206,
+      "step": 5696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.244446811657911e-05,
+      "loss": 22.4343,
+      "step": 5697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.243804396355425e-05,
+      "loss": 0.7988,
+      "step": 5698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.243161888572823e-05,
+      "loss": 0.7048,
+      "step": 5699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.242519288328421e-05,
+      "loss": 0.0584,
+      "step": 5700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.241876595640541e-05,
+      "loss": 0.7438,
+      "step": 5701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.241233810527504e-05,
+      "loss": 0.7723,
+      "step": 5702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.240590933007637e-05,
+      "loss": 0.7443,
+      "step": 5703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.23994796309927e-05,
+      "loss": 0.7084,
+      "step": 5704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.239304900820732e-05,
+      "loss": 20.8306,
+      "step": 5705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.238661746190357e-05,
+      "loss": 0.8106,
+      "step": 5706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.238018499226481e-05,
+      "loss": 0.0506,
+      "step": 5707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237375159947444e-05,
+      "loss": 0.7623,
+      "step": 5708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.236731728371585e-05,
+      "loss": 20.8156,
+      "step": 5709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.236088204517249e-05,
+      "loss": 20.9676,
+      "step": 5710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.235444588402782e-05,
+      "loss": 0.7713,
+      "step": 5711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.234800880046538e-05,
+      "loss": 21.9323,
+      "step": 5712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.234157079466863e-05,
+      "loss": 0.7311,
+      "step": 5713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233513186682115e-05,
+      "loss": 0.708,
+      "step": 5714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.232869201710649e-05,
+      "loss": 20.6162,
+      "step": 5715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.232225124570828e-05,
+      "loss": 21.0939,
+      "step": 5716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.231580955281011e-05,
+      "loss": 0.0683,
+      "step": 5717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.230936693859567e-05,
+      "loss": 0.7185,
+      "step": 5718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.230292340324859e-05,
+      "loss": 20.8129,
+      "step": 5719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22964789469526e-05,
+      "loss": 0.7299,
+      "step": 5720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.229003356989144e-05,
+      "loss": 0.7142,
+      "step": 5721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.228358727224887e-05,
+      "loss": 0.7885,
+      "step": 5722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.227714005420864e-05,
+      "loss": 19.8216,
+      "step": 5723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.227069191595458e-05,
+      "loss": 0.8116,
+      "step": 5724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.226424285767052e-05,
+      "loss": 21.4657,
+      "step": 5725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.225779287954031e-05,
+      "loss": 0.7793,
+      "step": 5726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.225134198174786e-05,
+      "loss": 19.7941,
+      "step": 5727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.224489016447708e-05,
+      "loss": 21.2266,
+      "step": 5728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.223843742791188e-05,
+      "loss": 0.741,
+      "step": 5729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.223198377223629e-05,
+      "loss": 20.3979,
+      "step": 5730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.222552919763422e-05,
+      "loss": 20.4279,
+      "step": 5731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.221907370428976e-05,
+      "loss": 20.0497,
+      "step": 5732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22126172923869e-05,
+      "loss": 0.7479,
+      "step": 5733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.220615996210975e-05,
+      "loss": 20.8482,
+      "step": 5734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21997017136424e-05,
+      "loss": 0.7466,
+      "step": 5735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219324254716896e-05,
+      "loss": 19.9847,
+      "step": 5736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.218678246287356e-05,
+      "loss": 19.0424,
+      "step": 5737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.218032146094043e-05,
+      "loss": 20.0551,
+      "step": 5738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.217385954155374e-05,
+      "loss": 20.3785,
+      "step": 5739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21673967048977e-05,
+      "loss": 21.5094,
+      "step": 5740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.216093295115661e-05,
+      "loss": 20.6462,
+      "step": 5741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21544682805147e-05,
+      "loss": 18.8496,
+      "step": 5742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.214800269315633e-05,
+      "loss": 22.625,
+      "step": 5743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.214153618926577e-05,
+      "loss": 0.8275,
+      "step": 5744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.213506876902742e-05,
+      "loss": 21.7267,
+      "step": 5745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.212860043262567e-05,
+      "loss": 0.6844,
+      "step": 5746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21221311802449e-05,
+      "loss": 0.0796,
+      "step": 5747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.211566101206958e-05,
+      "loss": 21.0893,
+      "step": 5748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.210918992828414e-05,
+      "loss": 0.0783,
+      "step": 5749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.210271792907309e-05,
+      "loss": 0.7138,
+      "step": 5750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.209624501462095e-05,
+      "loss": 21.884,
+      "step": 5751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.208977118511225e-05,
+      "loss": 19.5709,
+      "step": 5752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.208329644073155e-05,
+      "loss": 19.5579,
+      "step": 5753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.207682078166345e-05,
+      "loss": 0.7417,
+      "step": 5754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.207034420809259e-05,
+      "loss": 18.1705,
+      "step": 5755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.206386672020359e-05,
+      "loss": 19.8952,
+      "step": 5756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205738831818112e-05,
+      "loss": 18.6321,
+      "step": 5757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.20509090022099e-05,
+      "loss": 0.7286,
+      "step": 5758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.204442877247464e-05,
+      "loss": 19.5934,
+      "step": 5759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.20379476291601e-05,
+      "loss": 0.0889,
+      "step": 5760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.203146557245102e-05,
+      "loss": 0.7375,
+      "step": 5761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.202498260253226e-05,
+      "loss": 0.7496,
+      "step": 5762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.20184987195886e-05,
+      "loss": 0.7199,
+      "step": 5763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.201201392380493e-05,
+      "loss": 19.0381,
+      "step": 5764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.200552821536608e-05,
+      "loss": 20.5906,
+      "step": 5765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.199904159445702e-05,
+      "loss": 19.1345,
+      "step": 5766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.199255406126263e-05,
+      "loss": 21.2132,
+      "step": 5767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.198606561596788e-05,
+      "loss": 19.6066,
+      "step": 5768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197957625875779e-05,
+      "loss": 0.7281,
+      "step": 5769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19730859898173e-05,
+      "loss": 20.1278,
+      "step": 5770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.196659480933153e-05,
+      "loss": 20.2463,
+      "step": 5771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.196010271748546e-05,
+      "loss": 18.6283,
+      "step": 5772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.195360971446424e-05,
+      "loss": 19.1442,
+      "step": 5773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.194711580045297e-05,
+      "loss": 19.4589,
+      "step": 5774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.194062097563677e-05,
+      "loss": 19.2393,
+      "step": 5775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.193412524020082e-05,
+      "loss": 0.0799,
+      "step": 5776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19276285943303e-05,
+      "loss": 0.7227,
+      "step": 5777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.192113103821046e-05,
+      "loss": 0.779,
+      "step": 5778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.191463257202651e-05,
+      "loss": 0.7758,
+      "step": 5779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.190813319596373e-05,
+      "loss": 18.1951,
+      "step": 5780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.190163291020741e-05,
+      "loss": 0.811,
+      "step": 5781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189513171494288e-05,
+      "loss": 0.7342,
+      "step": 5782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.188862961035551e-05,
+      "loss": 23.4055,
+      "step": 5783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.188212659663061e-05,
+      "loss": 21.7482,
+      "step": 5784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.187562267395364e-05,
+      "loss": 19.5718,
+      "step": 5785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.186911784251001e-05,
+      "loss": 19.0811,
+      "step": 5786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.186261210248516e-05,
+      "loss": 0.7446,
+      "step": 5787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.185610545406459e-05,
+      "loss": 0.0693,
+      "step": 5788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184959789743375e-05,
+      "loss": 0.7408,
+      "step": 5789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184308943277822e-05,
+      "loss": 21.6295,
+      "step": 5790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.183658006028355e-05,
+      "loss": 20.8055,
+      "step": 5791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18300697801353e-05,
+      "loss": 0.0775,
+      "step": 5792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.182355859251909e-05,
+      "loss": 0.7378,
+      "step": 5793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.181704649762054e-05,
+      "loss": 19.3034,
+      "step": 5794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.181053349562533e-05,
+      "loss": 21.881,
+      "step": 5795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.180401958671913e-05,
+      "loss": 21.4194,
+      "step": 5796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179750477108766e-05,
+      "loss": 0.6975,
+      "step": 5797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179098904891663e-05,
+      "loss": 18.5126,
+      "step": 5798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.178447242039182e-05,
+      "loss": 0.7385,
+      "step": 5799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.177795488569904e-05,
+      "loss": 0.0621,
+      "step": 5800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.177143644502406e-05,
+      "loss": 21.8598,
+      "step": 5801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.176491709855274e-05,
+      "loss": 0.7715,
+      "step": 5802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.175839684647095e-05,
+      "loss": 21.5954,
+      "step": 5803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.175187568896459e-05,
+      "loss": 0.6965,
+      "step": 5804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.174535362621953e-05,
+      "loss": 19.0123,
+      "step": 5805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.173883065842176e-05,
+      "loss": 0.672,
+      "step": 5806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.173230678575725e-05,
+      "loss": 21.3108,
+      "step": 5807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.172578200841197e-05,
+      "loss": 18.7576,
+      "step": 5808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.171925632657193e-05,
+      "loss": 18.6293,
+      "step": 5809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17127297404232e-05,
+      "loss": 20.9926,
+      "step": 5810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170620225015184e-05,
+      "loss": 0.7063,
+      "step": 5811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.169967385594396e-05,
+      "loss": 22.2428,
+      "step": 5812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.169314455798569e-05,
+      "loss": 22.9947,
+      "step": 5813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.168661435646314e-05,
+      "loss": 17.5861,
+      "step": 5814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.168008325156251e-05,
+      "loss": 20.4317,
+      "step": 5815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.167355124347001e-05,
+      "loss": 0.7549,
+      "step": 5816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.166701833237185e-05,
+      "loss": 21.949,
+      "step": 5817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.166048451845429e-05,
+      "loss": 20.49,
+      "step": 5818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.165394980190359e-05,
+      "loss": 0.7324,
+      "step": 5819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.164741418290607e-05,
+      "loss": 0.0843,
+      "step": 5820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.164087766164808e-05,
+      "loss": 22.4068,
+      "step": 5821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.163434023831593e-05,
+      "loss": 20.9157,
+      "step": 5822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.162780191309601e-05,
+      "loss": 17.8538,
+      "step": 5823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.162126268617476e-05,
+      "loss": 20.0206,
+      "step": 5824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.161472255773857e-05,
+      "loss": 20.9512,
+      "step": 5825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.160818152797391e-05,
+      "loss": 20.1346,
+      "step": 5826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16016395970673e-05,
+      "loss": 0.7998,
+      "step": 5827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15950967652052e-05,
+      "loss": 22.7481,
+      "step": 5828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.158855303257418e-05,
+      "loss": 19.8428,
+      "step": 5829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.158200839936076e-05,
+      "loss": 19.5912,
+      "step": 5830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157546286575156e-05,
+      "loss": 19.5358,
+      "step": 5831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.156891643193319e-05,
+      "loss": 20.2953,
+      "step": 5832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.156236909809228e-05,
+      "loss": 0.075,
+      "step": 5833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.155582086441546e-05,
+      "loss": 0.811,
+      "step": 5834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.154927173108948e-05,
+      "loss": 20.8693,
+      "step": 5835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.154272169830101e-05,
+      "loss": 0.8116,
+      "step": 5836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.153617076623683e-05,
+      "loss": 0.7408,
+      "step": 5837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.152961893508366e-05,
+      "loss": 21.2418,
+      "step": 5838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.152306620502832e-05,
+      "loss": 21.7449,
+      "step": 5839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15165125762576e-05,
+      "loss": 0.6774,
+      "step": 5840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.150995804895838e-05,
+      "loss": 20.3444,
+      "step": 5841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15034026233175e-05,
+      "loss": 21.9965,
+      "step": 5842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149684629952186e-05,
+      "loss": 0.0745,
+      "step": 5843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149028907775837e-05,
+      "loss": 21.2729,
+      "step": 5844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.148373095821402e-05,
+      "loss": 20.1893,
+      "step": 5845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.147717194107572e-05,
+      "loss": 0.0554,
+      "step": 5846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.147061202653051e-05,
+      "loss": 20.1954,
+      "step": 5847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.146405121476537e-05,
+      "loss": 19.5334,
+      "step": 5848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.145748950596739e-05,
+      "loss": 20.1955,
+      "step": 5849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.145092690032363e-05,
+      "loss": 23.6689,
+      "step": 5850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.144436339802118e-05,
+      "loss": 0.779,
+      "step": 5851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.143779899924716e-05,
+      "loss": 0.73,
+      "step": 5852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.143123370418872e-05,
+      "loss": 0.6673,
+      "step": 5853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.142466751303306e-05,
+      "loss": 0.7313,
+      "step": 5854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.141810042596735e-05,
+      "loss": 0.7066,
+      "step": 5855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.141153244317883e-05,
+      "loss": 0.0756,
+      "step": 5856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.140496356485476e-05,
+      "loss": 0.7297,
+      "step": 5857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13983937911824e-05,
+      "loss": 0.7083,
+      "step": 5858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.139182312234906e-05,
+      "loss": 22.2889,
+      "step": 5859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138525155854206e-05,
+      "loss": 21.0067,
+      "step": 5860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.137867909994877e-05,
+      "loss": 20.4315,
+      "step": 5861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.137210574675656e-05,
+      "loss": 0.737,
+      "step": 5862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.136553149915284e-05,
+      "loss": 20.802,
+      "step": 5863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.135895635732503e-05,
+      "loss": 19.6507,
+      "step": 5864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13523803214606e-05,
+      "loss": 19.7733,
+      "step": 5865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.134580339174701e-05,
+      "loss": 21.1574,
+      "step": 5866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.133922556837179e-05,
+      "loss": 0.7768,
+      "step": 5867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.133264685152246e-05,
+      "loss": 0.0702,
+      "step": 5868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.132606724138659e-05,
+      "loss": 20.6831,
+      "step": 5869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.131948673815174e-05,
+      "loss": 0.7599,
+      "step": 5870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.131290534200553e-05,
+      "loss": 19.0696,
+      "step": 5871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13063230531356e-05,
+      "loss": 20.5228,
+      "step": 5872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12997398717296e-05,
+      "loss": 0.0506,
+      "step": 5873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.129315579797522e-05,
+      "loss": 20.1123,
+      "step": 5874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.128657083206019e-05,
+      "loss": 21.1794,
+      "step": 5875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.127998497417222e-05,
+      "loss": 0.7178,
+      "step": 5876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.127339822449909e-05,
+      "loss": 0.7973,
+      "step": 5877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.126681058322855e-05,
+      "loss": 0.7672,
+      "step": 5878
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.126022205054844e-05,
+      "loss": 0.7454,
+      "step": 5879
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12536326266466e-05,
+      "loss": 19.6749,
+      "step": 5880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.124704231171089e-05,
+      "loss": 20.1071,
+      "step": 5881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12404511059292e-05,
+      "loss": 20.7782,
+      "step": 5882
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.123385900948942e-05,
+      "loss": 0.7499,
+      "step": 5883
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.122726602257952e-05,
+      "loss": 0.7805,
+      "step": 5884
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.122067214538747e-05,
+      "loss": 0.6492,
+      "step": 5885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12140773781012e-05,
+      "loss": 24.4638,
+      "step": 5886
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.120748172090882e-05,
+      "loss": 20.5435,
+      "step": 5887
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.120088517399829e-05,
+      "loss": 19.6015,
+      "step": 5888
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.119428773755771e-05,
+      "loss": 19.2983,
+      "step": 5889
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.118768941177517e-05,
+      "loss": 0.7451,
+      "step": 5890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.118109019683878e-05,
+      "loss": 0.7085,
+      "step": 5891
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.117449009293668e-05,
+      "loss": 0.7367,
+      "step": 5892
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.116788910025706e-05,
+      "loss": 0.678,
+      "step": 5893
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.116128721898809e-05,
+      "loss": 0.7268,
+      "step": 5894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.115468444931797e-05,
+      "loss": 22.4665,
+      "step": 5895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.114808079143499e-05,
+      "loss": 21.6282,
+      "step": 5896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.114147624552739e-05,
+      "loss": 0.7836,
+      "step": 5897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.113487081178347e-05,
+      "loss": 21.4477,
+      "step": 5898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.112826449039154e-05,
+      "loss": 0.7346,
+      "step": 5899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.112165728153997e-05,
+      "loss": 22.075,
+      "step": 5900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.111504918541712e-05,
+      "loss": 0.7315,
+      "step": 5901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110844020221136e-05,
+      "loss": 22.5742,
+      "step": 5902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110183033211114e-05,
+      "loss": 0.7135,
+      "step": 5903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10952195753049e-05,
+      "loss": 22.6392,
+      "step": 5904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10886079319811e-05,
+      "loss": 0.759,
+      "step": 5905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.108199540232825e-05,
+      "loss": 20.7233,
+      "step": 5906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.107538198653486e-05,
+      "loss": 21.7564,
+      "step": 5907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.106876768478947e-05,
+      "loss": 19.97,
+      "step": 5908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.106215249728069e-05,
+      "loss": 22.6345,
+      "step": 5909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.105553642419707e-05,
+      "loss": 20.5229,
+      "step": 5910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.104891946572726e-05,
+      "loss": 20.1001,
+      "step": 5911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10423016220599e-05,
+      "loss": 0.0785,
+      "step": 5912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.103568289338369e-05,
+      "loss": 0.0719,
+      "step": 5913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.102906327988727e-05,
+      "loss": 0.697,
+      "step": 5914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.102244278175942e-05,
+      "loss": 22.0944,
+      "step": 5915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101582139918887e-05,
+      "loss": 21.2041,
+      "step": 5916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.100919913236438e-05,
+      "loss": 22.5479,
+      "step": 5917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.100257598147475e-05,
+      "loss": 19.7733,
+      "step": 5918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.099595194670883e-05,
+      "loss": 21.516,
+      "step": 5919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.098932702825545e-05,
+      "loss": 19.6309,
+      "step": 5920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09827012263035e-05,
+      "loss": 0.0637,
+      "step": 5921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.097607454104184e-05,
+      "loss": 17.7598,
+      "step": 5922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.096944697265945e-05,
+      "loss": 0.7163,
+      "step": 5923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.096281852134524e-05,
+      "loss": 0.0928,
+      "step": 5924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09561891872882e-05,
+      "loss": 20.9613,
+      "step": 5925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.094955897067735e-05,
+      "loss": 0.0662,
+      "step": 5926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.094292787170169e-05,
+      "loss": 0.6488,
+      "step": 5927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093629589055027e-05,
+      "loss": 20.459,
+      "step": 5928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.092966302741219e-05,
+      "loss": 0.7056,
+      "step": 5929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.092302928247652e-05,
+      "loss": 17.5025,
+      "step": 5930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.091639465593241e-05,
+      "loss": 20.5396,
+      "step": 5931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.090975914796901e-05,
+      "loss": 0.771,
+      "step": 5932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.090312275877548e-05,
+      "loss": 21.0336,
+      "step": 5933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089648548854104e-05,
+      "loss": 19.1336,
+      "step": 5934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.088984733745491e-05,
+      "loss": 0.0715,
+      "step": 5935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.088320830570635e-05,
+      "loss": 21.6923,
+      "step": 5936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087656839348461e-05,
+      "loss": 21.5293,
+      "step": 5937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.086992760097902e-05,
+      "loss": 0.8208,
+      "step": 5938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08632859283789e-05,
+      "loss": 0.7375,
+      "step": 5939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.085664337587361e-05,
+      "loss": 21.1509,
+      "step": 5940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.084999994365252e-05,
+      "loss": 0.7154,
+      "step": 5941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.084335563190501e-05,
+      "loss": 20.5935,
+      "step": 5942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.083671044082055e-05,
+      "loss": 19.1515,
+      "step": 5943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.083006437058858e-05,
+      "loss": 20.9729,
+      "step": 5944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.082341742139856e-05,
+      "loss": 21.6606,
+      "step": 5945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.081676959344e-05,
+      "loss": 0.7358,
+      "step": 5946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.081012088690242e-05,
+      "loss": 0.6945,
+      "step": 5947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.080347130197541e-05,
+      "loss": 0.7802,
+      "step": 5948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.079682083884852e-05,
+      "loss": 0.7345,
+      "step": 5949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.079016949771135e-05,
+      "loss": 19.7768,
+      "step": 5950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.078351727875352e-05,
+      "loss": 19.8653,
+      "step": 5951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.077686418216473e-05,
+      "loss": 21.3034,
+      "step": 5952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07702102081346e-05,
+      "loss": 0.0677,
+      "step": 5953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.076355535685286e-05,
+      "loss": 20.8046,
+      "step": 5954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.075689962850926e-05,
+      "loss": 0.0593,
+      "step": 5955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07502430232935e-05,
+      "loss": 0.6572,
+      "step": 5956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074358554139542e-05,
+      "loss": 19.894,
+      "step": 5957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.073692718300478e-05,
+      "loss": 19.73,
+      "step": 5958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.073026794831141e-05,
+      "loss": 18.1514,
+      "step": 5959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.072360783750519e-05,
+      "loss": 0.7349,
+      "step": 5960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.071694685077597e-05,
+      "loss": 0.742,
+      "step": 5961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.071028498831367e-05,
+      "loss": 0.7533,
+      "step": 5962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.070362225030821e-05,
+      "loss": 0.0893,
+      "step": 5963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.069695863694955e-05,
+      "loss": 0.7336,
+      "step": 5964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.069029414842768e-05,
+      "loss": 21.6721,
+      "step": 5965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.068362878493257e-05,
+      "loss": 20.8562,
+      "step": 5966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.067696254665426e-05,
+      "loss": 20.2225,
+      "step": 5967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.067029543378281e-05,
+      "loss": 0.7405,
+      "step": 5968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.066362744650831e-05,
+      "loss": 20.3064,
+      "step": 5969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065695858502084e-05,
+      "loss": 19.3137,
+      "step": 5970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065028884951054e-05,
+      "loss": 0.6944,
+      "step": 5971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.064361824016756e-05,
+      "loss": 0.0621,
+      "step": 5972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.063694675718207e-05,
+      "loss": 20.1915,
+      "step": 5973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.063027440074429e-05,
+      "loss": 20.2352,
+      "step": 5974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.062360117104444e-05,
+      "loss": 0.7135,
+      "step": 5975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.061692706827276e-05,
+      "loss": 19.4407,
+      "step": 5976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.061025209261954e-05,
+      "loss": 20.067,
+      "step": 5977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.060357624427508e-05,
+      "loss": 0.7261,
+      "step": 5978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05968995234297e-05,
+      "loss": 0.7612,
+      "step": 5979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.059022193027377e-05,
+      "loss": 0.7237,
+      "step": 5980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.058354346499764e-05,
+      "loss": 0.0948,
+      "step": 5981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.057686412779172e-05,
+      "loss": 20.8633,
+      "step": 5982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.057018391884646e-05,
+      "loss": 18.9478,
+      "step": 5983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.056350283835228e-05,
+      "loss": 19.2309,
+      "step": 5984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.055682088649968e-05,
+      "loss": 0.6855,
+      "step": 5985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.055013806347913e-05,
+      "loss": 19.538,
+      "step": 5986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.054345436948119e-05,
+      "loss": 20.2244,
+      "step": 5987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.053676980469639e-05,
+      "loss": 20.3223,
+      "step": 5988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.053008436931532e-05,
+      "loss": 20.9583,
+      "step": 5989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052339806352855e-05,
+      "loss": 0.743,
+      "step": 5990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.051671088752676e-05,
+      "loss": 21.2017,
+      "step": 5991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.051002284150054e-05,
+      "loss": 0.7902,
+      "step": 5992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.050333392564061e-05,
+      "loss": 22.7407,
+      "step": 5993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.049664414013764e-05,
+      "loss": 0.7633,
+      "step": 5994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048995348518235e-05,
+      "loss": 0.7934,
+      "step": 5995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048326196096554e-05,
+      "loss": 0.7182,
+      "step": 5996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.047656956767793e-05,
+      "loss": 0.6941,
+      "step": 5997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046987630551032e-05,
+      "loss": 0.7908,
+      "step": 5998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046318217465357e-05,
+      "loss": 20.2049,
+      "step": 5999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04564871752985e-05,
+      "loss": 19.3474,
+      "step": 6000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.0449791307636e-05,
+      "loss": 0.0826,
+      "step": 6001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.044309457185695e-05,
+      "loss": 0.7506,
+      "step": 6002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043639696815228e-05,
+      "loss": 0.7359,
+      "step": 6003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.042969849671294e-05,
+      "loss": 19.522,
+      "step": 6004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.042299915772991e-05,
+      "loss": 0.742,
+      "step": 6005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.041629895139417e-05,
+      "loss": 0.6882,
+      "step": 6006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.040959787789674e-05,
+      "loss": 21.2208,
+      "step": 6007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.040289593742868e-05,
+      "loss": 20.573,
+      "step": 6008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.039619313018106e-05,
+      "loss": 0.6957,
+      "step": 6009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.038948945634495e-05,
+      "loss": 20.5352,
+      "step": 6010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.038278491611151e-05,
+      "loss": 0.6932,
+      "step": 6011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.037607950967185e-05,
+      "loss": 0.7359,
+      "step": 6012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.036937323721718e-05,
+      "loss": 0.0961,
+      "step": 6013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.036266609893863e-05,
+      "loss": 0.7456,
+      "step": 6014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03559580950275e-05,
+      "loss": 0.7283,
+      "step": 6015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.034924922567497e-05,
+      "loss": 0.712,
+      "step": 6016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.034253949107232e-05,
+      "loss": 18.5576,
+      "step": 6017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.033582889141087e-05,
+      "loss": 0.7105,
+      "step": 6018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.032911742688192e-05,
+      "loss": 19.1491,
+      "step": 6019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03224050976768e-05,
+      "loss": 21.8547,
+      "step": 6020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.031569190398691e-05,
+      "loss": 21.5402,
+      "step": 6021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.030897784600362e-05,
+      "loss": 21.5927,
+      "step": 6022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.030226292391832e-05,
+      "loss": 0.0825,
+      "step": 6023
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.029554713792251e-05,
+      "loss": 0.7356,
+      "step": 6024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.028883048820761e-05,
+      "loss": 19.3333,
+      "step": 6025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.028211297496513e-05,
+      "loss": 20.2801,
+      "step": 6026
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.027539459838657e-05,
+      "loss": 20.0423,
+      "step": 6027
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.026867535866348e-05,
+      "loss": 0.747,
+      "step": 6028
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.026195525598743e-05,
+      "loss": 19.3377,
+      "step": 6029
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.025523429055e-05,
+      "loss": 24.1742,
+      "step": 6030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.024851246254279e-05,
+      "loss": 0.7288,
+      "step": 6031
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.024178977215747e-05,
+      "loss": 20.625,
+      "step": 6032
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.023506621958568e-05,
+      "loss": 19.3367,
+      "step": 6033
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022834180501911e-05,
+      "loss": 0.0632,
+      "step": 6034
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022161652864947e-05,
+      "loss": 19.9035,
+      "step": 6035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.021489039066849e-05,
+      "loss": 20.9001,
+      "step": 6036
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.020816339126795e-05,
+      "loss": 21.0851,
+      "step": 6037
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.020143553063963e-05,
+      "loss": 0.7208,
+      "step": 6038
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.019470680897533e-05,
+      "loss": 0.7471,
+      "step": 6039
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.01879772264669e-05,
+      "loss": 21.1135,
+      "step": 6040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.018124678330616e-05,
+      "loss": 19.9089,
+      "step": 6041
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.017451547968504e-05,
+      "loss": 21.1751,
+      "step": 6042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.016778331579541e-05,
+      "loss": 18.0645,
+      "step": 6043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.016105029182925e-05,
+      "loss": 18.9224,
+      "step": 6044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.015431640797848e-05,
+      "loss": 20.2657,
+      "step": 6045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.014758166443507e-05,
+      "loss": 20.4715,
+      "step": 6046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.014084606139107e-05,
+      "loss": 0.7112,
+      "step": 6047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.013410959903848e-05,
+      "loss": 21.4753,
+      "step": 6048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.012737227756935e-05,
+      "loss": 17.9555,
+      "step": 6049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.012063409717578e-05,
+      "loss": 0.7138,
+      "step": 6050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.011389505804988e-05,
+      "loss": 21.5376,
+      "step": 6051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010715516038374e-05,
+      "loss": 0.7783,
+      "step": 6052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010041440436953e-05,
+      "loss": 0.7751,
+      "step": 6053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.009367279019945e-05,
+      "loss": 22.0067,
+      "step": 6054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.008693031806569e-05,
+      "loss": 0.0842,
+      "step": 6055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.008018698816046e-05,
+      "loss": 0.0699,
+      "step": 6056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.007344280067603e-05,
+      "loss": 19.3375,
+      "step": 6057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.006669775580465e-05,
+      "loss": 20.0647,
+      "step": 6058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.005995185373867e-05,
+      "loss": 19.4529,
+      "step": 6059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.005320509467035e-05,
+      "loss": 21.9297,
+      "step": 6060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00464574787921e-05,
+      "loss": 20.6123,
+      "step": 6061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.003970900629623e-05,
+      "loss": 19.7509,
+      "step": 6062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.003295967737519e-05,
+      "loss": 20.8919,
+      "step": 6063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002620949222138e-05,
+      "loss": 0.0855,
+      "step": 6064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.001945845102723e-05,
+      "loss": 0.0834,
+      "step": 6065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.001270655398525e-05,
+      "loss": 0.7555,
+      "step": 6066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.000595380128791e-05,
+      "loss": 0.7282,
+      "step": 6067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999920019312775e-05,
+      "loss": 0.7563,
+      "step": 6068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999244572969727e-05,
+      "loss": 19.8721,
+      "step": 6069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.998569041118909e-05,
+      "loss": 0.7189,
+      "step": 6070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.997893423779575e-05,
+      "loss": 0.7562,
+      "step": 6071
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99721772097099e-05,
+      "loss": 19.3282,
+      "step": 6072
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99654193271242e-05,
+      "loss": 19.6499,
+      "step": 6073
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.995866059023129e-05,
+      "loss": 20.0466,
+      "step": 6074
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.995190099922385e-05,
+      "loss": 22.2407,
+      "step": 6075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.994514055429461e-05,
+      "loss": 0.7477,
+      "step": 6076
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.993837925563629e-05,
+      "loss": 19.5043,
+      "step": 6077
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.993161710344167e-05,
+      "loss": 0.0586,
+      "step": 6078
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.992485409790356e-05,
+      "loss": 19.2247,
+      "step": 6079
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.991809023921474e-05,
+      "loss": 0.7674,
+      "step": 6080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.991132552756804e-05,
+      "loss": 0.0556,
+      "step": 6081
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.990455996315633e-05,
+      "loss": 0.0877,
+      "step": 6082
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.989779354617252e-05,
+      "loss": 20.3713,
+      "step": 6083
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.989102627680949e-05,
+      "loss": 0.674,
+      "step": 6084
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.988425815526017e-05,
+      "loss": 0.7973,
+      "step": 6085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.987748918171753e-05,
+      "loss": 19.4691,
+      "step": 6086
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.987071935637456e-05,
+      "loss": 19.7108,
+      "step": 6087
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.986394867942426e-05,
+      "loss": 0.7168,
+      "step": 6088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.985717715105964e-05,
+      "loss": 18.8377,
+      "step": 6089
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.98504047714738e-05,
+      "loss": 0.7175,
+      "step": 6090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.984363154085977e-05,
+      "loss": 17.808,
+      "step": 6091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.983685745941067e-05,
+      "loss": 21.8511,
+      "step": 6092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.983008252731964e-05,
+      "loss": 0.0823,
+      "step": 6093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.982330674477983e-05,
+      "loss": 0.741,
+      "step": 6094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.98165301119844e-05,
+      "loss": 22.0312,
+      "step": 6095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.980975262912655e-05,
+      "loss": 20.1608,
+      "step": 6096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.980297429639953e-05,
+      "loss": 0.062,
+      "step": 6097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.979619511399657e-05,
+      "loss": 19.2342,
+      "step": 6098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.978941508211093e-05,
+      "loss": 21.0835,
+      "step": 6099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.978263420093595e-05,
+      "loss": 21.5241,
+      "step": 6100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.977585247066489e-05,
+      "loss": 21.6809,
+      "step": 6101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.976906989149116e-05,
+      "loss": 0.0776,
+      "step": 6102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.976228646360808e-05,
+      "loss": 19.7572,
+      "step": 6103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.975550218720908e-05,
+      "loss": 17.3822,
+      "step": 6104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.974871706248755e-05,
+      "loss": 19.2927,
+      "step": 6105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.974193108963692e-05,
+      "loss": 22.0757,
+      "step": 6106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973514426885071e-05,
+      "loss": 0.802,
+      "step": 6107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.972835660032238e-05,
+      "loss": 19.5281,
+      "step": 6108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.972156808424545e-05,
+      "loss": 18.0352,
+      "step": 6109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.971477872081345e-05,
+      "loss": 20.3891,
+      "step": 6110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.970798851021994e-05,
+      "loss": 20.1485,
+      "step": 6111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.970119745265851e-05,
+      "loss": 0.7148,
+      "step": 6112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96944055483228e-05,
+      "loss": 18.0356,
+      "step": 6113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.968761279740642e-05,
+      "loss": 0.7976,
+      "step": 6114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.968081920010301e-05,
+      "loss": 20.1828,
+      "step": 6115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96740247566063e-05,
+      "loss": 18.908,
+      "step": 6116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.966722946710997e-05,
+      "loss": 19.2579,
+      "step": 6117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.966043333180773e-05,
+      "loss": 0.7006,
+      "step": 6118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.965363635089338e-05,
+      "loss": 0.7291,
+      "step": 6119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96468385245607e-05,
+      "loss": 0.7331,
+      "step": 6120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.964003985300346e-05,
+      "loss": 0.7326,
+      "step": 6121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.963324033641551e-05,
+      "loss": 21.8307,
+      "step": 6122
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.962643997499068e-05,
+      "loss": 19.2355,
+      "step": 6123
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.961963876892289e-05,
+      "loss": 20.5596,
+      "step": 6124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.9612836718406e-05,
+      "loss": 0.7578,
+      "step": 6125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.960603382363398e-05,
+      "loss": 21.0387,
+      "step": 6126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.959923008480072e-05,
+      "loss": 19.3505,
+      "step": 6127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.959242550210024e-05,
+      "loss": 17.2325,
+      "step": 6128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.958562007572651e-05,
+      "loss": 19.486,
+      "step": 6129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.957881380587355e-05,
+      "loss": 20.3061,
+      "step": 6130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.957200669273544e-05,
+      "loss": 0.7005,
+      "step": 6131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.956519873650622e-05,
+      "loss": 0.7787,
+      "step": 6132
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.955838993737998e-05,
+      "loss": 0.847,
+      "step": 6133
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.955158029555085e-05,
+      "loss": 20.1092,
+      "step": 6134
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.954476981121297e-05,
+      "loss": 0.7163,
+      "step": 6135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.953795848456051e-05,
+      "loss": 20.0919,
+      "step": 6136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.953114631578765e-05,
+      "loss": 21.7906,
+      "step": 6137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95243333050886e-05,
+      "loss": 0.7649,
+      "step": 6138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.951751945265761e-05,
+      "loss": 0.0582,
+      "step": 6139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.951070475868891e-05,
+      "loss": 18.4221,
+      "step": 6140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.950388922337683e-05,
+      "loss": 18.5795,
+      "step": 6141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.949707284691566e-05,
+      "loss": 0.7105,
+      "step": 6142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.949025562949975e-05,
+      "loss": 18.8022,
+      "step": 6143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.948343757132341e-05,
+      "loss": 0.7789,
+      "step": 6144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.947661867258104e-05,
+      "loss": 17.3426,
+      "step": 6145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.94697989334671e-05,
+      "loss": 20.7379,
+      "step": 6146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.946297835417595e-05,
+      "loss": 0.745,
+      "step": 6147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.945615693490206e-05,
+      "loss": 0.7059,
+      "step": 6148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944933467583993e-05,
+      "loss": 21.6573,
+      "step": 6149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944251157718406e-05,
+      "loss": 0.7842,
+      "step": 6150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.943568763912893e-05,
+      "loss": 19.4405,
+      "step": 6151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.942886286186914e-05,
+      "loss": 22.7972,
+      "step": 6152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.942203724559925e-05,
+      "loss": 0.0848,
+      "step": 6153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.941521079051381e-05,
+      "loss": 0.7571,
+      "step": 6154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.940838349680752e-05,
+      "loss": 19.8173,
+      "step": 6155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.940155536467498e-05,
+      "loss": 0.7528,
+      "step": 6156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939472639431086e-05,
+      "loss": 0.704,
+      "step": 6157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.938789658590986e-05,
+      "loss": 0.6929,
+      "step": 6158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.938106593966668e-05,
+      "loss": 19.4111,
+      "step": 6159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.937423445577609e-05,
+      "loss": 0.7439,
+      "step": 6160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.936740213443282e-05,
+      "loss": 0.6913,
+      "step": 6161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.936056897583169e-05,
+      "loss": 21.0868,
+      "step": 6162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.935373498016749e-05,
+      "loss": 0.7583,
+      "step": 6163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.934690014763506e-05,
+      "loss": 0.7558,
+      "step": 6164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.934006447842925e-05,
+      "loss": 22.7026,
+      "step": 6165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.933322797274498e-05,
+      "loss": 19.21,
+      "step": 6166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.932639063077712e-05,
+      "loss": 0.0771,
+      "step": 6167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93195524527206e-05,
+      "loss": 0.0697,
+      "step": 6168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93127134387704e-05,
+      "loss": 20.2401,
+      "step": 6169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.930587358912149e-05,
+      "loss": 20.1461,
+      "step": 6170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.929903290396887e-05,
+      "loss": 19.0014,
+      "step": 6171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.929219138350757e-05,
+      "loss": 18.1464,
+      "step": 6172
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.928534902793261e-05,
+      "loss": 19.8242,
+      "step": 6173
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.927850583743912e-05,
+      "loss": 19.8131,
+      "step": 6174
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.927166181222214e-05,
+      "loss": 0.0893,
+      "step": 6175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.926481695247684e-05,
+      "loss": 0.7603,
+      "step": 6176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925797125839835e-05,
+      "loss": 19.7201,
+      "step": 6177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925112473018181e-05,
+      "loss": 19.2228,
+      "step": 6178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.924427736802246e-05,
+      "loss": 21.3805,
+      "step": 6179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923742917211548e-05,
+      "loss": 0.7159,
+      "step": 6180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923058014265612e-05,
+      "loss": 0.0719,
+      "step": 6181
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.922373027983966e-05,
+      "loss": 17.5041,
+      "step": 6182
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.921687958386138e-05,
+      "loss": 0.0675,
+      "step": 6183
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.921002805491658e-05,
+      "loss": 21.0163,
+      "step": 6184
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92031756932006e-05,
+      "loss": 0.7165,
+      "step": 6185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91963224989088e-05,
+      "loss": 20.5259,
+      "step": 6186
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.918946847223657e-05,
+      "loss": 20.2512,
+      "step": 6187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.918261361337931e-05,
+      "loss": 22.0888,
+      "step": 6188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.917575792253246e-05,
+      "loss": 20.6605,
+      "step": 6189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.916890139989145e-05,
+      "loss": 0.7435,
+      "step": 6190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91620440456518e-05,
+      "loss": 20.4715,
+      "step": 6191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.915518586000897e-05,
+      "loss": 17.3323,
+      "step": 6192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914832684315849e-05,
+      "loss": 20.0146,
+      "step": 6193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914146699529594e-05,
+      "loss": 21.9066,
+      "step": 6194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.913460631661687e-05,
+      "loss": 0.7583,
+      "step": 6195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.912774480731689e-05,
+      "loss": 20.4084,
+      "step": 6196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91208824675916e-05,
+      "loss": 20.5025,
+      "step": 6197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.911401929763664e-05,
+      "loss": 19.8624,
+      "step": 6198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.910715529764772e-05,
+      "loss": 0.08,
+      "step": 6199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.910029046782048e-05,
+      "loss": 0.6611,
+      "step": 6200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.909342480835067e-05,
+      "loss": 0.7057,
+      "step": 6201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.9086558319434e-05,
+      "loss": 20.3033,
+      "step": 6202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.907969100126626e-05,
+      "loss": 20.1468,
+      "step": 6203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.907282285404321e-05,
+      "loss": 0.0733,
+      "step": 6204
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.906595387796068e-05,
+      "loss": 20.2589,
+      "step": 6205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90590840732145e-05,
+      "loss": 0.7142,
+      "step": 6206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.905221344000051e-05,
+      "loss": 22.0648,
+      "step": 6207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.904534197851461e-05,
+      "loss": 19.5831,
+      "step": 6208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90384696889527e-05,
+      "loss": 20.5062,
+      "step": 6209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90315965715107e-05,
+      "loss": 0.7058,
+      "step": 6210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.902472262638456e-05,
+      "loss": 0.0701,
+      "step": 6211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.901784785377026e-05,
+      "loss": 20.4902,
+      "step": 6212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.901097225386381e-05,
+      "loss": 0.0478,
+      "step": 6213
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90040958268612e-05,
+      "loss": 0.7444,
+      "step": 6214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.899721857295848e-05,
+      "loss": 18.6865,
+      "step": 6215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.899034049235175e-05,
+      "loss": 0.721,
+      "step": 6216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.898346158523709e-05,
+      "loss": 0.7351,
+      "step": 6217
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89765818518106e-05,
+      "loss": 21.1679,
+      "step": 6218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.896970129226843e-05,
+      "loss": 21.9048,
+      "step": 6219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.896281990680678e-05,
+      "loss": 19.3677,
+      "step": 6220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.895593769562176e-05,
+      "loss": 21.7988,
+      "step": 6221
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894905465890963e-05,
+      "loss": 19.1725,
+      "step": 6222
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894217079686663e-05,
+      "loss": 19.4709,
+      "step": 6223
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.893528610968899e-05,
+      "loss": 18.9705,
+      "step": 6224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.8928400597573e-05,
+      "loss": 19.056,
+      "step": 6225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.892151426071497e-05,
+      "loss": 20.6046,
+      "step": 6226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.891462709931124e-05,
+      "loss": 19.7925,
+      "step": 6227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.890773911355813e-05,
+      "loss": 16.1612,
+      "step": 6228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.890085030365203e-05,
+      "loss": 0.7177,
+      "step": 6229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.889396066978934e-05,
+      "loss": 0.0653,
+      "step": 6230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.888707021216649e-05,
+      "loss": 21.0437,
+      "step": 6231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.888017893097991e-05,
+      "loss": 22.2197,
+      "step": 6232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.887328682642608e-05,
+      "loss": 0.7327,
+      "step": 6233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.88663938987015e-05,
+      "loss": 20.8871,
+      "step": 6234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.885950014800266e-05,
+      "loss": 0.7867,
+      "step": 6235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.885260557452612e-05,
+      "loss": 22.8715,
+      "step": 6236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.884571017846842e-05,
+      "loss": 0.7734,
+      "step": 6237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.883881396002618e-05,
+      "loss": 0.7139,
+      "step": 6238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.883191691939598e-05,
+      "loss": 21.2203,
+      "step": 6239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.882501905677449e-05,
+      "loss": 20.3938,
+      "step": 6240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881812037235831e-05,
+      "loss": 20.2601,
+      "step": 6241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881122086634416e-05,
+      "loss": 19.8964,
+      "step": 6242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.880432053892875e-05,
+      "loss": 0.7688,
+      "step": 6243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.879741939030877e-05,
+      "loss": 0.7644,
+      "step": 6244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8790517420681e-05,
+      "loss": 19.6183,
+      "step": 6245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.87836146302422e-05,
+      "loss": 0.084,
+      "step": 6246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.877671101918917e-05,
+      "loss": 21.8567,
+      "step": 6247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.876980658771875e-05,
+      "loss": 0.0714,
+      "step": 6248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.876290133602776e-05,
+      "loss": 0.8138,
+      "step": 6249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.875599526431306e-05,
+      "loss": 20.0627,
+      "step": 6250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874908837277157e-05,
+      "loss": 20.2968,
+      "step": 6251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874218066160018e-05,
+      "loss": 18.2098,
+      "step": 6252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.873527213099584e-05,
+      "loss": 0.6993,
+      "step": 6253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.872836278115551e-05,
+      "loss": 0.0654,
+      "step": 6254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.872145261227614e-05,
+      "loss": 23.1496,
+      "step": 6255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.87145416245548e-05,
+      "loss": 19.7803,
+      "step": 6256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.870762981818848e-05,
+      "loss": 18.9607,
+      "step": 6257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.870071719337423e-05,
+      "loss": 18.8092,
+      "step": 6258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.869380375030915e-05,
+      "loss": 19.0072,
+      "step": 6259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.868688948919032e-05,
+      "loss": 19.8997,
+      "step": 6260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867997441021488e-05,
+      "loss": 19.6853,
+      "step": 6261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867305851357996e-05,
+      "loss": 19.613,
+      "step": 6262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.866614179948275e-05,
+      "loss": 20.6147,
+      "step": 6263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865922426812045e-05,
+      "loss": 0.7103,
+      "step": 6264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865230591969023e-05,
+      "loss": 19.7587,
+      "step": 6265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.86453867543894e-05,
+      "loss": 0.6762,
+      "step": 6266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.863846677241517e-05,
+      "loss": 0.0535,
+      "step": 6267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.863154597396482e-05,
+      "loss": 17.8698,
+      "step": 6268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.862462435923572e-05,
+      "loss": 20.7759,
+      "step": 6269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.861770192842516e-05,
+      "loss": 20.1323,
+      "step": 6270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.86107786817305e-05,
+      "loss": 0.7183,
+      "step": 6271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.860385461934913e-05,
+      "loss": 19.9885,
+      "step": 6272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.859692974147844e-05,
+      "loss": 19.1782,
+      "step": 6273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.859000404831587e-05,
+      "loss": 20.1011,
+      "step": 6274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.858307754005886e-05,
+      "loss": 0.0671,
+      "step": 6275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85761502169049e-05,
+      "loss": 0.7879,
+      "step": 6276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.856922207905146e-05,
+      "loss": 0.7143,
+      "step": 6277
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.856229312669609e-05,
+      "loss": 21.3181,
+      "step": 6278
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.855536336003631e-05,
+      "loss": 0.7396,
+      "step": 6279
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.854843277926968e-05,
+      "loss": 19.5512,
+      "step": 6280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.854150138459382e-05,
+      "loss": 0.7195,
+      "step": 6281
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.853456917620632e-05,
+      "loss": 0.7356,
+      "step": 6282
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.852763615430481e-05,
+      "loss": 19.8353,
+      "step": 6283
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.852070231908697e-05,
+      "loss": 0.0729,
+      "step": 6284
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.851376767075047e-05,
+      "loss": 19.4835,
+      "step": 6285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.850683220949302e-05,
+      "loss": 18.7286,
+      "step": 6286
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.849989593551234e-05,
+      "loss": 18.6896,
+      "step": 6287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84929588490062e-05,
+      "loss": 0.0908,
+      "step": 6288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.848602095017235e-05,
+      "loss": 20.5085,
+      "step": 6289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847908223920859e-05,
+      "loss": 0.7628,
+      "step": 6290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847214271631276e-05,
+      "loss": 0.7589,
+      "step": 6291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.846520238168269e-05,
+      "loss": 0.0822,
+      "step": 6292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.845826123551626e-05,
+      "loss": 19.2253,
+      "step": 6293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.845131927801135e-05,
+      "loss": 20.1934,
+      "step": 6294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.844437650936586e-05,
+      "loss": 0.7827,
+      "step": 6295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.843743292977777e-05,
+      "loss": 21.0311,
+      "step": 6296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.843048853944499e-05,
+      "loss": 20.9048,
+      "step": 6297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.842354333856554e-05,
+      "loss": 23.0482,
+      "step": 6298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.841659732733742e-05,
+      "loss": 0.7349,
+      "step": 6299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.840965050595863e-05,
+      "loss": 23.1596,
+      "step": 6300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.840270287462726e-05,
+      "loss": 20.8696,
+      "step": 6301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.839575443354135e-05,
+      "loss": 0.7185,
+      "step": 6302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.838880518289903e-05,
+      "loss": 19.9677,
+      "step": 6303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.83818551228984e-05,
+      "loss": 19.8559,
+      "step": 6304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.837490425373762e-05,
+      "loss": 21.295,
+      "step": 6305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.836795257561484e-05,
+      "loss": 0.7376,
+      "step": 6306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.836100008872829e-05,
+      "loss": 19.1025,
+      "step": 6307
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.835404679327612e-05,
+      "loss": 0.696,
+      "step": 6308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.834709268945662e-05,
+      "loss": 18.5091,
+      "step": 6309
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.834013777746804e-05,
+      "loss": 18.0652,
+      "step": 6310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.833318205750863e-05,
+      "loss": 21.0973,
+      "step": 6311
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832622552977675e-05,
+      "loss": 0.722,
+      "step": 6312
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.831926819447068e-05,
+      "loss": 20.2129,
+      "step": 6313
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.83123100517888e-05,
+      "loss": 19.4534,
+      "step": 6314
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.830535110192947e-05,
+      "loss": 19.1609,
+      "step": 6315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82983913450911e-05,
+      "loss": 0.7611,
+      "step": 6316
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.829143078147211e-05,
+      "loss": 19.3605,
+      "step": 6317
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.828446941127095e-05,
+      "loss": 20.277,
+      "step": 6318
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.827750723468605e-05,
+      "loss": 20.4816,
+      "step": 6319
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.827054425191595e-05,
+      "loss": 0.7478,
+      "step": 6320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.826358046315911e-05,
+      "loss": 0.7661,
+      "step": 6321
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.825661586861414e-05,
+      "loss": 21.4706,
+      "step": 6322
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.824965046847952e-05,
+      "loss": 0.0675,
+      "step": 6323
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.824268426295389e-05,
+      "loss": 0.6951,
+      "step": 6324
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.823571725223582e-05,
+      "loss": 19.8462,
+      "step": 6325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.822874943652394e-05,
+      "loss": 17.7387,
+      "step": 6326
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.822178081601694e-05,
+      "loss": 0.0702,
+      "step": 6327
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.821481139091345e-05,
+      "loss": 21.1123,
+      "step": 6328
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820784116141218e-05,
+      "loss": 21.7353,
+      "step": 6329
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820087012771184e-05,
+      "loss": 0.0653,
+      "step": 6330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81938982900112e-05,
+      "loss": 0.0779,
+      "step": 6331
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8186925648509e-05,
+      "loss": 18.6026,
+      "step": 6332
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.817995220340404e-05,
+      "loss": 0.7481,
+      "step": 6333
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.817297795489513e-05,
+      "loss": 20.9661,
+      "step": 6334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81660029031811e-05,
+      "loss": 21.1168,
+      "step": 6335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.815902704846082e-05,
+      "loss": 0.0724,
+      "step": 6336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.815205039093315e-05,
+      "loss": 20.6189,
+      "step": 6337
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8145072930797e-05,
+      "loss": 0.6735,
+      "step": 6338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.813809466825132e-05,
+      "loss": 19.1939,
+      "step": 6339
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.813111560349503e-05,
+      "loss": 19.9262,
+      "step": 6340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81241357367271e-05,
+      "loss": 18.424,
+      "step": 6341
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.811715506814653e-05,
+      "loss": 17.9715,
+      "step": 6342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.811017359795235e-05,
+      "loss": 0.0831,
+      "step": 6343
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.810319132634358e-05,
+      "loss": 19.6546,
+      "step": 6344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80962082535193e-05,
+      "loss": 0.7746,
+      "step": 6345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.808922437967859e-05,
+      "loss": 0.7216,
+      "step": 6346
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.808223970502055e-05,
+      "loss": 0.0743,
+      "step": 6347
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.807525422974434e-05,
+      "loss": 0.764,
+      "step": 6348
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.806826795404908e-05,
+      "loss": 0.7709,
+      "step": 6349
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.806128087813395e-05,
+      "loss": 19.8262,
+      "step": 6350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80542930021982e-05,
+      "loss": 21.903,
+      "step": 6351
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.804730432644097e-05,
+      "loss": 0.7525,
+      "step": 6352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.804031485106158e-05,
+      "loss": 19.7193,
+      "step": 6353
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.803332457625925e-05,
+      "loss": 17.6275,
+      "step": 6354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.802633350223329e-05,
+      "loss": 18.9965,
+      "step": 6355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.801934162918302e-05,
+      "loss": 0.0616,
+      "step": 6356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.801234895730776e-05,
+      "loss": 20.8481,
+      "step": 6357
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80053554868069e-05,
+      "loss": 18.6243,
+      "step": 6358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.799836121787978e-05,
+      "loss": 20.6869,
+      "step": 6359
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.799136615072583e-05,
+      "loss": 19.1481,
+      "step": 6360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.798437028554447e-05,
+      "loss": 18.9402,
+      "step": 6361
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.797737362253516e-05,
+      "loss": 0.7153,
+      "step": 6362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.797037616189738e-05,
+      "loss": 22.0233,
+      "step": 6363
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.79633779038306e-05,
+      "loss": 0.0765,
+      "step": 6364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.795637884853435e-05,
+      "loss": 18.4546,
+      "step": 6365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.794937899620818e-05,
+      "loss": 19.0528,
+      "step": 6366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.794237834705163e-05,
+      "loss": 0.0665,
+      "step": 6367
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.793537690126434e-05,
+      "loss": 21.1221,
+      "step": 6368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.792837465904586e-05,
+      "loss": 20.2449,
+      "step": 6369
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.792137162059586e-05,
+      "loss": 18.9501,
+      "step": 6370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.791436778611396e-05,
+      "loss": 20.2075,
+      "step": 6371
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790736315579988e-05,
+      "loss": 17.476,
+      "step": 6372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790035772985328e-05,
+      "loss": 0.7549,
+      "step": 6373
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78933515084739e-05,
+      "loss": 17.9366,
+      "step": 6374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.788634449186152e-05,
+      "loss": 20.5436,
+      "step": 6375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787933668021586e-05,
+      "loss": 0.6845,
+      "step": 6376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787232807373672e-05,
+      "loss": 19.0421,
+      "step": 6377
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.786531867262393e-05,
+      "loss": 0.0958,
+      "step": 6378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.785830847707731e-05,
+      "loss": 20.9937,
+      "step": 6379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.785129748729674e-05,
+      "loss": 0.815,
+      "step": 6380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78442857034821e-05,
+      "loss": 0.7302,
+      "step": 6381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783727312583326e-05,
+      "loss": 0.0743,
+      "step": 6382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783025975455018e-05,
+      "loss": 19.0645,
+      "step": 6383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78232455898328e-05,
+      "loss": 0.6819,
+      "step": 6384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78162306318811e-05,
+      "loss": 20.58,
+      "step": 6385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.780921488089506e-05,
+      "loss": 19.3441,
+      "step": 6386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78021983370747e-05,
+      "loss": 0.0705,
+      "step": 6387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.779518100062006e-05,
+      "loss": 0.734,
+      "step": 6388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.778816287173123e-05,
+      "loss": 20.897,
+      "step": 6389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.778114395060825e-05,
+      "loss": 0.7418,
+      "step": 6390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.777412423745125e-05,
+      "loss": 19.2158,
+      "step": 6391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.776710373246038e-05,
+      "loss": 19.8028,
+      "step": 6392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.776008243583576e-05,
+      "loss": 0.7007,
+      "step": 6393
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.775306034777758e-05,
+      "loss": 19.0724,
+      "step": 6394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.774603746848605e-05,
+      "loss": 0.7199,
+      "step": 6395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.773901379816135e-05,
+      "loss": 0.0454,
+      "step": 6396
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.773198933700379e-05,
+      "loss": 0.0881,
+      "step": 6397
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.772496408521357e-05,
+      "loss": 0.7677,
+      "step": 6398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.771793804299101e-05,
+      "loss": 20.0708,
+      "step": 6399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.771091121053641e-05,
+      "loss": 20.3066,
+      "step": 6400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.770388358805012e-05,
+      "loss": 20.3988,
+      "step": 6401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.769685517573248e-05,
+      "loss": 20.4696,
+      "step": 6402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.768982597378387e-05,
+      "loss": 21.2461,
+      "step": 6403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.768279598240471e-05,
+      "loss": 19.8932,
+      "step": 6404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.76757652017954e-05,
+      "loss": 0.7182,
+      "step": 6405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766873363215638e-05,
+      "loss": 0.0727,
+      "step": 6406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766170127368815e-05,
+      "loss": 18.2539,
+      "step": 6407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.765466812659119e-05,
+      "loss": 0.7429,
+      "step": 6408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.764763419106599e-05,
+      "loss": 20.6906,
+      "step": 6409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.764059946731312e-05,
+      "loss": 0.7307,
+      "step": 6410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.763356395553312e-05,
+      "loss": 21.0373,
+      "step": 6411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.762652765592655e-05,
+      "loss": 23.3581,
+      "step": 6412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.761949056869404e-05,
+      "loss": 17.6301,
+      "step": 6413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.761245269403623e-05,
+      "loss": 0.0673,
+      "step": 6414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.760541403215374e-05,
+      "loss": 19.4585,
+      "step": 6415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.759837458324724e-05,
+      "loss": 0.7297,
+      "step": 6416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.759133434751744e-05,
+      "loss": 0.701,
+      "step": 6417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.758429332516505e-05,
+      "loss": 18.4613,
+      "step": 6418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.757725151639081e-05,
+      "loss": 18.2447,
+      "step": 6419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.757020892139547e-05,
+      "loss": 19.3136,
+      "step": 6420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.756316554037983e-05,
+      "loss": 21.0165,
+      "step": 6421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.755612137354467e-05,
+      "loss": 20.6675,
+      "step": 6422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.754907642109084e-05,
+      "loss": 22.3443,
+      "step": 6423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.754203068321917e-05,
+      "loss": 0.7796,
+      "step": 6424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.753498416013057e-05,
+      "loss": 0.7826,
+      "step": 6425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.75279368520259e-05,
+      "loss": 0.7373,
+      "step": 6426
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.752088875910608e-05,
+      "loss": 0.7045,
+      "step": 6427
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.751383988157206e-05,
+      "loss": 0.7304,
+      "step": 6428
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.750679021962479e-05,
+      "loss": 0.6863,
+      "step": 6429
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.749973977346528e-05,
+      "loss": 21.0729,
+      "step": 6430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74926885432945e-05,
+      "loss": 20.2616,
+      "step": 6431
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.748563652931349e-05,
+      "loss": 20.1604,
+      "step": 6432
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.747858373172334e-05,
+      "loss": 0.7867,
+      "step": 6433
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.747153015072505e-05,
+      "loss": 19.6169,
+      "step": 6434
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74644757865198e-05,
+      "loss": 18.7546,
+      "step": 6435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.745742063930865e-05,
+      "loss": 21.6665,
+      "step": 6436
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.745036470929274e-05,
+      "loss": 0.7687,
+      "step": 6437
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.744330799667329e-05,
+      "loss": 0.822,
+      "step": 6438
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.743625050165141e-05,
+      "loss": 0.7523,
+      "step": 6439
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.742919222442837e-05,
+      "loss": 18.8703,
+      "step": 6440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.742213316520535e-05,
+      "loss": 0.057,
+      "step": 6441
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.741507332418366e-05,
+      "loss": 19.4095,
+      "step": 6442
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74080127015645e-05,
+      "loss": 18.3884,
+      "step": 6443
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.740095129754921e-05,
+      "loss": 19.6907,
+      "step": 6444
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.739388911233913e-05,
+      "loss": 21.2169,
+      "step": 6445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.738682614613555e-05,
+      "loss": 19.0165,
+      "step": 6446
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.737976239913989e-05,
+      "loss": 20.8277,
+      "step": 6447
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.737269787155348e-05,
+      "loss": 18.3894,
+      "step": 6448
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.736563256357776e-05,
+      "loss": 0.7396,
+      "step": 6449
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.735856647541416e-05,
+      "loss": 0.6931,
+      "step": 6450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.735149960726413e-05,
+      "loss": 0.7272,
+      "step": 6451
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.734443195932913e-05,
+      "loss": 0.7352,
+      "step": 6452
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.733736353181067e-05,
+      "loss": 0.7757,
+      "step": 6453
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.733029432491028e-05,
+      "loss": 0.7062,
+      "step": 6454
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.732322433882949e-05,
+      "loss": 0.7552,
+      "step": 6455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.731615357376985e-05,
+      "loss": 20.8806,
+      "step": 6456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.730908202993296e-05,
+      "loss": 0.7095,
+      "step": 6457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.730200970752043e-05,
+      "loss": 0.7096,
+      "step": 6458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72949366067339e-05,
+      "loss": 20.8218,
+      "step": 6459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.7287862727775e-05,
+      "loss": 20.5598,
+      "step": 6460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72807880708454e-05,
+      "loss": 20.7925,
+      "step": 6461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.727371263614682e-05,
+      "loss": 20.2155,
+      "step": 6462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.726663642388098e-05,
+      "loss": 20.4182,
+      "step": 6463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72595594342496e-05,
+      "loss": 0.0772,
+      "step": 6464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.725248166745447e-05,
+      "loss": 0.7187,
+      "step": 6465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.724540312369734e-05,
+      "loss": 20.8605,
+      "step": 6466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.723832380318005e-05,
+      "loss": 19.9566,
+      "step": 6467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.723124370610442e-05,
+      "loss": 0.0772,
+      "step": 6468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72241628326723e-05,
+      "loss": 0.0589,
+      "step": 6469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.721708118308557e-05,
+      "loss": 19.3629,
+      "step": 6470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.720999875754611e-05,
+      "loss": 0.083,
+      "step": 6471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.720291555625583e-05,
+      "loss": 19.7795,
+      "step": 6472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.719583157941672e-05,
+      "loss": 20.5211,
+      "step": 6473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71887468272307e-05,
+      "loss": 21.4464,
+      "step": 6474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.718166129989976e-05,
+      "loss": 20.447,
+      "step": 6475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.717457499762593e-05,
+      "loss": 19.56,
+      "step": 6476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71674879206112e-05,
+      "loss": 0.7607,
+      "step": 6477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.716040006905765e-05,
+      "loss": 19.7519,
+      "step": 6478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.715331144316736e-05,
+      "loss": 0.8169,
+      "step": 6479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.714622204314239e-05,
+      "loss": 17.7289,
+      "step": 6480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.713913186918489e-05,
+      "loss": 19.6878,
+      "step": 6481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.713204092149698e-05,
+      "loss": 19.7673,
+      "step": 6482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.712494920028081e-05,
+      "loss": 20.1598,
+      "step": 6483
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711785670573862e-05,
+      "loss": 0.7374,
+      "step": 6484
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711076343807255e-05,
+      "loss": 19.6266,
+      "step": 6485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.710366939748488e-05,
+      "loss": 20.1108,
+      "step": 6486
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70965745841778e-05,
+      "loss": 19.0651,
+      "step": 6487
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.708947899835364e-05,
+      "loss": 18.2988,
+      "step": 6488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.708238264021466e-05,
+      "loss": 20.771,
+      "step": 6489
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70752855099632e-05,
+      "loss": 18.5306,
+      "step": 6490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.706818760780156e-05,
+      "loss": 18.7117,
+      "step": 6491
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.706108893393212e-05,
+      "loss": 19.7448,
+      "step": 6492
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.705398948855729e-05,
+      "loss": 0.701,
+      "step": 6493
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.704688927187942e-05,
+      "loss": 21.6245,
+      "step": 6494
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.703978828410096e-05,
+      "loss": 19.5918,
+      "step": 6495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.703268652542439e-05,
+      "loss": 0.7508,
+      "step": 6496
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.702558399605213e-05,
+      "loss": 0.7135,
+      "step": 6497
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.701848069618669e-05,
+      "loss": 0.7202,
+      "step": 6498
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.701137662603059e-05,
+      "loss": 19.5717,
+      "step": 6499
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.700427178578635e-05,
+      "loss": 0.7278,
+      "step": 6500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699716617565655e-05,
+      "loss": 0.7044,
+      "step": 6501
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699005979584374e-05,
+      "loss": 21.5412,
+      "step": 6502
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.698295264655056e-05,
+      "loss": 0.7458,
+      "step": 6503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.697584472797958e-05,
+      "loss": 19.4117,
+      "step": 6504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69687360403335e-05,
+      "loss": 19.1646,
+      "step": 6505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.696162658381495e-05,
+      "loss": 0.6573,
+      "step": 6506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.695451635862664e-05,
+      "loss": 22.6002,
+      "step": 6507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.694740536497127e-05,
+      "loss": 19.8046,
+      "step": 6508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.694029360305157e-05,
+      "loss": 17.9749,
+      "step": 6509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.693318107307032e-05,
+      "loss": 0.7561,
+      "step": 6510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.692606777523025e-05,
+      "loss": 0.7164,
+      "step": 6511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69189537097342e-05,
+      "loss": 0.7664,
+      "step": 6512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.691183887678498e-05,
+      "loss": 0.714,
+      "step": 6513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.690472327658543e-05,
+      "loss": 19.7191,
+      "step": 6514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68976069093384e-05,
+      "loss": 19.1631,
+      "step": 6515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.689048977524677e-05,
+      "loss": 0.7622,
+      "step": 6516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.688337187451351e-05,
+      "loss": 19.5189,
+      "step": 6517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.687625320734147e-05,
+      "loss": 21.1123,
+      "step": 6518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.686913377393366e-05,
+      "loss": 18.921,
+      "step": 6519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.686201357449302e-05,
+      "loss": 19.9762,
+      "step": 6520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.685489260922254e-05,
+      "loss": 18.8075,
+      "step": 6521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.684777087832529e-05,
+      "loss": 0.7448,
+      "step": 6522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.684064838200424e-05,
+      "loss": 0.7444,
+      "step": 6523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68335251204625e-05,
+      "loss": 0.0772,
+      "step": 6524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.682640109390314e-05,
+      "loss": 19.9454,
+      "step": 6525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681927630252923e-05,
+      "loss": 20.0692,
+      "step": 6526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681215074654396e-05,
+      "loss": 21.9404,
+      "step": 6527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.680502442615041e-05,
+      "loss": 20.4598,
+      "step": 6528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67978973415518e-05,
+      "loss": 20.4945,
+      "step": 6529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.679076949295131e-05,
+      "loss": 0.0743,
+      "step": 6530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.678364088055215e-05,
+      "loss": 21.1496,
+      "step": 6531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.677651150455754e-05,
+      "loss": 0.7476,
+      "step": 6532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.676938136517076e-05,
+      "loss": 18.6242,
+      "step": 6533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.676225046259508e-05,
+      "loss": 18.6874,
+      "step": 6534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67551187970338e-05,
+      "loss": 20.0,
+      "step": 6535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.674798636869023e-05,
+      "loss": 17.5936,
+      "step": 6536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.674085317776773e-05,
+      "loss": 19.8956,
+      "step": 6537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.673371922446965e-05,
+      "loss": 18.9634,
+      "step": 6538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.672658450899943e-05,
+      "loss": 0.0714,
+      "step": 6539
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.671944903156038e-05,
+      "loss": 19.5961,
+      "step": 6540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.671231279235603e-05,
+      "loss": 20.4925,
+      "step": 6541
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.670517579158977e-05,
+      "loss": 0.7081,
+      "step": 6542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66980380294651e-05,
+      "loss": 0.7336,
+      "step": 6543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.669089950618552e-05,
+      "loss": 20.544,
+      "step": 6544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.668376022195453e-05,
+      "loss": 0.7505,
+      "step": 6545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.667662017697567e-05,
+      "loss": 0.6671,
+      "step": 6546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.666947937145252e-05,
+      "loss": 20.9608,
+      "step": 6547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.666233780558864e-05,
+      "loss": 19.6126,
+      "step": 6548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.665519547958767e-05,
+      "loss": 19.1982,
+      "step": 6549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66480523936532e-05,
+      "loss": 0.7508,
+      "step": 6550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.664090854798889e-05,
+      "loss": 0.739,
+      "step": 6551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66337639427984e-05,
+      "loss": 18.485,
+      "step": 6552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.662661857828543e-05,
+      "loss": 17.334,
+      "step": 6553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.661947245465371e-05,
+      "loss": 0.0893,
+      "step": 6554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.661232557210693e-05,
+      "loss": 19.5648,
+      "step": 6555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66051779308489e-05,
+      "loss": 0.7236,
+      "step": 6556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.659802953108336e-05,
+      "loss": 21.7744,
+      "step": 6557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.659088037301412e-05,
+      "loss": 23.3537,
+      "step": 6558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.6583730456845e-05,
+      "loss": 0.0807,
+      "step": 6559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.657657978277984e-05,
+      "loss": 17.2448,
+      "step": 6560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656942835102252e-05,
+      "loss": 0.7517,
+      "step": 6561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656227616177691e-05,
+      "loss": 0.7174,
+      "step": 6562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.655512321524691e-05,
+      "loss": 19.2167,
+      "step": 6563
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.654796951163647e-05,
+      "loss": 0.7583,
+      "step": 6564
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.654081505114952e-05,
+      "loss": 0.0644,
+      "step": 6565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.653365983399005e-05,
+      "loss": 18.927,
+      "step": 6566
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.652650386036204e-05,
+      "loss": 0.7152,
+      "step": 6567
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.651934713046951e-05,
+      "loss": 20.3906,
+      "step": 6568
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.651218964451651e-05,
+      "loss": 19.8764,
+      "step": 6569
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.650503140270708e-05,
+      "loss": 21.1439,
+      "step": 6570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.649787240524529e-05,
+      "loss": 19.434,
+      "step": 6571
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.649071265233528e-05,
+      "loss": 0.0888,
+      "step": 6572
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.648355214418113e-05,
+      "loss": 0.0627,
+      "step": 6573
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.647639088098702e-05,
+      "loss": 0.7364,
+      "step": 6574
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.646922886295708e-05,
+      "loss": 0.7133,
+      "step": 6575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.646206609029555e-05,
+      "loss": 18.0728,
+      "step": 6576
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.645490256320657e-05,
+      "loss": 20.0845,
+      "step": 6577
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644773828189445e-05,
+      "loss": 0.6982,
+      "step": 6578
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644057324656336e-05,
+      "loss": 21.1253,
+      "step": 6579
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.643340745741762e-05,
+      "loss": 22.2439,
+      "step": 6580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.642624091466153e-05,
+      "loss": 20.5708,
+      "step": 6581
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.641907361849939e-05,
+      "loss": 0.6961,
+      "step": 6582
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.641190556913553e-05,
+      "loss": 20.4349,
+      "step": 6583
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.640473676677432e-05,
+      "loss": 20.607,
+      "step": 6584
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.639756721162016e-05,
+      "loss": 0.7197,
+      "step": 6585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.639039690387742e-05,
+      "loss": 0.0827,
+      "step": 6586
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.638322584375055e-05,
+      "loss": 0.755,
+      "step": 6587
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.637605403144398e-05,
+      "loss": 20.0814,
+      "step": 6588
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.636888146716217e-05,
+      "loss": 21.6992,
+      "step": 6589
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.636170815110963e-05,
+      "loss": 19.5355,
+      "step": 6590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.635453408349086e-05,
+      "loss": 19.6906,
+      "step": 6591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634735926451039e-05,
+      "loss": 21.779,
+      "step": 6592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634018369437277e-05,
+      "loss": 18.4945,
+      "step": 6593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.633300737328256e-05,
+      "loss": 21.5973,
+      "step": 6594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.632583030144439e-05,
+      "loss": 20.5728,
+      "step": 6595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.631865247906284e-05,
+      "loss": 20.667,
+      "step": 6596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.631147390634259e-05,
+      "loss": 0.7748,
+      "step": 6597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.630429458348825e-05,
+      "loss": 20.8405,
+      "step": 6598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.629711451070455e-05,
+      "loss": 20.8389,
+      "step": 6599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.628993368819615e-05,
+      "loss": 21.8623,
+      "step": 6600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.62827521161678e-05,
+      "loss": 19.0761,
+      "step": 6601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.627556979482424e-05,
+      "loss": 19.2963,
+      "step": 6602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.626838672437023e-05,
+      "loss": 20.0228,
+      "step": 6603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.626120290501055e-05,
+      "loss": 0.7688,
+      "step": 6604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.625401833695003e-05,
+      "loss": 18.3981,
+      "step": 6605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.624683302039348e-05,
+      "loss": 20.1157,
+      "step": 6606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.623964695554578e-05,
+      "loss": 0.7122,
+      "step": 6607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.623246014261176e-05,
+      "loss": 21.931,
+      "step": 6608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.622527258179636e-05,
+      "loss": 0.747,
+      "step": 6609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621808427330447e-05,
+      "loss": 18.3367,
+      "step": 6610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621089521734104e-05,
+      "loss": 0.7357,
+      "step": 6611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.6203705414111e-05,
+      "loss": 18.7638,
+      "step": 6612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.619651486381936e-05,
+      "loss": 19.356,
+      "step": 6613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.618932356667111e-05,
+      "loss": 0.6714,
+      "step": 6614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.618213152287129e-05,
+      "loss": 21.1142,
+      "step": 6615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.61749387326249e-05,
+      "loss": 19.8128,
+      "step": 6616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.616774519613703e-05,
+      "loss": 0.0637,
+      "step": 6617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.61605509136128e-05,
+      "loss": 0.6906,
+      "step": 6618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.615335588525725e-05,
+      "loss": 0.7224,
+      "step": 6619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.614616011127557e-05,
+      "loss": 22.1106,
+      "step": 6620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.613896359187287e-05,
+      "loss": 0.0549,
+      "step": 6621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.613176632725435e-05,
+      "loss": 21.7761,
+      "step": 6622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612456831762516e-05,
+      "loss": 20.9354,
+      "step": 6623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.611736956319056e-05,
+      "loss": 20.2516,
+      "step": 6624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.611017006415575e-05,
+      "loss": 0.7392,
+      "step": 6625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.610296982072602e-05,
+      "loss": 20.5717,
+      "step": 6626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.60957688331066e-05,
+      "loss": 0.7778,
+      "step": 6627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.608856710150285e-05,
+      "loss": 0.721,
+      "step": 6628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.608136462612003e-05,
+      "loss": 0.0681,
+      "step": 6629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.607416140716353e-05,
+      "loss": 0.0795,
+      "step": 6630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.606695744483868e-05,
+      "loss": 19.0037,
+      "step": 6631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.605975273935087e-05,
+      "loss": 21.6893,
+      "step": 6632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.60525472909055e-05,
+      "loss": 18.894,
+      "step": 6633
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.604534109970802e-05,
+      "loss": 19.2441,
+      "step": 6634
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.603813416596384e-05,
+      "loss": 18.8986,
+      "step": 6635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.603092648987846e-05,
+      "loss": 19.2684,
+      "step": 6636
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.602371807165735e-05,
+      "loss": 17.008,
+      "step": 6637
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.601650891150606e-05,
+      "loss": 0.725,
+      "step": 6638
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.600929900963005e-05,
+      "loss": 19.0972,
+      "step": 6639
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.600208836623494e-05,
+      "loss": 20.1383,
+      "step": 6640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.599487698152625e-05,
+      "loss": 21.0663,
+      "step": 6641
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.598766485570962e-05,
+      "loss": 19.5464,
+      "step": 6642
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.598045198899064e-05,
+      "loss": 19.392,
+      "step": 6643
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.597323838157495e-05,
+      "loss": 0.069,
+      "step": 6644
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.596602403366821e-05,
+      "loss": 0.7633,
+      "step": 6645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.595880894547611e-05,
+      "loss": 18.4883,
+      "step": 6646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.595159311720434e-05,
+      "loss": 0.0689,
+      "step": 6647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.594437654905862e-05,
+      "loss": 19.1173,
+      "step": 6648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.593715924124469e-05,
+      "loss": 0.07,
+      "step": 6649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.592994119396834e-05,
+      "loss": 19.4635,
+      "step": 6650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.592272240743533e-05,
+      "loss": 20.4961,
+      "step": 6651
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.591550288185146e-05,
+      "loss": 19.3129,
+      "step": 6652
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.590828261742257e-05,
+      "loss": 17.5342,
+      "step": 6653
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.590106161435449e-05,
+      "loss": 20.563,
+      "step": 6654
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.589383987285314e-05,
+      "loss": 20.1711,
+      "step": 6655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588661739312435e-05,
+      "loss": 19.9293,
+      "step": 6656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.587939417537406e-05,
+      "loss": 19.6305,
+      "step": 6657
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.587217021980821e-05,
+      "loss": 20.4832,
+      "step": 6658
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.586494552663273e-05,
+      "loss": 20.2047,
+      "step": 6659
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.58577200960536e-05,
+      "loss": 0.7177,
+      "step": 6660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.585049392827685e-05,
+      "loss": 0.6939,
+      "step": 6661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.584326702350845e-05,
+      "loss": 17.9925,
+      "step": 6662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.583603938195445e-05,
+      "loss": 19.5355,
+      "step": 6663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.582881100382094e-05,
+      "loss": 19.6652,
+      "step": 6664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.582158188931395e-05,
+      "loss": 20.4859,
+      "step": 6665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.58143520386396e-05,
+      "loss": 0.0813,
+      "step": 6666
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.580712145200401e-05,
+      "loss": 18.2725,
+      "step": 6667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579989012961335e-05,
+      "loss": 20.4076,
+      "step": 6668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579265807167375e-05,
+      "loss": 19.6428,
+      "step": 6669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.578542527839141e-05,
+      "loss": 19.6746,
+      "step": 6670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.577819174997251e-05,
+      "loss": 0.7345,
+      "step": 6671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.577095748662333e-05,
+      "loss": 22.3876,
+      "step": 6672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.576372248855007e-05,
+      "loss": 20.2622,
+      "step": 6673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5756486755959e-05,
+      "loss": 0.7602,
+      "step": 6674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.574925028905643e-05,
+      "loss": 0.0919,
+      "step": 6675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.574201308804865e-05,
+      "loss": 0.7653,
+      "step": 6676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.573477515314202e-05,
+      "loss": 0.7946,
+      "step": 6677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.572753648454285e-05,
+      "loss": 0.0633,
+      "step": 6678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.572029708245755e-05,
+      "loss": 21.2183,
+      "step": 6679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.571305694709252e-05,
+      "loss": 0.0609,
+      "step": 6680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.570581607865412e-05,
+      "loss": 0.794,
+      "step": 6681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.569857447734885e-05,
+      "loss": 0.0565,
+      "step": 6682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.569133214338311e-05,
+      "loss": 19.6732,
+      "step": 6683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.568408907696343e-05,
+      "loss": 0.7,
+      "step": 6684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.567684527829626e-05,
+      "loss": 21.0565,
+      "step": 6685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.566960074758816e-05,
+      "loss": 19.4792,
+      "step": 6686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.566235548504564e-05,
+      "loss": 0.0623,
+      "step": 6687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.565510949087526e-05,
+      "loss": 0.0745,
+      "step": 6688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.564786276528364e-05,
+      "loss": 0.7461,
+      "step": 6689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.564061530847735e-05,
+      "loss": 19.8001,
+      "step": 6690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.563336712066301e-05,
+      "loss": 0.7107,
+      "step": 6691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.562611820204728e-05,
+      "loss": 21.1287,
+      "step": 6692
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.56188685528368e-05,
+      "loss": 20.8834,
+      "step": 6693
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.56116181732383e-05,
+      "loss": 18.604,
+      "step": 6694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.560436706345843e-05,
+      "loss": 18.387,
+      "step": 6695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.559711522370398e-05,
+      "loss": 18.7472,
+      "step": 6696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.558986265418165e-05,
+      "loss": 0.7357,
+      "step": 6697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.558260935509825e-05,
+      "loss": 19.797,
+      "step": 6698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.557535532666052e-05,
+      "loss": 19.5741,
+      "step": 6699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.556810056907529e-05,
+      "loss": 21.7518,
+      "step": 6700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.556084508254942e-05,
+      "loss": 18.7017,
+      "step": 6701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.555358886728971e-05,
+      "loss": 0.7783,
+      "step": 6702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55463319235031e-05,
+      "loss": 19.3351,
+      "step": 6703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.553907425139642e-05,
+      "loss": 0.8023,
+      "step": 6704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.553181585117661e-05,
+      "loss": 0.7308,
+      "step": 6705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.552455672305064e-05,
+      "loss": 0.7672,
+      "step": 6706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55172968672254e-05,
+      "loss": 0.7129,
+      "step": 6707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.551003628390791e-05,
+      "loss": 0.7725,
+      "step": 6708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.550277497330516e-05,
+      "loss": 20.05,
+      "step": 6709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.549551293562416e-05,
+      "loss": 0.7523,
+      "step": 6710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.548825017107195e-05,
+      "loss": 18.399,
+      "step": 6711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.548098667985562e-05,
+      "loss": 0.7207,
+      "step": 6712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.547372246218219e-05,
+      "loss": 21.0503,
+      "step": 6713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.546645751825883e-05,
+      "loss": 0.7159,
+      "step": 6714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.54591918482926e-05,
+      "loss": 19.9554,
+      "step": 6715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.545192545249069e-05,
+      "loss": 19.7384,
+      "step": 6716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.544465833106023e-05,
+      "loss": 0.7108,
+      "step": 6717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.543739048420842e-05,
+      "loss": 0.7365,
+      "step": 6718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.543012191214246e-05,
+      "loss": 0.0654,
+      "step": 6719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.542285261506958e-05,
+      "loss": 20.2418,
+      "step": 6720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.541558259319703e-05,
+      "loss": 21.5103,
+      "step": 6721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.540831184673207e-05,
+      "loss": 20.8262,
+      "step": 6722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.540104037588198e-05,
+      "loss": 16.4233,
+      "step": 6723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.539376818085407e-05,
+      "loss": 0.6881,
+      "step": 6724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.538649526185568e-05,
+      "loss": 0.7254,
+      "step": 6725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.537922161909417e-05,
+      "loss": 0.0822,
+      "step": 6726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.537194725277688e-05,
+      "loss": 0.7425,
+      "step": 6727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.53646721631112e-05,
+      "loss": 18.5809,
+      "step": 6728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.535739635030456e-05,
+      "loss": 18.841,
+      "step": 6729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.535011981456439e-05,
+      "loss": 0.7747,
+      "step": 6730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.534284255609814e-05,
+      "loss": 18.8235,
+      "step": 6731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.53355645751133e-05,
+      "loss": 21.2522,
+      "step": 6732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.532828587181733e-05,
+      "loss": 21.0618,
+      "step": 6733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.532100644641775e-05,
+      "loss": 0.795,
+      "step": 6734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.531372629912211e-05,
+      "loss": 0.6906,
+      "step": 6735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.530644543013796e-05,
+      "loss": 19.9676,
+      "step": 6736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.529916383967288e-05,
+      "loss": 19.2387,
+      "step": 6737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.529188152793446e-05,
+      "loss": 21.14,
+      "step": 6738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.528459849513032e-05,
+      "loss": 18.3787,
+      "step": 6739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52773147414681e-05,
+      "loss": 0.0837,
+      "step": 6740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.527003026715544e-05,
+      "loss": 19.9831,
+      "step": 6741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.526274507240006e-05,
+      "loss": 0.0934,
+      "step": 6742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.525545915740961e-05,
+      "loss": 0.6763,
+      "step": 6743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.524817252239185e-05,
+      "loss": 0.7652,
+      "step": 6744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.524088516755448e-05,
+      "loss": 20.6471,
+      "step": 6745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52335970931053e-05,
+      "loss": 20.5707,
+      "step": 6746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.522630829925207e-05,
+      "loss": 21.41,
+      "step": 6747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.521901878620259e-05,
+      "loss": 0.7225,
+      "step": 6748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.521172855416469e-05,
+      "loss": 19.4085,
+      "step": 6749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.520443760334623e-05,
+      "loss": 19.3663,
+      "step": 6750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.519714593395504e-05,
+      "loss": 0.7355,
+      "step": 6751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.518985354619903e-05,
+      "loss": 20.5732,
+      "step": 6752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.518256044028608e-05,
+      "loss": 19.0333,
+      "step": 6753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.517526661642414e-05,
+      "loss": 0.796,
+      "step": 6754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.516797207482113e-05,
+      "loss": 21.3227,
+      "step": 6755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.516067681568503e-05,
+      "loss": 0.7291,
+      "step": 6756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.515338083922385e-05,
+      "loss": 19.9688,
+      "step": 6757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.514608414564554e-05,
+      "loss": 0.7969,
+      "step": 6758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.513878673515818e-05,
+      "loss": 19.1067,
+      "step": 6759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.51314886079698e-05,
+      "loss": 0.7417,
+      "step": 6760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.512418976428844e-05,
+      "loss": 0.7176,
+      "step": 6761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.511689020432222e-05,
+      "loss": 18.8343,
+      "step": 6762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.510958992827924e-05,
+      "loss": 18.483,
+      "step": 6763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.510228893636764e-05,
+      "loss": 0.7752,
+      "step": 6764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.509498722879556e-05,
+      "loss": 22.0879,
+      "step": 6765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.508768480577118e-05,
+      "loss": 20.0164,
+      "step": 6766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.508038166750266e-05,
+      "loss": 0.0825,
+      "step": 6767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.507307781419825e-05,
+      "loss": 0.0507,
+      "step": 6768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.506577324606614e-05,
+      "loss": 0.7164,
+      "step": 6769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505846796331462e-05,
+      "loss": 18.2847,
+      "step": 6770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505116196615194e-05,
+      "loss": 19.0336,
+      "step": 6771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50438552547864e-05,
+      "loss": 22.3066,
+      "step": 6772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.503654782942632e-05,
+      "loss": 20.8872,
+      "step": 6773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.502923969028002e-05,
+      "loss": 0.698,
+      "step": 6774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.502193083755584e-05,
+      "loss": 0.6834,
+      "step": 6775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50146212714622e-05,
+      "loss": 21.1295,
+      "step": 6776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500731099220743e-05,
+      "loss": 0.7296,
+      "step": 6777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.7236,
+      "step": 6778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.499268829504833e-05,
+      "loss": 22.9867,
+      "step": 6779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.498537587756086e-05,
+      "loss": 0.7281,
+      "step": 6780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.497806274774609e-05,
+      "loss": 20.1842,
+      "step": 6781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.497074890581247e-05,
+      "loss": 20.249,
+      "step": 6782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.496343435196857e-05,
+      "loss": 19.7648,
+      "step": 6783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.49561190864229e-05,
+      "loss": 0.725,
+      "step": 6784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.494880310938401e-05,
+      "loss": 19.1031,
+      "step": 6785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.49414864210605e-05,
+      "loss": 0.0876,
+      "step": 6786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.493416902166092e-05,
+      "loss": 20.0815,
+      "step": 6787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.492685091139395e-05,
+      "loss": 19.7171,
+      "step": 6788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.49195320904682e-05,
+      "loss": 19.3303,
+      "step": 6789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.491221255909232e-05,
+      "loss": 0.7293,
+      "step": 6790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.490489231747498e-05,
+      "loss": 0.0548,
+      "step": 6791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.48975713658249e-05,
+      "loss": 19.8691,
+      "step": 6792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.489024970435081e-05,
+      "loss": 0.7145,
+      "step": 6793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488292733326141e-05,
+      "loss": 0.7475,
+      "step": 6794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.487560425276549e-05,
+      "loss": 18.6276,
+      "step": 6795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.486828046307181e-05,
+      "loss": 0.7272,
+      "step": 6796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.486095596438918e-05,
+      "loss": 20.1826,
+      "step": 6797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.485363075692642e-05,
+      "loss": 20.8018,
+      "step": 6798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.484630484089236e-05,
+      "loss": 18.6587,
+      "step": 6799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483897821649587e-05,
+      "loss": 18.5595,
+      "step": 6800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483165088394584e-05,
+      "loss": 0.0468,
+      "step": 6801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.482432284345115e-05,
+      "loss": 20.9036,
+      "step": 6802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.481699409522074e-05,
+      "loss": 19.1082,
+      "step": 6803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.480966463946352e-05,
+      "loss": 0.7559,
+      "step": 6804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.480233447638849e-05,
+      "loss": 0.0528,
+      "step": 6805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.47950036062046e-05,
+      "loss": 19.1879,
+      "step": 6806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.478767202912087e-05,
+      "loss": 20.0609,
+      "step": 6807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.47803397453463e-05,
+      "loss": 0.0455,
+      "step": 6808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.477300675508997e-05,
+      "loss": 19.2939,
+      "step": 6809
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.476567305856092e-05,
+      "loss": 20.1718,
+      "step": 6810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.475833865596822e-05,
+      "loss": 18.8338,
+      "step": 6811
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.475100354752098e-05,
+      "loss": 0.6595,
+      "step": 6812
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.474366773342834e-05,
+      "loss": 0.75,
+      "step": 6813
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.473633121389943e-05,
+      "loss": 0.7157,
+      "step": 6814
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472899398914339e-05,
+      "loss": 20.1842,
+      "step": 6815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472165605936943e-05,
+      "loss": 17.5232,
+      "step": 6816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.471431742478675e-05,
+      "loss": 19.3745,
+      "step": 6817
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.470697808560457e-05,
+      "loss": 17.1474,
+      "step": 6818
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.469963804203214e-05,
+      "loss": 18.8718,
+      "step": 6819
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46922972942787e-05,
+      "loss": 0.0681,
+      "step": 6820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.468495584255355e-05,
+      "loss": 17.7411,
+      "step": 6821
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4677613687066e-05,
+      "loss": 19.3905,
+      "step": 6822
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.467027082802535e-05,
+      "loss": 20.22,
+      "step": 6823
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.466292726564096e-05,
+      "loss": 20.7837,
+      "step": 6824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46555830001222e-05,
+      "loss": 18.9598,
+      "step": 6825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.464823803167844e-05,
+      "loss": 19.4823,
+      "step": 6826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.464089236051907e-05,
+      "loss": 0.7459,
+      "step": 6827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.463354598685355e-05,
+      "loss": 20.1059,
+      "step": 6828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.462619891089131e-05,
+      "loss": 19.4181,
+      "step": 6829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.461885113284179e-05,
+      "loss": 20.1467,
+      "step": 6830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46115026529145e-05,
+      "loss": 19.8758,
+      "step": 6831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.460415347131892e-05,
+      "loss": 19.6815,
+      "step": 6832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.459680358826459e-05,
+      "loss": 21.8452,
+      "step": 6833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.458945300396107e-05,
+      "loss": 19.6183,
+      "step": 6834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.45821017186179e-05,
+      "loss": 20.4369,
+      "step": 6835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.457474973244467e-05,
+      "loss": 0.7064,
+      "step": 6836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.456739704565096e-05,
+      "loss": 18.0874,
+      "step": 6837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.456004365844644e-05,
+      "loss": 0.0745,
+      "step": 6838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.455268957104072e-05,
+      "loss": 0.056,
+      "step": 6839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.454533478364346e-05,
+      "loss": 0.7786,
+      "step": 6840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.453797929646436e-05,
+      "loss": 18.281,
+      "step": 6841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.453062310971313e-05,
+      "loss": 18.0124,
+      "step": 6842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.452326622359947e-05,
+      "loss": 19.5571,
+      "step": 6843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.451590863833313e-05,
+      "loss": 20.1209,
+      "step": 6844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.45085503541239e-05,
+      "loss": 0.0616,
+      "step": 6845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.450119137118153e-05,
+      "loss": 0.7129,
+      "step": 6846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.449383168971581e-05,
+      "loss": 20.9476,
+      "step": 6847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.448647130993663e-05,
+      "loss": 18.7872,
+      "step": 6848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.447911023205377e-05,
+      "loss": 20.0746,
+      "step": 6849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.447174845627708e-05,
+      "loss": 20.29,
+      "step": 6850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.44643859828165e-05,
+      "loss": 0.7388,
+      "step": 6851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.445702281188191e-05,
+      "loss": 19.0373,
+      "step": 6852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.444965894368323e-05,
+      "loss": 19.5553,
+      "step": 6853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.44422943784304e-05,
+      "loss": 20.2944,
+      "step": 6854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.443492911633339e-05,
+      "loss": 0.6543,
+      "step": 6855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.442756315760216e-05,
+      "loss": 19.3861,
+      "step": 6856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.442019650244673e-05,
+      "loss": 0.69,
+      "step": 6857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.441282915107712e-05,
+      "loss": 18.4271,
+      "step": 6858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.440546110370337e-05,
+      "loss": 0.7043,
+      "step": 6859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.439809236053555e-05,
+      "loss": 20.5295,
+      "step": 6860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.439072292178371e-05,
+      "loss": 0.7832,
+      "step": 6861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.438335278765798e-05,
+      "loss": 0.7759,
+      "step": 6862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.437598195836849e-05,
+      "loss": 21.4387,
+      "step": 6863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.436861043412536e-05,
+      "loss": 0.7234,
+      "step": 6864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.436123821513875e-05,
+      "loss": 21.3997,
+      "step": 6865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.435386530161885e-05,
+      "loss": 20.7929,
+      "step": 6866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.434649169377586e-05,
+      "loss": 0.0616,
+      "step": 6867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.433911739181997e-05,
+      "loss": 19.5294,
+      "step": 6868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.433174239596148e-05,
+      "loss": 18.332,
+      "step": 6869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.43243667064106e-05,
+      "loss": 0.6748,
+      "step": 6870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.431699032337762e-05,
+      "loss": 0.0861,
+      "step": 6871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.430961324707286e-05,
+      "loss": 0.7813,
+      "step": 6872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.43022354777066e-05,
+      "loss": 20.3789,
+      "step": 6873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.429485701548919e-05,
+      "loss": 19.242,
+      "step": 6874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.428747786063103e-05,
+      "loss": 19.0254,
+      "step": 6875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.428009801334245e-05,
+      "loss": 17.806,
+      "step": 6876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.427271747383387e-05,
+      "loss": 17.9592,
+      "step": 6877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.42653362423157e-05,
+      "loss": 19.6103,
+      "step": 6878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.425795431899837e-05,
+      "loss": 0.7301,
+      "step": 6879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.425057170409235e-05,
+      "loss": 18.1057,
+      "step": 6880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.424318839780811e-05,
+      "loss": 18.3233,
+      "step": 6881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.423580440035616e-05,
+      "loss": 0.7916,
+      "step": 6882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.422841971194701e-05,
+      "loss": 18.8285,
+      "step": 6883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.422103433279117e-05,
+      "loss": 18.9498,
+      "step": 6884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.421364826309924e-05,
+      "loss": 20.3079,
+      "step": 6885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.420626150308174e-05,
+      "loss": 0.6656,
+      "step": 6886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.419887405294932e-05,
+      "loss": 0.8074,
+      "step": 6887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.419148591291256e-05,
+      "loss": 20.0606,
+      "step": 6888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.418409708318212e-05,
+      "loss": 0.8192,
+      "step": 6889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.417670756396863e-05,
+      "loss": 19.8183,
+      "step": 6890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416931735548275e-05,
+      "loss": 19.4749,
+      "step": 6891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416192645793524e-05,
+      "loss": 18.4879,
+      "step": 6892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415453487153675e-05,
+      "loss": 0.6679,
+      "step": 6893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.414714259649801e-05,
+      "loss": 0.7164,
+      "step": 6894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.413974963302981e-05,
+      "loss": 19.5639,
+      "step": 6895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.413235598134289e-05,
+      "loss": 19.4287,
+      "step": 6896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.412496164164808e-05,
+      "loss": 17.5826,
+      "step": 6897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.411756661415614e-05,
+      "loss": 0.7515,
+      "step": 6898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.411017089907795e-05,
+      "loss": 19.7237,
+      "step": 6899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.410277449662432e-05,
+      "loss": 0.7695,
+      "step": 6900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.409537740700613e-05,
+      "loss": 0.7016,
+      "step": 6901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.40879796304343e-05,
+      "loss": 0.0606,
+      "step": 6902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.40805811671197e-05,
+      "loss": 0.0742,
+      "step": 6903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407318201727328e-05,
+      "loss": 19.4697,
+      "step": 6904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.406578218110598e-05,
+      "loss": 0.7189,
+      "step": 6905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405838165882877e-05,
+      "loss": 19.7148,
+      "step": 6906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405098045065263e-05,
+      "loss": 20.681,
+      "step": 6907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.404357855678859e-05,
+      "loss": 18.933,
+      "step": 6908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.403617597744763e-05,
+      "loss": 19.3655,
+      "step": 6909
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.402877271284085e-05,
+      "loss": 20.5986,
+      "step": 6910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.402136876317928e-05,
+      "loss": 19.4583,
+      "step": 6911
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401396412867402e-05,
+      "loss": 0.7339,
+      "step": 6912
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.400655880953617e-05,
+      "loss": 0.7084,
+      "step": 6913
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.399915280597685e-05,
+      "loss": 18.6296,
+      "step": 6914
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.39917461182072e-05,
+      "loss": 20.4695,
+      "step": 6915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.398433874643841e-05,
+      "loss": 19.5895,
+      "step": 6916
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.397693069088163e-05,
+      "loss": 0.7586,
+      "step": 6917
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.39695219517481e-05,
+      "loss": 20.3813,
+      "step": 6918
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.396211252924897e-05,
+      "loss": 0.672,
+      "step": 6919
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.395470242359554e-05,
+      "loss": 18.7976,
+      "step": 6920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.394729163499908e-05,
+      "loss": 20.6751,
+      "step": 6921
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.393988016367084e-05,
+      "loss": 21.0895,
+      "step": 6922
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.393246800982212e-05,
+      "loss": 0.0651,
+      "step": 6923
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.392505517366424e-05,
+      "loss": 19.9627,
+      "step": 6924
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.391764165540857e-05,
+      "loss": 0.734,
+      "step": 6925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.391022745526642e-05,
+      "loss": 20.6204,
+      "step": 6926
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.390281257344919e-05,
+      "loss": 21.6957,
+      "step": 6927
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.389539701016828e-05,
+      "loss": 20.0481,
+      "step": 6928
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.38879807656351e-05,
+      "loss": 20.5402,
+      "step": 6929
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388056384006108e-05,
+      "loss": 0.7658,
+      "step": 6930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.387314623365769e-05,
+      "loss": 18.2128,
+      "step": 6931
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.38657279466364e-05,
+      "loss": 21.5127,
+      "step": 6932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.385830897920868e-05,
+      "loss": 19.5571,
+      "step": 6933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.385088933158608e-05,
+      "loss": 19.6609,
+      "step": 6934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.384346900398012e-05,
+      "loss": 17.3748,
+      "step": 6935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.383604799660233e-05,
+      "loss": 0.7717,
+      "step": 6936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.38286263096643e-05,
+      "loss": 19.396,
+      "step": 6937
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.382120394337763e-05,
+      "loss": 20.8739,
+      "step": 6938
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.38137808979539e-05,
+      "loss": 0.6851,
+      "step": 6939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.380635717360478e-05,
+      "loss": 0.7137,
+      "step": 6940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.379893277054188e-05,
+      "loss": 20.5842,
+      "step": 6941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.379150768897689e-05,
+      "loss": 0.7009,
+      "step": 6942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.37840819291215e-05,
+      "loss": 19.6014,
+      "step": 6943
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.377665549118739e-05,
+      "loss": 0.6791,
+      "step": 6944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.376922837538633e-05,
+      "loss": 20.239,
+      "step": 6945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.376180058193e-05,
+      "loss": 0.7511,
+      "step": 6946
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.375437211103025e-05,
+      "loss": 19.4279,
+      "step": 6947
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.374694296289879e-05,
+      "loss": 0.7385,
+      "step": 6948
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.373951313774746e-05,
+      "loss": 0.0971,
+      "step": 6949
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.373208263578808e-05,
+      "loss": 0.7625,
+      "step": 6950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.372465145723248e-05,
+      "loss": 20.3817,
+      "step": 6951
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.371721960229253e-05,
+      "loss": 20.1752,
+      "step": 6952
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.37097870711801e-05,
+      "loss": 0.7075,
+      "step": 6953
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.37023538641071e-05,
+      "loss": 0.7449,
+      "step": 6954
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.369491998128546e-05,
+      "loss": 0.7463,
+      "step": 6955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.368748542292709e-05,
+      "loss": 0.6979,
+      "step": 6956
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.368005018924398e-05,
+      "loss": 20.2435,
+      "step": 6957
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.367261428044808e-05,
+      "loss": 0.752,
+      "step": 6958
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.366517769675139e-05,
+      "loss": 0.7322,
+      "step": 6959
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.365774043836592e-05,
+      "loss": 22.1094,
+      "step": 6960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.365030250550373e-05,
+      "loss": 19.2479,
+      "step": 6961
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.364286389837686e-05,
+      "loss": 17.7967,
+      "step": 6962
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.36354246171974e-05,
+      "loss": 0.7636,
+      "step": 6963
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.36279846621774e-05,
+      "loss": 0.7498,
+      "step": 6964
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3620544033529e-05,
+      "loss": 20.3799,
+      "step": 6965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.361310273146433e-05,
+      "loss": 19.6325,
+      "step": 6966
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.360566075619554e-05,
+      "loss": 19.7838,
+      "step": 6967
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.359821810793478e-05,
+      "loss": 19.2738,
+      "step": 6968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.359077478689426e-05,
+      "loss": 0.0682,
+      "step": 6969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.358333079328619e-05,
+      "loss": 18.808,
+      "step": 6970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.357588612732278e-05,
+      "loss": 0.733,
+      "step": 6971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356844078921628e-05,
+      "loss": 0.7471,
+      "step": 6972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356099477917896e-05,
+      "loss": 18.6024,
+      "step": 6973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.355354809742311e-05,
+      "loss": 0.7185,
+      "step": 6974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.354610074416102e-05,
+      "loss": 0.6914,
+      "step": 6975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.353865271960502e-05,
+      "loss": 17.6618,
+      "step": 6976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.353120402396744e-05,
+      "loss": 19.5187,
+      "step": 6977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.352375465746066e-05,
+      "loss": 0.7365,
+      "step": 6978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.351630462029704e-05,
+      "loss": 21.3283,
+      "step": 6979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3508853912689e-05,
+      "loss": 18.4745,
+      "step": 6980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.350140253484893e-05,
+      "loss": 20.2244,
+      "step": 6981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.34939504869893e-05,
+      "loss": 0.7231,
+      "step": 6982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.348649776932253e-05,
+      "loss": 0.674,
+      "step": 6983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.347904438206113e-05,
+      "loss": 20.6443,
+      "step": 6984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.347159032541757e-05,
+      "loss": 18.8614,
+      "step": 6985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.346413559960438e-05,
+      "loss": 19.2404,
+      "step": 6986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.345668020483408e-05,
+      "loss": 19.5047,
+      "step": 6987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.344922414131923e-05,
+      "loss": 0.7166,
+      "step": 6988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.344176740927239e-05,
+      "loss": 0.7143,
+      "step": 6989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.343431000890615e-05,
+      "loss": 0.7048,
+      "step": 6990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.342685194043314e-05,
+      "loss": 20.2272,
+      "step": 6991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.341939320406597e-05,
+      "loss": 19.9607,
+      "step": 6992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.34119338000173e-05,
+      "loss": 21.838,
+      "step": 6993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.340447372849976e-05,
+      "loss": 22.7396,
+      "step": 6994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.339701298972609e-05,
+      "loss": 20.0902,
+      "step": 6995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.338955158390894e-05,
+      "loss": 0.0568,
+      "step": 6996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.338208951126107e-05,
+      "loss": 17.5091,
+      "step": 6997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.337462677199522e-05,
+      "loss": 0.0588,
+      "step": 6998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.336716336632412e-05,
+      "loss": 0.7054,
+      "step": 6999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.335969929446058e-05,
+      "loss": 19.2259,
+      "step": 7000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.33522345566174e-05,
+      "loss": 19.339,
+      "step": 7001
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.334476915300737e-05,
+      "loss": 18.4146,
+      "step": 7002
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.333730308384335e-05,
+      "loss": 0.7175,
+      "step": 7003
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.332983634933819e-05,
+      "loss": 17.9532,
+      "step": 7004
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.332236894970477e-05,
+      "loss": 20.9281,
+      "step": 7005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.331490088515598e-05,
+      "loss": 0.0513,
+      "step": 7006
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.330743215590474e-05,
+      "loss": 20.1197,
+      "step": 7007
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.329996276216395e-05,
+      "loss": 20.8834,
+      "step": 7008
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.32924927041466e-05,
+      "loss": 19.8754,
+      "step": 7009
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.328502198206565e-05,
+      "loss": 0.7374,
+      "step": 7010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327755059613407e-05,
+      "loss": 19.1962,
+      "step": 7011
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327007854656489e-05,
+      "loss": 20.2099,
+      "step": 7012
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.326260583357111e-05,
+      "loss": 0.0734,
+      "step": 7013
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.325513245736581e-05,
+      "loss": 18.6494,
+      "step": 7014
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.324765841816203e-05,
+      "loss": 18.8668,
+      "step": 7015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.324018371617286e-05,
+      "loss": 0.7471,
+      "step": 7016
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.323270835161141e-05,
+      "loss": 0.0805,
+      "step": 7017
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.322523232469078e-05,
+      "loss": 0.7181,
+      "step": 7018
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.321775563562411e-05,
+      "loss": 19.4353,
+      "step": 7019
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.321027828462462e-05,
+      "loss": 0.7131,
+      "step": 7020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.320280027190538e-05,
+      "loss": 20.1336,
+      "step": 7021
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.319532159767967e-05,
+      "loss": 20.6525,
+      "step": 7022
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.31878422621607e-05,
+      "loss": 0.7319,
+      "step": 7023
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.318036226556167e-05,
+      "loss": 20.4153,
+      "step": 7024
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.317288160809584e-05,
+      "loss": 0.7127,
+      "step": 7025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.316540028997648e-05,
+      "loss": 20.5555,
+      "step": 7026
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315791831141689e-05,
+      "loss": 19.7914,
+      "step": 7027
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315043567263041e-05,
+      "loss": 19.6419,
+      "step": 7028
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.314295237383032e-05,
+      "loss": 20.7105,
+      "step": 7029
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.313546841522998e-05,
+      "loss": 0.7582,
+      "step": 7030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.312798379704275e-05,
+      "loss": 0.7625,
+      "step": 7031
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.312049851948205e-05,
+      "loss": 21.4027,
+      "step": 7032
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.311301258276124e-05,
+      "loss": 21.1401,
+      "step": 7033
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.310552598709378e-05,
+      "loss": 20.2293,
+      "step": 7034
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.309803873269308e-05,
+      "loss": 0.7256,
+      "step": 7035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.309055081977261e-05,
+      "loss": 19.3253,
+      "step": 7036
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.308306224854584e-05,
+      "loss": 18.3006,
+      "step": 7037
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.30755730192263e-05,
+      "loss": 20.3192,
+      "step": 7038
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.306808313202748e-05,
+      "loss": 0.7554,
+      "step": 7039
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.306059258716292e-05,
+      "loss": 0.0792,
+      "step": 7040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.305310138484616e-05,
+      "loss": 0.7269,
+      "step": 7041
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.30456095252908e-05,
+      "loss": 0.7524,
+      "step": 7042
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.30381170087104e-05,
+      "loss": 20.0745,
+      "step": 7043
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.30306238353186e-05,
+      "loss": 21.2626,
+      "step": 7044
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.302313000532901e-05,
+      "loss": 19.1336,
+      "step": 7045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.30156355189553e-05,
+      "loss": 17.9163,
+      "step": 7046
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.30081403764111e-05,
+      "loss": 18.3631,
+      "step": 7047
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.300064457791012e-05,
+      "loss": 0.0629,
+      "step": 7048
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.299314812366604e-05,
+      "loss": 21.8807,
+      "step": 7049
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.298565101389262e-05,
+      "loss": 0.722,
+      "step": 7050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.297815324880359e-05,
+      "loss": 18.5831,
+      "step": 7051
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.297065482861267e-05,
+      "loss": 0.7429,
+      "step": 7052
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.29631557535337e-05,
+      "loss": 19.5432,
+      "step": 7053
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.295565602378042e-05,
+      "loss": 0.7857,
+      "step": 7054
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.294815563956667e-05,
+      "loss": 0.7516,
+      "step": 7055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.29406546011063e-05,
+      "loss": 19.0928,
+      "step": 7056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.293315290861313e-05,
+      "loss": 20.4776,
+      "step": 7057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.292565056230108e-05,
+      "loss": 19.4764,
+      "step": 7058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.291814756238398e-05,
+      "loss": 20.3979,
+      "step": 7059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.291064390907576e-05,
+      "loss": 18.323,
+      "step": 7060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.290313960259038e-05,
+      "loss": 19.078,
+      "step": 7061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.289563464314175e-05,
+      "loss": 0.7765,
+      "step": 7062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.288812903094386e-05,
+      "loss": 20.1958,
+      "step": 7063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.288062276621066e-05,
+      "loss": 0.1099,
+      "step": 7064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.287311584915619e-05,
+      "loss": 19.9853,
+      "step": 7065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.286560827999444e-05,
+      "loss": 19.4126,
+      "step": 7066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.285810005893947e-05,
+      "loss": 0.0776,
+      "step": 7067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.285059118620532e-05,
+      "loss": 0.7565,
+      "step": 7068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.284308166200608e-05,
+      "loss": 0.761,
+      "step": 7069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.283557148655584e-05,
+      "loss": 18.9909,
+      "step": 7070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.282806066006871e-05,
+      "loss": 19.495,
+      "step": 7071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.282054918275881e-05,
+      "loss": 20.4708,
+      "step": 7072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.281303705484033e-05,
+      "loss": 0.6905,
+      "step": 7073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.28055242765274e-05,
+      "loss": 18.3269,
+      "step": 7074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.279801084803424e-05,
+      "loss": 20.3813,
+      "step": 7075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2790496769575e-05,
+      "loss": 20.9044,
+      "step": 7076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.278298204136398e-05,
+      "loss": 0.0737,
+      "step": 7077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.277546666361536e-05,
+      "loss": 0.8015,
+      "step": 7078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.276795063654345e-05,
+      "loss": 22.5031,
+      "step": 7079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27604339603625e-05,
+      "loss": 0.7324,
+      "step": 7080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.275291663528682e-05,
+      "loss": 0.7236,
+      "step": 7081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.274539866153072e-05,
+      "loss": 19.3,
+      "step": 7082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.273788003930852e-05,
+      "loss": 18.5525,
+      "step": 7083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.273036076883461e-05,
+      "loss": 0.7828,
+      "step": 7084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.272284085032336e-05,
+      "loss": 0.7353,
+      "step": 7085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.271532028398913e-05,
+      "loss": 0.6813,
+      "step": 7086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.270779907004637e-05,
+      "loss": 19.3577,
+      "step": 7087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.270027720870946e-05,
+      "loss": 0.785,
+      "step": 7088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269275470019287e-05,
+      "loss": 19.4108,
+      "step": 7089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.268523154471108e-05,
+      "loss": 19.1127,
+      "step": 7090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.267770774247855e-05,
+      "loss": 0.0737,
+      "step": 7091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.267018329370981e-05,
+      "loss": 0.6928,
+      "step": 7092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.266265819861935e-05,
+      "loss": 0.7483,
+      "step": 7093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.265513245742173e-05,
+      "loss": 0.7148,
+      "step": 7094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264760607033148e-05,
+      "loss": 0.7676,
+      "step": 7095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.26400790375632e-05,
+      "loss": 0.6452,
+      "step": 7096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.263255135933147e-05,
+      "loss": 19.7591,
+      "step": 7097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.262502303585093e-05,
+      "loss": 0.7804,
+      "step": 7098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.261749406733618e-05,
+      "loss": 19.8747,
+      "step": 7099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.260996445400189e-05,
+      "loss": 19.4473,
+      "step": 7100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.26024341960627e-05,
+      "loss": 19.2805,
+      "step": 7101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.259490329373333e-05,
+      "loss": 0.786,
+      "step": 7102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.258737174722845e-05,
+      "loss": 19.087,
+      "step": 7103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.257983955676281e-05,
+      "loss": 19.0974,
+      "step": 7104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.257230672255115e-05,
+      "loss": 17.9162,
+      "step": 7105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.256477324480821e-05,
+      "loss": 17.9445,
+      "step": 7106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.255723912374879e-05,
+      "loss": 20.38,
+      "step": 7107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.254970435958765e-05,
+      "loss": 0.7707,
+      "step": 7108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.254216895253966e-05,
+      "loss": 0.7156,
+      "step": 7109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.25346329028196e-05,
+      "loss": 0.0766,
+      "step": 7110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.252709621064235e-05,
+      "loss": 20.5551,
+      "step": 7111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251955887622277e-05,
+      "loss": 18.1827,
+      "step": 7112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251202089977575e-05,
+      "loss": 0.0655,
+      "step": 7113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.25044822815162e-05,
+      "loss": 20.4852,
+      "step": 7114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.249694302165904e-05,
+      "loss": 17.9504,
+      "step": 7115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.248940312041921e-05,
+      "loss": 19.4199,
+      "step": 7116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.248186257801169e-05,
+      "loss": 0.785,
+      "step": 7117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.247432139465141e-05,
+      "loss": 0.0747,
+      "step": 7118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.246677957055342e-05,
+      "loss": 0.7261,
+      "step": 7119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.245923710593273e-05,
+      "loss": 20.8123,
+      "step": 7120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.245169400100433e-05,
+      "loss": 19.2469,
+      "step": 7121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.244415025598331e-05,
+      "loss": 18.8938,
+      "step": 7122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.243660587108474e-05,
+      "loss": 18.6985,
+      "step": 7123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.242906084652368e-05,
+      "loss": 17.312,
+      "step": 7124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.242151518251528e-05,
+      "loss": 18.2199,
+      "step": 7125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.241396887927465e-05,
+      "loss": 20.1954,
+      "step": 7126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.24064219370169e-05,
+      "loss": 20.0013,
+      "step": 7127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.239887435595723e-05,
+      "loss": 20.9817,
+      "step": 7128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.239132613631081e-05,
+      "loss": 19.1361,
+      "step": 7129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.238377727829282e-05,
+      "loss": 0.723,
+      "step": 7130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.237622778211852e-05,
+      "loss": 19.7753,
+      "step": 7131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23686776480031e-05,
+      "loss": 19.8985,
+      "step": 7132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.236112687616183e-05,
+      "loss": 19.8114,
+      "step": 7133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.235357546680998e-05,
+      "loss": 0.0691,
+      "step": 7134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.234602342016285e-05,
+      "loss": 0.0667,
+      "step": 7135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.233847073643571e-05,
+      "loss": 18.8657,
+      "step": 7136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.233091741584393e-05,
+      "loss": 20.3406,
+      "step": 7137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.232336345860282e-05,
+      "loss": 0.7917,
+      "step": 7138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.231580886492777e-05,
+      "loss": 19.8942,
+      "step": 7139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.230825363503414e-05,
+      "loss": 19.2191,
+      "step": 7140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.230069776913733e-05,
+      "loss": 19.7399,
+      "step": 7141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.229314126745277e-05,
+      "loss": 18.843,
+      "step": 7142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.228558413019588e-05,
+      "loss": 0.7602,
+      "step": 7143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.227802635758211e-05,
+      "loss": 20.2744,
+      "step": 7144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.227046794982694e-05,
+      "loss": 0.7776,
+      "step": 7145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.226290890714586e-05,
+      "loss": 18.8263,
+      "step": 7146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.225534922975436e-05,
+      "loss": 19.5298,
+      "step": 7147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.224778891786799e-05,
+      "loss": 0.0794,
+      "step": 7148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.224022797170227e-05,
+      "loss": 0.0952,
+      "step": 7149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.223266639147277e-05,
+      "loss": 20.719,
+      "step": 7150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.222510417739507e-05,
+      "loss": 19.2621,
+      "step": 7151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.221754132968476e-05,
+      "loss": 0.7078,
+      "step": 7152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.220997784855747e-05,
+      "loss": 19.7955,
+      "step": 7153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.22024137342288e-05,
+      "loss": 20.7831,
+      "step": 7154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.219484898691446e-05,
+      "loss": 19.8766,
+      "step": 7155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.218728360683006e-05,
+      "loss": 20.1421,
+      "step": 7156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.217971759419132e-05,
+      "loss": 0.7792,
+      "step": 7157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.217215094921393e-05,
+      "loss": 0.6837,
+      "step": 7158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.216458367211363e-05,
+      "loss": 0.08,
+      "step": 7159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.215701576310615e-05,
+      "loss": 19.459,
+      "step": 7160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.214944722240724e-05,
+      "loss": 19.6491,
+      "step": 7161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.214187805023269e-05,
+      "loss": 19.6523,
+      "step": 7162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.213430824679831e-05,
+      "loss": 19.1078,
+      "step": 7163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.212673781231987e-05,
+      "loss": 19.2952,
+      "step": 7164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.211916674701326e-05,
+      "loss": 17.0923,
+      "step": 7165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.211159505109427e-05,
+      "loss": 0.8126,
+      "step": 7166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.21040227247788e-05,
+      "loss": 19.044,
+      "step": 7167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.209644976828274e-05,
+      "loss": 0.7188,
+      "step": 7168
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.208887618182198e-05,
+      "loss": 0.7147,
+      "step": 7169
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.208130196561243e-05,
+      "loss": 19.0946,
+      "step": 7170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.207372711987005e-05,
+      "loss": 20.4045,
+      "step": 7171
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.206615164481081e-05,
+      "loss": 19.1256,
+      "step": 7172
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.205857554065064e-05,
+      "loss": 0.0807,
+      "step": 7173
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.205099880760557e-05,
+      "loss": 18.3055,
+      "step": 7174
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.20434214458916e-05,
+      "loss": 0.7953,
+      "step": 7175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.203584345572476e-05,
+      "loss": 19.3895,
+      "step": 7176
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.202826483732108e-05,
+      "loss": 0.0785,
+      "step": 7177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.202068559089665e-05,
+      "loss": 0.7303,
+      "step": 7178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.201310571666755e-05,
+      "loss": 0.7446,
+      "step": 7179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.200552521484985e-05,
+      "loss": 0.7359,
+      "step": 7180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.199794408565971e-05,
+      "loss": 19.5583,
+      "step": 7181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.199036232931326e-05,
+      "loss": 0.0826,
+      "step": 7182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.198277994602662e-05,
+      "loss": 0.7753,
+      "step": 7183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.197519693601599e-05,
+      "loss": 0.7473,
+      "step": 7184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.196761329949756e-05,
+      "loss": 21.9574,
+      "step": 7185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.196002903668752e-05,
+      "loss": 0.7159,
+      "step": 7186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.19524441478021e-05,
+      "loss": 19.4095,
+      "step": 7187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.194485863305756e-05,
+      "loss": 0.6997,
+      "step": 7188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.193727249267015e-05,
+      "loss": 0.8095,
+      "step": 7189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.192968572685615e-05,
+      "loss": 20.4167,
+      "step": 7190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.192209833583186e-05,
+      "loss": 18.8027,
+      "step": 7191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.19145103198136e-05,
+      "loss": 18.0582,
+      "step": 7192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.190692167901769e-05,
+      "loss": 20.7341,
+      "step": 7193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189933241366046e-05,
+      "loss": 18.8028,
+      "step": 7194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189174252395834e-05,
+      "loss": 0.7285,
+      "step": 7195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.188415201012768e-05,
+      "loss": 0.7204,
+      "step": 7196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.187656087238486e-05,
+      "loss": 0.0865,
+      "step": 7197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.186896911094633e-05,
+      "loss": 0.7284,
+      "step": 7198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.186137672602853e-05,
+      "loss": 19.0681,
+      "step": 7199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18537837178479e-05,
+      "loss": 0.7613,
+      "step": 7200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.184619008662094e-05,
+      "loss": 22.4824,
+      "step": 7201
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.183859583256411e-05,
+      "loss": 0.6354,
+      "step": 7202
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.183100095589397e-05,
+      "loss": 17.3985,
+      "step": 7203
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1823405456827e-05,
+      "loss": 18.3345,
+      "step": 7204
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.181580933557976e-05,
+      "loss": 18.9915,
+      "step": 7205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.180821259236883e-05,
+      "loss": 18.8077,
+      "step": 7206
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.180061522741077e-05,
+      "loss": 18.5012,
+      "step": 7207
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.179301724092219e-05,
+      "loss": 19.9875,
+      "step": 7208
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.17854186331197e-05,
+      "loss": 19.0567,
+      "step": 7209
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.177781940421995e-05,
+      "loss": 0.743,
+      "step": 7210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.177021955443957e-05,
+      "loss": 0.0949,
+      "step": 7211
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.176261908399523e-05,
+      "loss": 19.322,
+      "step": 7212
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.175501799310366e-05,
+      "loss": 18.7961,
+      "step": 7213
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.17474162819815e-05,
+      "loss": 0.7811,
+      "step": 7214
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.173981395084554e-05,
+      "loss": 0.7362,
+      "step": 7215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.173221099991246e-05,
+      "loss": 19.0646,
+      "step": 7216
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.172460742939906e-05,
+      "loss": 0.7115,
+      "step": 7217
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.171700323952211e-05,
+      "loss": 0.7432,
+      "step": 7218
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.170939843049838e-05,
+      "loss": 20.5169,
+      "step": 7219
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.170179300254472e-05,
+      "loss": 19.2469,
+      "step": 7220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.169418695587791e-05,
+      "loss": 20.0709,
+      "step": 7221
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.168658029071484e-05,
+      "loss": 19.8266,
+      "step": 7222
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167897300727236e-05,
+      "loss": 18.7122,
+      "step": 7223
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167136510576733e-05,
+      "loss": 19.0136,
+      "step": 7224
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.166375658641668e-05,
+      "loss": 19.1751,
+      "step": 7225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.165614744943731e-05,
+      "loss": 19.5473,
+      "step": 7226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.164853769504618e-05,
+      "loss": 20.0521,
+      "step": 7227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.164092732346022e-05,
+      "loss": 0.7356,
+      "step": 7228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.163331633489638e-05,
+      "loss": 19.2954,
+      "step": 7229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.162570472957169e-05,
+      "loss": 0.0734,
+      "step": 7230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.161809250770312e-05,
+      "loss": 19.9,
+      "step": 7231
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.161047966950774e-05,
+      "loss": 0.078,
+      "step": 7232
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.160286621520252e-05,
+      "loss": 0.6887,
+      "step": 7233
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.159525214500459e-05,
+      "loss": 0.7503,
+      "step": 7234
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.158763745913098e-05,
+      "loss": 0.7274,
+      "step": 7235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.158002215779878e-05,
+      "loss": 0.6874,
+      "step": 7236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.157240624122513e-05,
+      "loss": 21.3889,
+      "step": 7237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.156478970962714e-05,
+      "loss": 19.8741,
+      "step": 7238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.155717256322196e-05,
+      "loss": 21.4857,
+      "step": 7239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.154955480222675e-05,
+      "loss": 20.8207,
+      "step": 7240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.154193642685867e-05,
+      "loss": 20.3695,
+      "step": 7241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.153431743733499e-05,
+      "loss": 19.2312,
+      "step": 7242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.152669783387285e-05,
+      "loss": 0.0702,
+      "step": 7243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.15190776166895e-05,
+      "loss": 0.7918,
+      "step": 7244
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.151145678600221e-05,
+      "loss": 19.794,
+      "step": 7245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.150383534202825e-05,
+      "loss": 18.5461,
+      "step": 7246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.149621328498488e-05,
+      "loss": 0.7597,
+      "step": 7247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.148859061508942e-05,
+      "loss": 17.9615,
+      "step": 7248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.148096733255917e-05,
+      "loss": 0.779,
+      "step": 7249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.147334343761152e-05,
+      "loss": 0.7595,
+      "step": 7250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.146571893046377e-05,
+      "loss": 0.7214,
+      "step": 7251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.145809381133333e-05,
+      "loss": 17.4856,
+      "step": 7252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.145046808043754e-05,
+      "loss": 17.0421,
+      "step": 7253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.144284173799388e-05,
+      "loss": 0.7299,
+      "step": 7254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.143521478421972e-05,
+      "loss": 20.0169,
+      "step": 7255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.142758721933253e-05,
+      "loss": 0.7426,
+      "step": 7256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.141995904354975e-05,
+      "loss": 18.6205,
+      "step": 7257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.141233025708886e-05,
+      "loss": 19.2654,
+      "step": 7258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.140470086016738e-05,
+      "loss": 19.7089,
+      "step": 7259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.139707085300281e-05,
+      "loss": 17.8925,
+      "step": 7260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.138944023581265e-05,
+      "loss": 19.4048,
+      "step": 7261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.138180900881449e-05,
+      "loss": 0.7281,
+      "step": 7262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.137417717222588e-05,
+      "loss": 0.723,
+      "step": 7263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.136654472626439e-05,
+      "loss": 0.7393,
+      "step": 7264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.135891167114761e-05,
+      "loss": 19.7856,
+      "step": 7265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.13512780070932e-05,
+      "loss": 18.7105,
+      "step": 7266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.134364373431876e-05,
+      "loss": 18.9165,
+      "step": 7267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.133600885304195e-05,
+      "loss": 0.691,
+      "step": 7268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.132837336348044e-05,
+      "loss": 18.6766,
+      "step": 7269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.132073726585191e-05,
+      "loss": 20.864,
+      "step": 7270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.131310056037407e-05,
+      "loss": 20.0346,
+      "step": 7271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.130546324726465e-05,
+      "loss": 0.777,
+      "step": 7272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.129782532674136e-05,
+      "loss": 19.2493,
+      "step": 7273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.129018679902199e-05,
+      "loss": 0.6905,
+      "step": 7274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.128254766432429e-05,
+      "loss": 0.8462,
+      "step": 7275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.127490792286604e-05,
+      "loss": 0.7592,
+      "step": 7276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.126726757486506e-05,
+      "loss": 17.4455,
+      "step": 7277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12596266205392e-05,
+      "loss": 0.7169,
+      "step": 7278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.125198506010626e-05,
+      "loss": 0.7074,
+      "step": 7279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.124434289378412e-05,
+      "loss": 19.2883,
+      "step": 7280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.123670012179067e-05,
+      "loss": 18.5965,
+      "step": 7281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.122905674434375e-05,
+      "loss": 20.5548,
+      "step": 7282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.122141276166135e-05,
+      "loss": 0.7958,
+      "step": 7283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.121376817396134e-05,
+      "loss": 19.2925,
+      "step": 7284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12061229814617e-05,
+      "loss": 0.0743,
+      "step": 7285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.119847718438033e-05,
+      "loss": 0.0652,
+      "step": 7286
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.119083078293529e-05,
+      "loss": 0.7523,
+      "step": 7287
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.118318377734452e-05,
+      "loss": 0.6714,
+      "step": 7288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.117553616782608e-05,
+      "loss": 0.7648,
+      "step": 7289
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.116788795459797e-05,
+      "loss": 19.319,
+      "step": 7290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.116023913787823e-05,
+      "loss": 0.6919,
+      "step": 7291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.115258971788497e-05,
+      "loss": 0.7157,
+      "step": 7292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.114493969483623e-05,
+      "loss": 18.3487,
+      "step": 7293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.113728906895013e-05,
+      "loss": 0.7511,
+      "step": 7294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.112963784044478e-05,
+      "loss": 18.647,
+      "step": 7295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.112198600953833e-05,
+      "loss": 0.818,
+      "step": 7296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11143335764489e-05,
+      "loss": 0.748,
+      "step": 7297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11066805413947e-05,
+      "loss": 0.7926,
+      "step": 7298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.109902690459389e-05,
+      "loss": 0.6908,
+      "step": 7299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.109137266626468e-05,
+      "loss": 0.0588,
+      "step": 7300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.108371782662527e-05,
+      "loss": 0.7121,
+      "step": 7301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.107606238589394e-05,
+      "loss": 20.6565,
+      "step": 7302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.106840634428892e-05,
+      "loss": 19.8639,
+      "step": 7303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.106074970202849e-05,
+      "loss": 20.9366,
+      "step": 7304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.105309245933092e-05,
+      "loss": 19.2893,
+      "step": 7305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.104543461641453e-05,
+      "loss": 20.5288,
+      "step": 7306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.103777617349764e-05,
+      "loss": 0.6592,
+      "step": 7307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.103011713079861e-05,
+      "loss": 20.7929,
+      "step": 7308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.102245748853578e-05,
+      "loss": 0.786,
+      "step": 7309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.10147972469275e-05,
+      "loss": 0.7799,
+      "step": 7310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.10071364061922e-05,
+      "loss": 20.0336,
+      "step": 7311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.099947496654829e-05,
+      "loss": 18.211,
+      "step": 7312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.099181292821418e-05,
+      "loss": 21.0734,
+      "step": 7313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09841502914083e-05,
+      "loss": 18.8603,
+      "step": 7314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.097648705634913e-05,
+      "loss": 18.6332,
+      "step": 7315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096882322325514e-05,
+      "loss": 0.7382,
+      "step": 7316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096115879234483e-05,
+      "loss": 21.2734,
+      "step": 7317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.095349376383671e-05,
+      "loss": 17.1479,
+      "step": 7318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.094582813794931e-05,
+      "loss": 0.7214,
+      "step": 7319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.093816191490117e-05,
+      "loss": 19.6893,
+      "step": 7320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.093049509491086e-05,
+      "loss": 0.8094,
+      "step": 7321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.092282767819692e-05,
+      "loss": 0.7071,
+      "step": 7322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0915159664978e-05,
+      "loss": 18.077,
+      "step": 7323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09074910554727e-05,
+      "loss": 0.7381,
+      "step": 7324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.089982184989963e-05,
+      "loss": 20.8154,
+      "step": 7325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.089215204847746e-05,
+      "loss": 19.5758,
+      "step": 7326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.088448165142483e-05,
+      "loss": 0.7017,
+      "step": 7327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.087681065896044e-05,
+      "loss": 21.562,
+      "step": 7328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.086913907130299e-05,
+      "loss": 0.0688,
+      "step": 7329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.086146688867118e-05,
+      "loss": 0.7673,
+      "step": 7330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.085379411128376e-05,
+      "loss": 17.9881,
+      "step": 7331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.084612073935946e-05,
+      "loss": 19.2825,
+      "step": 7332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.083844677311708e-05,
+      "loss": 19.3111,
+      "step": 7333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.083077221277534e-05,
+      "loss": 0.0783,
+      "step": 7334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.08230970585531e-05,
+      "loss": 19.0171,
+      "step": 7335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.081542131066916e-05,
+      "loss": 17.8903,
+      "step": 7336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.080774496934233e-05,
+      "loss": 19.7631,
+      "step": 7337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.08000680347915e-05,
+      "loss": 18.2617,
+      "step": 7338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.079239050723551e-05,
+      "loss": 0.7083,
+      "step": 7339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.078471238689326e-05,
+      "loss": 0.7172,
+      "step": 7340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.077703367398364e-05,
+      "loss": 0.7812,
+      "step": 7341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.076935436872557e-05,
+      "loss": 18.4399,
+      "step": 7342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.076167447133799e-05,
+      "loss": 18.1851,
+      "step": 7343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.075399398203984e-05,
+      "loss": 17.3893,
+      "step": 7344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.074631290105011e-05,
+      "loss": 18.5642,
+      "step": 7345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073863122858778e-05,
+      "loss": 17.911,
+      "step": 7346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073094896487182e-05,
+      "loss": 17.6676,
+      "step": 7347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.07232661101213e-05,
+      "loss": 20.1251,
+      "step": 7348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.071558266455523e-05,
+      "loss": 0.7177,
+      "step": 7349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.070789862839267e-05,
+      "loss": 0.0652,
+      "step": 7350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.070021400185267e-05,
+      "loss": 20.711,
+      "step": 7351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.069252878515434e-05,
+      "loss": 0.7119,
+      "step": 7352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.068484297851679e-05,
+      "loss": 21.6688,
+      "step": 7353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.06771565821591e-05,
+      "loss": 0.7407,
+      "step": 7354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.066946959630047e-05,
+      "loss": 0.0581,
+      "step": 7355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.066178202116e-05,
+      "loss": 0.0843,
+      "step": 7356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.065409385695687e-05,
+      "loss": 19.569,
+      "step": 7357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.06464051039103e-05,
+      "loss": 0.7517,
+      "step": 7358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063871576223946e-05,
+      "loss": 19.1987,
+      "step": 7359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063102583216359e-05,
+      "loss": 19.2577,
+      "step": 7360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.062333531390192e-05,
+      "loss": 19.5445,
+      "step": 7361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.06156442076737e-05,
+      "loss": 20.8757,
+      "step": 7362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.060795251369822e-05,
+      "loss": 0.7548,
+      "step": 7363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.060026023219474e-05,
+      "loss": 19.9781,
+      "step": 7364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05925673633826e-05,
+      "loss": 0.6998,
+      "step": 7365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05848739074811e-05,
+      "loss": 0.7082,
+      "step": 7366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.057717986470959e-05,
+      "loss": 20.2768,
+      "step": 7367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05694852352874e-05,
+      "loss": 0.0674,
+      "step": 7368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.056179001943392e-05,
+      "loss": 19.2915,
+      "step": 7369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.055409421736856e-05,
+      "loss": 0.658,
+      "step": 7370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05463978293107e-05,
+      "loss": 0.7536,
+      "step": 7371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053870085547975e-05,
+      "loss": 17.8009,
+      "step": 7372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053100329609519e-05,
+      "loss": 18.5149,
+      "step": 7373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.052330515137645e-05,
+      "loss": 20.1962,
+      "step": 7374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.051560642154298e-05,
+      "loss": 18.7714,
+      "step": 7375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05079071068143e-05,
+      "loss": 20.5754,
+      "step": 7376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.050020720740991e-05,
+      "loss": 0.0806,
+      "step": 7377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.049250672354934e-05,
+      "loss": 19.6209,
+      "step": 7378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.048480565545213e-05,
+      "loss": 0.7236,
+      "step": 7379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.04771040033378e-05,
+      "loss": 0.0722,
+      "step": 7380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.046940176742596e-05,
+      "loss": 0.0657,
+      "step": 7381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.046169894793618e-05,
+      "loss": 18.7603,
+      "step": 7382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.04539955450881e-05,
+      "loss": 0.793,
+      "step": 7383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.044629155910128e-05,
+      "loss": 21.1653,
+      "step": 7384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.043858699019542e-05,
+      "loss": 18.8072,
+      "step": 7385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.043088183859012e-05,
+      "loss": 16.8563,
+      "step": 7386
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.042317610450511e-05,
+      "loss": 17.6986,
+      "step": 7387
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.041546978816004e-05,
+      "loss": 18.8764,
+      "step": 7388
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.040776288977462e-05,
+      "loss": 0.7492,
+      "step": 7389
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.040005540956858e-05,
+      "loss": 0.6885,
+      "step": 7390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.039234734776166e-05,
+      "loss": 19.7789,
+      "step": 7391
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.038463870457361e-05,
+      "loss": 19.5242,
+      "step": 7392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.037692948022418e-05,
+      "loss": 19.2788,
+      "step": 7393
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.036921967493322e-05,
+      "loss": 0.7283,
+      "step": 7394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.036150928892047e-05,
+      "loss": 20.3789,
+      "step": 7395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035379832240579e-05,
+      "loss": 17.6477,
+      "step": 7396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.034608677560899e-05,
+      "loss": 20.0412,
+      "step": 7397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.033837464874995e-05,
+      "loss": 0.05,
+      "step": 7398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.033066194204852e-05,
+      "loss": 19.989,
+      "step": 7399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.032294865572461e-05,
+      "loss": 17.2128,
+      "step": 7400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.03152347899981e-05,
+      "loss": 0.8088,
+      "step": 7401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.030752034508893e-05,
+      "loss": 0.7269,
+      "step": 7402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.029980532121703e-05,
+      "loss": 20.8661,
+      "step": 7403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.029208971860235e-05,
+      "loss": 19.5831,
+      "step": 7404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.028437353746486e-05,
+      "loss": 0.7158,
+      "step": 7405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.027665677802455e-05,
+      "loss": 20.3614,
+      "step": 7406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.026893944050142e-05,
+      "loss": 17.2481,
+      "step": 7407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.026122152511551e-05,
+      "loss": 19.2699,
+      "step": 7408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02535030320868e-05,
+      "loss": 19.8125,
+      "step": 7409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.024578396163541e-05,
+      "loss": 0.1004,
+      "step": 7410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.023806431398137e-05,
+      "loss": 17.3723,
+      "step": 7411
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02303440893448e-05,
+      "loss": 0.7769,
+      "step": 7412
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.022262328794574e-05,
+      "loss": 0.6992,
+      "step": 7413
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.021490191000437e-05,
+      "loss": 19.9005,
+      "step": 7414
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.020717995574078e-05,
+      "loss": 20.6227,
+      "step": 7415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.019945742537515e-05,
+      "loss": 20.3314,
+      "step": 7416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.019173431912763e-05,
+      "loss": 0.0505,
+      "step": 7417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.018401063721842e-05,
+      "loss": 22.6468,
+      "step": 7418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.017628637986771e-05,
+      "loss": 17.7497,
+      "step": 7419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.016856154729573e-05,
+      "loss": 18.6383,
+      "step": 7420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.016083613972269e-05,
+      "loss": 0.7616,
+      "step": 7421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.015311015736885e-05,
+      "loss": 0.7329,
+      "step": 7422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.014538360045449e-05,
+      "loss": 19.2247,
+      "step": 7423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.013765646919986e-05,
+      "loss": 17.5679,
+      "step": 7424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.012992876382531e-05,
+      "loss": 17.177,
+      "step": 7425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.01222004845511e-05,
+      "loss": 17.9129,
+      "step": 7426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.011447163159759e-05,
+      "loss": 0.745,
+      "step": 7427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.010674220518512e-05,
+      "loss": 0.0752,
+      "step": 7428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.009901220553408e-05,
+      "loss": 0.7532,
+      "step": 7429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.00912816328648e-05,
+      "loss": 0.6768,
+      "step": 7430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.008355048739772e-05,
+      "loss": 19.491,
+      "step": 7431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.007581876935323e-05,
+      "loss": 0.6477,
+      "step": 7432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.006808647895179e-05,
+      "loss": 19.1308,
+      "step": 7433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.00603536164138e-05,
+      "loss": 19.8465,
+      "step": 7434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.005262018195976e-05,
+      "loss": 0.7359,
+      "step": 7435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.004488617581012e-05,
+      "loss": 0.72,
+      "step": 7436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.00371515981854e-05,
+      "loss": 19.4353,
+      "step": 7437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.002941644930609e-05,
+      "loss": 0.7004,
+      "step": 7438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.002168072939272e-05,
+      "loss": 18.0702,
+      "step": 7439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.001394443866587e-05,
+      "loss": 0.7183,
+      "step": 7440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.000620757734604e-05,
+      "loss": 0.0887,
+      "step": 7441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.999847014565386e-05,
+      "loss": 0.7175,
+      "step": 7442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.999073214380989e-05,
+      "loss": 18.3966,
+      "step": 7443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.998299357203475e-05,
+      "loss": 20.7175,
+      "step": 7444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.997525443054905e-05,
+      "loss": 0.7589,
+      "step": 7445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.996751471957346e-05,
+      "loss": 18.9757,
+      "step": 7446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.995977443932862e-05,
+      "loss": 0.0666,
+      "step": 7447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.99520335900352e-05,
+      "loss": 19.584,
+      "step": 7448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.99442921719139e-05,
+      "loss": 19.7207,
+      "step": 7449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.99365501851854e-05,
+      "loss": 20.8032,
+      "step": 7450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.992880763007046e-05,
+      "loss": 19.9069,
+      "step": 7451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.99210645067898e-05,
+      "loss": 0.0887,
+      "step": 7452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.991332081556417e-05,
+      "loss": 0.7584,
+      "step": 7453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.990557655661436e-05,
+      "loss": 0.7091,
+      "step": 7454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.989783173016113e-05,
+      "loss": 17.9988,
+      "step": 7455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.989008633642531e-05,
+      "loss": 20.1764,
+      "step": 7456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.988234037562771e-05,
+      "loss": 0.7918,
+      "step": 7457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.987459384798914e-05,
+      "loss": 0.0673,
+      "step": 7458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.98668467537305e-05,
+      "loss": 17.6706,
+      "step": 7459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.985909909307261e-05,
+      "loss": 19.1841,
+      "step": 7460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.985135086623638e-05,
+      "loss": 0.7608,
+      "step": 7461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.984360207344271e-05,
+      "loss": 21.0427,
+      "step": 7462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.983585271491251e-05,
+      "loss": 0.7397,
+      "step": 7463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.982810279086672e-05,
+      "loss": 19.6648,
+      "step": 7464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.982035230152628e-05,
+      "loss": 17.6212,
+      "step": 7465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.981260124711216e-05,
+      "loss": 0.6874,
+      "step": 7466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.980484962784534e-05,
+      "loss": 0.0596,
+      "step": 7467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.979709744394681e-05,
+      "loss": 0.7157,
+      "step": 7468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.97893446956376e-05,
+      "loss": 18.4258,
+      "step": 7469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978159138313871e-05,
+      "loss": 19.9145,
+      "step": 7470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.977383750667122e-05,
+      "loss": 20.1207,
+      "step": 7471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.976608306645616e-05,
+      "loss": 18.172,
+      "step": 7472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.975832806271463e-05,
+      "loss": 19.9935,
+      "step": 7473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.975057249566771e-05,
+      "loss": 18.8705,
+      "step": 7474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.974281636553653e-05,
+      "loss": 0.7462,
+      "step": 7475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.973505967254217e-05,
+      "loss": 18.5857,
+      "step": 7476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.972730241690583e-05,
+      "loss": 0.7233,
+      "step": 7477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.971954459884862e-05,
+      "loss": 19.3147,
+      "step": 7478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.971178621859175e-05,
+      "loss": 0.7401,
+      "step": 7479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.970402727635637e-05,
+      "loss": 19.216,
+      "step": 7480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.969626777236372e-05,
+      "loss": 20.6003,
+      "step": 7481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9688507706835e-05,
+      "loss": 0.7705,
+      "step": 7482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.968074707999147e-05,
+      "loss": 0.8063,
+      "step": 7483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.967298589205436e-05,
+      "loss": 19.2343,
+      "step": 7484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.966522414324496e-05,
+      "loss": 20.3358,
+      "step": 7485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.965746183378454e-05,
+      "loss": 0.6689,
+      "step": 7486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.964969896389441e-05,
+      "loss": 20.3475,
+      "step": 7487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.964193553379589e-05,
+      "loss": 19.3688,
+      "step": 7488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.963417154371029e-05,
+      "loss": 20.0973,
+      "step": 7489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.962640699385901e-05,
+      "loss": 0.7743,
+      "step": 7490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.961864188446336e-05,
+      "loss": 0.6989,
+      "step": 7491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.961087621574474e-05,
+      "loss": 19.6694,
+      "step": 7492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.960310998792456e-05,
+      "loss": 21.666,
+      "step": 7493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.959534320122422e-05,
+      "loss": 20.02,
+      "step": 7494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.958757585586519e-05,
+      "loss": 0.7887,
+      "step": 7495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.957980795206885e-05,
+      "loss": 18.8051,
+      "step": 7496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.957203949005668e-05,
+      "loss": 19.8837,
+      "step": 7497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.956427047005019e-05,
+      "loss": 19.3657,
+      "step": 7498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.955650089227085e-05,
+      "loss": 0.797,
+      "step": 7499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.954873075694018e-05,
+      "loss": 19.5699,
+      "step": 7500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.954096006427968e-05,
+      "loss": 18.9446,
+      "step": 7501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.953318881451091e-05,
+      "loss": 0.7437,
+      "step": 7502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.952541700785543e-05,
+      "loss": 18.2569,
+      "step": 7503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.95176446445348e-05,
+      "loss": 0.0864,
+      "step": 7504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.950987172477062e-05,
+      "loss": 0.7068,
+      "step": 7505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.950209824878447e-05,
+      "loss": 17.5304,
+      "step": 7506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9494324216798e-05,
+      "loss": 20.6315,
+      "step": 7507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.948654962903284e-05,
+      "loss": 19.728,
+      "step": 7508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.947877448571062e-05,
+      "loss": 0.0633,
+      "step": 7509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.947099878705303e-05,
+      "loss": 0.7648,
+      "step": 7510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.946322253328174e-05,
+      "loss": 19.4367,
+      "step": 7511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.945544572461847e-05,
+      "loss": 20.7879,
+      "step": 7512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.944766836128491e-05,
+      "loss": 0.7304,
+      "step": 7513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.943989044350281e-05,
+      "loss": 19.4593,
+      "step": 7514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.943211197149391e-05,
+      "loss": 18.4521,
+      "step": 7515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.942433294547995e-05,
+      "loss": 0.7748,
+      "step": 7516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.941655336568274e-05,
+      "loss": 0.7185,
+      "step": 7517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.940877323232406e-05,
+      "loss": 20.2927,
+      "step": 7518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.940099254562573e-05,
+      "loss": 0.7221,
+      "step": 7519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.939321130580958e-05,
+      "loss": 0.6543,
+      "step": 7520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.938542951309741e-05,
+      "loss": 19.252,
+      "step": 7521
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.937764716771114e-05,
+      "loss": 0.6753,
+      "step": 7522
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.936986426987258e-05,
+      "loss": 20.44,
+      "step": 7523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.936208081980367e-05,
+      "loss": 0.6933,
+      "step": 7524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.935429681772625e-05,
+      "loss": 0.0749,
+      "step": 7525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.934651226386232e-05,
+      "loss": 0.7035,
+      "step": 7526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.933872715843379e-05,
+      "loss": 0.7031,
+      "step": 7527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.933094150166256e-05,
+      "loss": 19.1023,
+      "step": 7528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.932315529377067e-05,
+      "loss": 18.9544,
+      "step": 7529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.931536853498005e-05,
+      "loss": 20.8056,
+      "step": 7530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.930758122551273e-05,
+      "loss": 0.0861,
+      "step": 7531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.929979336559071e-05,
+      "loss": 21.6049,
+      "step": 7532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.929200495543602e-05,
+      "loss": 21.4035,
+      "step": 7533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.928421599527072e-05,
+      "loss": 19.6329,
+      "step": 7534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.927642648531685e-05,
+      "loss": 0.6984,
+      "step": 7535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.92686364257965e-05,
+      "loss": 18.0136,
+      "step": 7536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.926084581693175e-05,
+      "loss": 0.8001,
+      "step": 7537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.925305465894474e-05,
+      "loss": 19.7537,
+      "step": 7538
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.924526295205757e-05,
+      "loss": 21.2214,
+      "step": 7539
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.923747069649236e-05,
+      "loss": 19.2358,
+      "step": 7540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.922967789247131e-05,
+      "loss": 0.7289,
+      "step": 7541
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.922188454021656e-05,
+      "loss": 0.701,
+      "step": 7542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.921409063995033e-05,
+      "loss": 20.8281,
+      "step": 7543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.920629619189476e-05,
+      "loss": 0.7402,
+      "step": 7544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.919850119627213e-05,
+      "loss": 0.066,
+      "step": 7545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.919070565330465e-05,
+      "loss": 19.8202,
+      "step": 7546
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.918290956321456e-05,
+      "loss": 0.7443,
+      "step": 7547
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.917511292622415e-05,
+      "loss": 0.7345,
+      "step": 7548
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.916731574255565e-05,
+      "loss": 21.0002,
+      "step": 7549
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.915951801243142e-05,
+      "loss": 0.7056,
+      "step": 7550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.915171973607373e-05,
+      "loss": 0.7346,
+      "step": 7551
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.914392091370492e-05,
+      "loss": 17.9835,
+      "step": 7552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.913612154554733e-05,
+      "loss": 0.7537,
+      "step": 7553
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.912832163182333e-05,
+      "loss": 0.0574,
+      "step": 7554
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.912052117275526e-05,
+      "loss": 20.5431,
+      "step": 7555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.911272016856554e-05,
+      "loss": 0.0989,
+      "step": 7556
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.910491861947658e-05,
+      "loss": 0.0628,
+      "step": 7557
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.909711652571079e-05,
+      "loss": 21.3554,
+      "step": 7558
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.908931388749059e-05,
+      "loss": 0.7488,
+      "step": 7559
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.908151070503845e-05,
+      "loss": 19.1477,
+      "step": 7560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.907370697857683e-05,
+      "loss": 0.7916,
+      "step": 7561
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.906590270832822e-05,
+      "loss": 20.6397,
+      "step": 7562
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90580978945151e-05,
+      "loss": 0.7212,
+      "step": 7563
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.905029253736002e-05,
+      "loss": 19.5172,
+      "step": 7564
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.904248663708545e-05,
+      "loss": 0.7271,
+      "step": 7565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.903468019391399e-05,
+      "loss": 18.9155,
+      "step": 7566
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.902687320806818e-05,
+      "loss": 0.7797,
+      "step": 7567
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90190656797706e-05,
+      "loss": 20.4614,
+      "step": 7568
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.901125760924381e-05,
+      "loss": 20.8176,
+      "step": 7569
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.900344899671047e-05,
+      "loss": 19.6249,
+      "step": 7570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.899563984239315e-05,
+      "loss": 17.177,
+      "step": 7571
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.898783014651453e-05,
+      "loss": 20.2745,
+      "step": 7572
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.898001990929723e-05,
+      "loss": 0.7268,
+      "step": 7573
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.897220913096393e-05,
+      "loss": 19.9123,
+      "step": 7574
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.896439781173731e-05,
+      "loss": 20.0025,
+      "step": 7575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.895658595184009e-05,
+      "loss": 0.7475,
+      "step": 7576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.894877355149496e-05,
+      "loss": 0.0695,
+      "step": 7577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.894096061092465e-05,
+      "loss": 20.9849,
+      "step": 7578
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.893314713035191e-05,
+      "loss": 0.7974,
+      "step": 7579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.892533310999954e-05,
+      "loss": 19.7403,
+      "step": 7580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.891751855009023e-05,
+      "loss": 18.3819,
+      "step": 7581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.890970345084685e-05,
+      "loss": 0.7395,
+      "step": 7582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.890188781249217e-05,
+      "loss": 0.0514,
+      "step": 7583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.889407163524901e-05,
+      "loss": 0.08,
+      "step": 7584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.888625491934023e-05,
+      "loss": 0.0766,
+      "step": 7585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.887843766498866e-05,
+      "loss": 0.0736,
+      "step": 7586
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.887061987241719e-05,
+      "loss": 0.7107,
+      "step": 7587
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.886280154184867e-05,
+      "loss": 18.5242,
+      "step": 7588
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.885498267350605e-05,
+      "loss": 20.354,
+      "step": 7589
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.884716326761218e-05,
+      "loss": 18.3885,
+      "step": 7590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.883934332439003e-05,
+      "loss": 19.9264,
+      "step": 7591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.883152284406255e-05,
+      "loss": 19.2462,
+      "step": 7592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.88237018268527e-05,
+      "loss": 18.4313,
+      "step": 7593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.881588027298342e-05,
+      "loss": 0.7705,
+      "step": 7594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.880805818267773e-05,
+      "loss": 0.7871,
+      "step": 7595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.880023555615864e-05,
+      "loss": 18.7303,
+      "step": 7596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.879241239364915e-05,
+      "loss": 20.5414,
+      "step": 7597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.878458869537231e-05,
+      "loss": 18.0949,
+      "step": 7598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.877676446155118e-05,
+      "loss": 17.3783,
+      "step": 7599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.876893969240881e-05,
+      "loss": 19.3206,
+      "step": 7600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.876111438816829e-05,
+      "loss": 0.7618,
+      "step": 7601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.875328854905272e-05,
+      "loss": 0.7584,
+      "step": 7602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.874546217528519e-05,
+      "loss": 0.0584,
+      "step": 7603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.873763526708887e-05,
+      "loss": 18.0135,
+      "step": 7604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.872980782468688e-05,
+      "loss": 0.7375,
+      "step": 7605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.872197984830237e-05,
+      "loss": 0.7405,
+      "step": 7606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.871415133815852e-05,
+      "loss": 20.3659,
+      "step": 7607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.870632229447856e-05,
+      "loss": 0.6825,
+      "step": 7608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.869849271748561e-05,
+      "loss": 0.7543,
+      "step": 7609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.869066260740297e-05,
+      "loss": 0.7281,
+      "step": 7610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.868283196445382e-05,
+      "loss": 18.1255,
+      "step": 7611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.867500078886144e-05,
+      "loss": 0.7421,
+      "step": 7612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.866716908084908e-05,
+      "loss": 0.6884,
+      "step": 7613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.865933684064004e-05,
+      "loss": 18.8121,
+      "step": 7614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.865150406845758e-05,
+      "loss": 19.6648,
+      "step": 7615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.864367076452505e-05,
+      "loss": 18.6153,
+      "step": 7616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.863583692906575e-05,
+      "loss": 0.7006,
+      "step": 7617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.862800256230303e-05,
+      "loss": 20.8118,
+      "step": 7618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.862016766446024e-05,
+      "loss": 16.8584,
+      "step": 7619
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.861233223576075e-05,
+      "loss": 21.0348,
+      "step": 7620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.860449627642798e-05,
+      "loss": 20.198,
+      "step": 7621
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.859665978668528e-05,
+      "loss": 19.0357,
+      "step": 7622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.85888227667561e-05,
+      "loss": 0.7175,
+      "step": 7623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.858098521686384e-05,
+      "loss": 0.6523,
+      "step": 7624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.857314713723199e-05,
+      "loss": 0.7655,
+      "step": 7625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.856530852808399e-05,
+      "loss": 17.9148,
+      "step": 7626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.85574693896433e-05,
+      "loss": 0.6903,
+      "step": 7627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.854962972213344e-05,
+      "loss": 19.6973,
+      "step": 7628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.85417895257779e-05,
+      "loss": 18.5476,
+      "step": 7629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.853394880080023e-05,
+      "loss": 20.6033,
+      "step": 7630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.852610754742393e-05,
+      "loss": 19.1929,
+      "step": 7631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.851826576587256e-05,
+      "loss": 19.766,
+      "step": 7632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.851042345636971e-05,
+      "loss": 0.7317,
+      "step": 7633
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.850258061913894e-05,
+      "loss": 20.1776,
+      "step": 7634
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.849473725440386e-05,
+      "loss": 18.6766,
+      "step": 7635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.848689336238808e-05,
+      "loss": 18.4514,
+      "step": 7636
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.84790489433152e-05,
+      "loss": 0.7134,
+      "step": 7637
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.847120399740892e-05,
+      "loss": 20.7692,
+      "step": 7638
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.846335852489285e-05,
+      "loss": 0.7649,
+      "step": 7639
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.845551252599069e-05,
+      "loss": 19.0771,
+      "step": 7640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.84476660009261e-05,
+      "loss": 0.7874,
+      "step": 7641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.843981894992282e-05,
+      "loss": 0.7203,
+      "step": 7642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.843197137320451e-05,
+      "loss": 0.7763,
+      "step": 7643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.842412327099497e-05,
+      "loss": 20.3387,
+      "step": 7644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.84162746435179e-05,
+      "loss": 0.7507,
+      "step": 7645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.840842549099709e-05,
+      "loss": 20.9113,
+      "step": 7646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.840057581365631e-05,
+      "loss": 21.6551,
+      "step": 7647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.839272561171933e-05,
+      "loss": 17.18,
+      "step": 7648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.838487488540998e-05,
+      "loss": 0.7696,
+      "step": 7649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.83770236349521e-05,
+      "loss": 17.9144,
+      "step": 7650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.836917186056949e-05,
+      "loss": 20.0615,
+      "step": 7651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.836131956248602e-05,
+      "loss": 0.7211,
+      "step": 7652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.835346674092556e-05,
+      "loss": 18.0211,
+      "step": 7653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.834561339611199e-05,
+      "loss": 0.7551,
+      "step": 7654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.833775952826918e-05,
+      "loss": 0.7207,
+      "step": 7655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.83299051376211e-05,
+      "loss": 0.7949,
+      "step": 7656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.832205022439162e-05,
+      "loss": 0.069,
+      "step": 7657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.831419478880473e-05,
+      "loss": 0.7727,
+      "step": 7658
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.830633883108434e-05,
+      "loss": 20.9455,
+      "step": 7659
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.829848235145443e-05,
+      "loss": 21.2973,
+      "step": 7660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.829062535013902e-05,
+      "loss": 20.3363,
+      "step": 7661
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82827678273621e-05,
+      "loss": 0.7231,
+      "step": 7662
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.827490978334766e-05,
+      "loss": 17.4783,
+      "step": 7663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.826705121831976e-05,
+      "loss": 0.7218,
+      "step": 7664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.825919213250242e-05,
+      "loss": 18.0634,
+      "step": 7665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.825133252611973e-05,
+      "loss": 19.6031,
+      "step": 7666
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.824347239939574e-05,
+      "loss": 0.7123,
+      "step": 7667
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.823561175255456e-05,
+      "loss": 20.6029,
+      "step": 7668
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82277505858203e-05,
+      "loss": 21.0122,
+      "step": 7669
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.821988889941704e-05,
+      "loss": 0.7378,
+      "step": 7670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.821202669356895e-05,
+      "loss": 18.9096,
+      "step": 7671
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.820416396850017e-05,
+      "loss": 0.7974,
+      "step": 7672
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.819630072443487e-05,
+      "loss": 20.1239,
+      "step": 7673
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.818843696159721e-05,
+      "loss": 21.0005,
+      "step": 7674
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.818057268021141e-05,
+      "loss": 18.8847,
+      "step": 7675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.817270788050168e-05,
+      "loss": 18.765,
+      "step": 7676
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.81648425626922e-05,
+      "loss": 18.0037,
+      "step": 7677
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.815697672700725e-05,
+      "loss": 0.7099,
+      "step": 7678
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.814911037367109e-05,
+      "loss": 18.2143,
+      "step": 7679
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.814124350290797e-05,
+      "loss": 0.6991,
+      "step": 7680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.813337611494216e-05,
+      "loss": 0.7349,
+      "step": 7681
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.812550820999797e-05,
+      "loss": 0.7701,
+      "step": 7682
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.81176397882997e-05,
+      "loss": 0.0924,
+      "step": 7683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.810977085007172e-05,
+      "loss": 0.6928,
+      "step": 7684
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.810190139553832e-05,
+      "loss": 17.7758,
+      "step": 7685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80940314249239e-05,
+      "loss": 0.696,
+      "step": 7686
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.808616093845279e-05,
+      "loss": 18.7044,
+      "step": 7687
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80782899363494e-05,
+      "loss": 0.7399,
+      "step": 7688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.807041841883813e-05,
+      "loss": 20.5316,
+      "step": 7689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.806254638614337e-05,
+      "loss": 0.6953,
+      "step": 7690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80546738384896e-05,
+      "loss": 0.7225,
+      "step": 7691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.804680077610121e-05,
+      "loss": 19.5101,
+      "step": 7692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80389271992027e-05,
+      "loss": 18.5702,
+      "step": 7693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.803105310801851e-05,
+      "loss": 0.7385,
+      "step": 7694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.802317850277316e-05,
+      "loss": 19.6879,
+      "step": 7695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.801530338369113e-05,
+      "loss": 21.6583,
+      "step": 7696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.800742775099696e-05,
+      "loss": 19.6962,
+      "step": 7697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.799955160491517e-05,
+      "loss": 0.7065,
+      "step": 7698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.79916749456703e-05,
+      "loss": 0.7492,
+      "step": 7699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798379777348691e-05,
+      "loss": 0.7793,
+      "step": 7700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.79759200885896e-05,
+      "loss": 0.7237,
+      "step": 7701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.796804189120293e-05,
+      "loss": 19.0229,
+      "step": 7702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.796016318155154e-05,
+      "loss": 20.0135,
+      "step": 7703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.795228395986002e-05,
+      "loss": 0.7939,
+      "step": 7704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.794440422635302e-05,
+      "loss": 19.2151,
+      "step": 7705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.793652398125519e-05,
+      "loss": 18.3993,
+      "step": 7706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.792864322479117e-05,
+      "loss": 19.1539,
+      "step": 7707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.792076195718567e-05,
+      "loss": 21.9669,
+      "step": 7708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.791288017866338e-05,
+      "loss": 18.6755,
+      "step": 7709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.790499788944899e-05,
+      "loss": 17.8413,
+      "step": 7710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.789711508976723e-05,
+      "loss": 20.4847,
+      "step": 7711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.788923177984284e-05,
+      "loss": 0.7349,
+      "step": 7712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.788134795990058e-05,
+      "loss": 0.0566,
+      "step": 7713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.787346363016518e-05,
+      "loss": 21.0825,
+      "step": 7714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.786557879086147e-05,
+      "loss": 20.2217,
+      "step": 7715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.785769344221421e-05,
+      "loss": 18.8775,
+      "step": 7716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.784980758444823e-05,
+      "loss": 0.7851,
+      "step": 7717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.784192121778833e-05,
+      "loss": 18.9025,
+      "step": 7718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.783403434245937e-05,
+      "loss": 19.7241,
+      "step": 7719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78261469586862e-05,
+      "loss": 19.3718,
+      "step": 7720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.781825906669367e-05,
+      "loss": 0.7123,
+      "step": 7721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78103706667067e-05,
+      "loss": 19.6509,
+      "step": 7722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.780248175895013e-05,
+      "loss": 0.7521,
+      "step": 7723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.779459234364891e-05,
+      "loss": 0.7112,
+      "step": 7724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.778670242102798e-05,
+      "loss": 0.7377,
+      "step": 7725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.777881199131224e-05,
+      "loss": 0.6731,
+      "step": 7726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.777092105472667e-05,
+      "loss": 0.7109,
+      "step": 7727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.776302961149622e-05,
+      "loss": 21.996,
+      "step": 7728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.77551376618459e-05,
+      "loss": 19.4589,
+      "step": 7729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.774724520600069e-05,
+      "loss": 0.7628,
+      "step": 7730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.773935224418558e-05,
+      "loss": 17.3214,
+      "step": 7731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.773145877662565e-05,
+      "loss": 18.9103,
+      "step": 7732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.77235648035459e-05,
+      "loss": 20.1907,
+      "step": 7733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.77156703251714e-05,
+      "loss": 18.1186,
+      "step": 7734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.77077753417272e-05,
+      "loss": 0.7218,
+      "step": 7735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.769987985343842e-05,
+      "loss": 19.1751,
+      "step": 7736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.769198386053014e-05,
+      "loss": 19.9132,
+      "step": 7737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.768408736322745e-05,
+      "loss": 19.1209,
+      "step": 7738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.767619036175553e-05,
+      "loss": 18.7544,
+      "step": 7739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.766829285633947e-05,
+      "loss": 17.757,
+      "step": 7740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.766039484720445e-05,
+      "loss": 17.0464,
+      "step": 7741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.765249633457565e-05,
+      "loss": 0.7125,
+      "step": 7742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.764459731867823e-05,
+      "loss": 20.9963,
+      "step": 7743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.763669779973741e-05,
+      "loss": 17.6261,
+      "step": 7744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.762879777797839e-05,
+      "loss": 18.5183,
+      "step": 7745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.762089725362641e-05,
+      "loss": 0.7881,
+      "step": 7746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.76129962269067e-05,
+      "loss": 19.9214,
+      "step": 7747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.760509469804452e-05,
+      "loss": 0.679,
+      "step": 7748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.759719266726515e-05,
+      "loss": 0.0711,
+      "step": 7749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.758929013479385e-05,
+      "loss": 0.7303,
+      "step": 7750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.758138710085596e-05,
+      "loss": 0.7587,
+      "step": 7751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.757348356567676e-05,
+      "loss": 17.78,
+      "step": 7752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.756557952948157e-05,
+      "loss": 0.7305,
+      "step": 7753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.755767499249578e-05,
+      "loss": 0.7257,
+      "step": 7754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.754976995494469e-05,
+      "loss": 0.7038,
+      "step": 7755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.754186441705372e-05,
+      "loss": 20.6809,
+      "step": 7756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.753395837904822e-05,
+      "loss": 0.7037,
+      "step": 7757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.752605184115362e-05,
+      "loss": 19.1143,
+      "step": 7758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.75181448035953e-05,
+      "loss": 0.0614,
+      "step": 7759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.751023726659871e-05,
+      "loss": 19.7975,
+      "step": 7760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.750232923038928e-05,
+      "loss": 19.3724,
+      "step": 7761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.749442069519247e-05,
+      "loss": 18.2046,
+      "step": 7762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.748651166123376e-05,
+      "loss": 0.8021,
+      "step": 7763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.747860212873863e-05,
+      "loss": 0.0709,
+      "step": 7764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.747069209793257e-05,
+      "loss": 19.3267,
+      "step": 7765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.74627815690411e-05,
+      "loss": 0.7265,
+      "step": 7766
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.745487054228974e-05,
+      "loss": 19.5107,
+      "step": 7767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.744695901790405e-05,
+      "loss": 20.564,
+      "step": 7768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.743904699610956e-05,
+      "loss": 19.6073,
+      "step": 7769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.743113447713189e-05,
+      "loss": 0.7156,
+      "step": 7770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742322146119656e-05,
+      "loss": 0.7426,
+      "step": 7771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.741530794852919e-05,
+      "loss": 0.7413,
+      "step": 7772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.740739393935541e-05,
+      "loss": 0.7248,
+      "step": 7773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.739947943390083e-05,
+      "loss": 20.0703,
+      "step": 7774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.73915644323911e-05,
+      "loss": 0.7847,
+      "step": 7775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.738364893505187e-05,
+      "loss": 0.7023,
+      "step": 7776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.737573294210882e-05,
+      "loss": 20.655,
+      "step": 7777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.736781645378761e-05,
+      "loss": 17.9059,
+      "step": 7778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.735989947031395e-05,
+      "loss": 0.7623,
+      "step": 7779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.735198199191357e-05,
+      "loss": 0.0628,
+      "step": 7780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.734406401881215e-05,
+      "loss": 19.4521,
+      "step": 7781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.733614555123546e-05,
+      "loss": 17.8931,
+      "step": 7782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.732822658940927e-05,
+      "loss": 20.6845,
+      "step": 7783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.732030713355932e-05,
+      "loss": 0.7493,
+      "step": 7784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.73123871839114e-05,
+      "loss": 0.7493,
+      "step": 7785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.730446674069131e-05,
+      "loss": 0.7698,
+      "step": 7786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.729654580412485e-05,
+      "loss": 18.5226,
+      "step": 7787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.728862437443786e-05,
+      "loss": 19.5126,
+      "step": 7788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.728070245185616e-05,
+      "loss": 20.3002,
+      "step": 7789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.727278003660561e-05,
+      "loss": 0.0796,
+      "step": 7790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.726485712891209e-05,
+      "loss": 19.9726,
+      "step": 7791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.725693372900143e-05,
+      "loss": 0.6568,
+      "step": 7792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.72490098370996e-05,
+      "loss": 0.7812,
+      "step": 7793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.724108545343246e-05,
+      "loss": 18.0641,
+      "step": 7794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.723316057822594e-05,
+      "loss": 21.5866,
+      "step": 7795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.722523521170597e-05,
+      "loss": 19.7952,
+      "step": 7796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.721730935409852e-05,
+      "loss": 0.7372,
+      "step": 7797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720938300562953e-05,
+      "loss": 0.0845,
+      "step": 7798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720145616652501e-05,
+      "loss": 0.0641,
+      "step": 7799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.719352883701092e-05,
+      "loss": 0.0752,
+      "step": 7800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.718560101731328e-05,
+      "loss": 20.6593,
+      "step": 7801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.717767270765813e-05,
+      "loss": 0.7436,
+      "step": 7802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.716974390827147e-05,
+      "loss": 17.6856,
+      "step": 7803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.716181461937936e-05,
+      "loss": 0.7265,
+      "step": 7804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.715388484120786e-05,
+      "loss": 19.0518,
+      "step": 7805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.714595457398306e-05,
+      "loss": 18.0996,
+      "step": 7806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.713802381793105e-05,
+      "loss": 0.7343,
+      "step": 7807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.713009257327792e-05,
+      "loss": 19.3606,
+      "step": 7808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.712216084024978e-05,
+      "loss": 0.0833,
+      "step": 7809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.711422861907279e-05,
+      "loss": 19.764,
+      "step": 7810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.710629590997308e-05,
+      "loss": 19.186,
+      "step": 7811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.709836271317681e-05,
+      "loss": 0.6742,
+      "step": 7812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.709042902891015e-05,
+      "loss": 17.7508,
+      "step": 7813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.70824948573993e-05,
+      "loss": 20.2764,
+      "step": 7814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.707456019887043e-05,
+      "loss": 18.9417,
+      "step": 7815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.706662505354981e-05,
+      "loss": 0.7475,
+      "step": 7816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.705868942166362e-05,
+      "loss": 0.7222,
+      "step": 7817
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.705075330343813e-05,
+      "loss": 0.6967,
+      "step": 7818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.704281669909958e-05,
+      "loss": 0.7102,
+      "step": 7819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.703487960887426e-05,
+      "loss": 0.7362,
+      "step": 7820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.702694203298843e-05,
+      "loss": 0.0691,
+      "step": 7821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.70190039716684e-05,
+      "loss": 0.7094,
+      "step": 7822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.70110654251405e-05,
+      "loss": 0.0539,
+      "step": 7823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.700312639363103e-05,
+      "loss": 0.7372,
+      "step": 7824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.699518687736634e-05,
+      "loss": 0.7003,
+      "step": 7825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.69872468765728e-05,
+      "loss": 18.9825,
+      "step": 7826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.697930639147673e-05,
+      "loss": 19.7714,
+      "step": 7827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.697136542230458e-05,
+      "loss": 0.731,
+      "step": 7828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696342396928268e-05,
+      "loss": 0.7593,
+      "step": 7829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.695548203263746e-05,
+      "loss": 19.3626,
+      "step": 7830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.694753961259538e-05,
+      "loss": 0.7453,
+      "step": 7831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.693959670938282e-05,
+      "loss": 20.2085,
+      "step": 7832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.693165332322629e-05,
+      "loss": 0.0594,
+      "step": 7833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.69237094543522e-05,
+      "loss": 20.8306,
+      "step": 7834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.691576510298703e-05,
+      "loss": 0.7492,
+      "step": 7835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.690782026935732e-05,
+      "loss": 0.0858,
+      "step": 7836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.689987495368954e-05,
+      "loss": 0.7365,
+      "step": 7837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.68919291562102e-05,
+      "loss": 20.2297,
+      "step": 7838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.688398287714587e-05,
+      "loss": 21.4552,
+      "step": 7839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.687603611672306e-05,
+      "loss": 18.4591,
+      "step": 7840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.686808887516834e-05,
+      "loss": 20.3011,
+      "step": 7841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.686014115270829e-05,
+      "loss": 0.0586,
+      "step": 7842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.68521929495695e-05,
+      "loss": 19.5322,
+      "step": 7843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.684424426597858e-05,
+      "loss": 0.7757,
+      "step": 7844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.683629510216212e-05,
+      "loss": 0.7223,
+      "step": 7845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.682834545834676e-05,
+      "loss": 18.7595,
+      "step": 7846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.682039533475915e-05,
+      "loss": 21.1023,
+      "step": 7847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.681244473162594e-05,
+      "loss": 20.0179,
+      "step": 7848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.680449364917379e-05,
+      "loss": 17.839,
+      "step": 7849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.679654208762943e-05,
+      "loss": 19.3082,
+      "step": 7850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.678859004721947e-05,
+      "loss": 17.7734,
+      "step": 7851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.67806375281707e-05,
+      "loss": 17.4669,
+      "step": 7852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.677268453070982e-05,
+      "loss": 22.0936,
+      "step": 7853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.676473105506354e-05,
+      "loss": 0.705,
+      "step": 7854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.675677710145867e-05,
+      "loss": 0.7362,
+      "step": 7855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.674882267012193e-05,
+      "loss": 0.7969,
+      "step": 7856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.67408677612801e-05,
+      "loss": 20.5619,
+      "step": 7857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.673291237515998e-05,
+      "loss": 20.8772,
+      "step": 7858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.67249565119884e-05,
+      "loss": 20.1037,
+      "step": 7859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.671700017199213e-05,
+      "loss": 0.0765,
+      "step": 7860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.670904335539806e-05,
+      "loss": 0.073,
+      "step": 7861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6701086062433e-05,
+      "loss": 0.7591,
+      "step": 7862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669312829332382e-05,
+      "loss": 20.1414,
+      "step": 7863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.66851700482974e-05,
+      "loss": 0.6926,
+      "step": 7864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667721132758062e-05,
+      "loss": 17.7136,
+      "step": 7865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.666925213140039e-05,
+      "loss": 0.7075,
+      "step": 7866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.66612924599836e-05,
+      "loss": 19.9964,
+      "step": 7867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.665333231355722e-05,
+      "loss": 0.7541,
+      "step": 7868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.664537169234815e-05,
+      "loss": 0.0844,
+      "step": 7869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.663741059658338e-05,
+      "loss": 19.5914,
+      "step": 7870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.662944902648985e-05,
+      "loss": 0.0606,
+      "step": 7871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.662148698229456e-05,
+      "loss": 19.1704,
+      "step": 7872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.661352446422448e-05,
+      "loss": 18.8775,
+      "step": 7873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.660556147250668e-05,
+      "loss": 0.7504,
+      "step": 7874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.659759800736812e-05,
+      "loss": 19.3446,
+      "step": 7875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.658963406903588e-05,
+      "loss": 20.8682,
+      "step": 7876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.658166965773698e-05,
+      "loss": 19.347,
+      "step": 7877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.657370477369848e-05,
+      "loss": 17.8582,
+      "step": 7878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.65657394171475e-05,
+      "loss": 0.7643,
+      "step": 7879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.655777358831109e-05,
+      "loss": 19.4762,
+      "step": 7880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.654980728741635e-05,
+      "loss": 19.5127,
+      "step": 7881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.654184051469044e-05,
+      "loss": 0.7018,
+      "step": 7882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.653387327036045e-05,
+      "loss": 19.213,
+      "step": 7883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.652590555465354e-05,
+      "loss": 19.1635,
+      "step": 7884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.651793736779687e-05,
+      "loss": 18.5955,
+      "step": 7885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.650996871001761e-05,
+      "loss": 0.7624,
+      "step": 7886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.650199958154292e-05,
+      "loss": 0.7425,
+      "step": 7887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649402998260005e-05,
+      "loss": 18.6155,
+      "step": 7888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.648605991341616e-05,
+      "loss": 18.7424,
+      "step": 7889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.647808937421851e-05,
+      "loss": 0.0729,
+      "step": 7890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.647011836523433e-05,
+      "loss": 22.0625,
+      "step": 7891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.646214688669084e-05,
+      "loss": 0.7947,
+      "step": 7892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.645417493881534e-05,
+      "loss": 0.7364,
+      "step": 7893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.644620252183512e-05,
+      "loss": 18.8407,
+      "step": 7894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.643822963597743e-05,
+      "loss": 17.5312,
+      "step": 7895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.643025628146961e-05,
+      "loss": 17.9507,
+      "step": 7896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.642228245853896e-05,
+      "loss": 0.6484,
+      "step": 7897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.641430816741283e-05,
+      "loss": 0.066,
+      "step": 7898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640633340831853e-05,
+      "loss": 0.6983,
+      "step": 7899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.639835818148345e-05,
+      "loss": 0.06,
+      "step": 7900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.639038248713496e-05,
+      "loss": 20.2221,
+      "step": 7901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.638240632550043e-05,
+      "loss": 18.7788,
+      "step": 7902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.637442969680727e-05,
+      "loss": 19.2267,
+      "step": 7903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.636645260128288e-05,
+      "loss": 0.7282,
+      "step": 7904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.63584750391547e-05,
+      "loss": 0.7639,
+      "step": 7905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.635049701065016e-05,
+      "loss": 19.8041,
+      "step": 7906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.634251851599671e-05,
+      "loss": 19.4067,
+      "step": 7907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.633453955542183e-05,
+      "loss": 0.7133,
+      "step": 7908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.632656012915298e-05,
+      "loss": 20.4681,
+      "step": 7909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.631858023741765e-05,
+      "loss": 20.8938,
+      "step": 7910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.631059988044337e-05,
+      "loss": 0.0794,
+      "step": 7911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.630261905845763e-05,
+      "loss": 19.1967,
+      "step": 7912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.629463777168799e-05,
+      "loss": 0.0622,
+      "step": 7913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.628665602036194e-05,
+      "loss": 0.0735,
+      "step": 7914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.627867380470712e-05,
+      "loss": 0.7357,
+      "step": 7915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.627069112495104e-05,
+      "loss": 0.7691,
+      "step": 7916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.626270798132128e-05,
+      "loss": 19.7253,
+      "step": 7917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.625472437404548e-05,
+      "loss": 0.7615,
+      "step": 7918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.624674030335123e-05,
+      "loss": 0.72,
+      "step": 7919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.623875576946616e-05,
+      "loss": 0.0915,
+      "step": 7920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.623077077261788e-05,
+      "loss": 0.7254,
+      "step": 7921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.622278531303407e-05,
+      "loss": 0.7382,
+      "step": 7922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.621479939094238e-05,
+      "loss": 19.0387,
+      "step": 7923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.620681300657051e-05,
+      "loss": 0.7091,
+      "step": 7924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.619882616014612e-05,
+      "loss": 19.1706,
+      "step": 7925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.619083885189692e-05,
+      "loss": 19.1436,
+      "step": 7926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.618285108205063e-05,
+      "loss": 0.7157,
+      "step": 7927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.617486285083499e-05,
+      "loss": 18.1575,
+      "step": 7928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.616687415847773e-05,
+      "loss": 21.2666,
+      "step": 7929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.615888500520661e-05,
+      "loss": 0.7114,
+      "step": 7930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.61508953912494e-05,
+      "loss": 20.7498,
+      "step": 7931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.614290531683387e-05,
+      "loss": 19.1777,
+      "step": 7932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.613491478218783e-05,
+      "loss": 17.6341,
+      "step": 7933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.612692378753907e-05,
+      "loss": 0.7636,
+      "step": 7934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.611893233311545e-05,
+      "loss": 0.6999,
+      "step": 7935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.611094041914477e-05,
+      "loss": 20.4797,
+      "step": 7936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.610294804585489e-05,
+      "loss": 18.5159,
+      "step": 7937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.609495521347365e-05,
+      "loss": 20.4954,
+      "step": 7938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.608696192222895e-05,
+      "loss": 18.0026,
+      "step": 7939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.607896817234866e-05,
+      "loss": 21.0452,
+      "step": 7940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.60709739640607e-05,
+      "loss": 0.0748,
+      "step": 7941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.606297929759299e-05,
+      "loss": 19.4052,
+      "step": 7942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.60549841731734e-05,
+      "loss": 20.2587,
+      "step": 7943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.604698859102993e-05,
+      "loss": 20.7393,
+      "step": 7944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.60389925513905e-05,
+      "loss": 0.0799,
+      "step": 7945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.603099605448309e-05,
+      "loss": 19.5288,
+      "step": 7946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.602299910053567e-05,
+      "loss": 19.2017,
+      "step": 7947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.601500168977623e-05,
+      "loss": 20.6418,
+      "step": 7948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.600700382243279e-05,
+      "loss": 18.2534,
+      "step": 7949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.599900549873335e-05,
+      "loss": 20.4602,
+      "step": 7950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.599100671890593e-05,
+      "loss": 19.2842,
+      "step": 7951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.598300748317861e-05,
+      "loss": 0.7441,
+      "step": 7952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59750077917794e-05,
+      "loss": 18.3941,
+      "step": 7953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.596700764493643e-05,
+      "loss": 18.0886,
+      "step": 7954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.595900704287772e-05,
+      "loss": 17.7895,
+      "step": 7955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.595100598583141e-05,
+      "loss": 18.9006,
+      "step": 7956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.594300447402558e-05,
+      "loss": 0.7574,
+      "step": 7957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.593500250768836e-05,
+      "loss": 0.7683,
+      "step": 7958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59270000870479e-05,
+      "loss": 20.3669,
+      "step": 7959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.591899721233231e-05,
+      "loss": 19.4363,
+      "step": 7960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.59109938837698e-05,
+      "loss": 0.7623,
+      "step": 7961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.590299010158849e-05,
+      "loss": 19.7402,
+      "step": 7962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58949858660166e-05,
+      "loss": 17.911,
+      "step": 7963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.588698117728235e-05,
+      "loss": 0.7679,
+      "step": 7964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58789760356139e-05,
+      "loss": 20.1159,
+      "step": 7965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58709704412395e-05,
+      "loss": 19.2333,
+      "step": 7966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.586296439438737e-05,
+      "loss": 0.0507,
+      "step": 7967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58549578952858e-05,
+      "loss": 0.706,
+      "step": 7968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.584695094416302e-05,
+      "loss": 20.7667,
+      "step": 7969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.583894354124732e-05,
+      "loss": 0.7372,
+      "step": 7970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.583093568676698e-05,
+      "loss": 0.7205,
+      "step": 7971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.582292738095032e-05,
+      "loss": 18.3569,
+      "step": 7972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.581491862402563e-05,
+      "loss": 0.7384,
+      "step": 7973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.580690941622126e-05,
+      "loss": 0.6786,
+      "step": 7974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.579889975776552e-05,
+      "loss": 0.0879,
+      "step": 7975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57908896488868e-05,
+      "loss": 0.7118,
+      "step": 7976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.578287908981345e-05,
+      "loss": 0.7301,
+      "step": 7977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.577486808077386e-05,
+      "loss": 0.7427,
+      "step": 7978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57668566219964e-05,
+      "loss": 22.2554,
+      "step": 7979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.575884471370948e-05,
+      "loss": 0.6855,
+      "step": 7980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.575083235614153e-05,
+      "loss": 19.6042,
+      "step": 7981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.574281954952098e-05,
+      "loss": 0.7435,
+      "step": 7982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.573480629407627e-05,
+      "loss": 19.1973,
+      "step": 7983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.572679259003584e-05,
+      "loss": 0.7131,
+      "step": 7984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.571877843762818e-05,
+      "loss": 20.8154,
+      "step": 7985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.571076383708177e-05,
+      "loss": 18.2125,
+      "step": 7986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.570274878862509e-05,
+      "loss": 19.3228,
+      "step": 7987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.569473329248667e-05,
+      "loss": 0.7457,
+      "step": 7988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5686717348895e-05,
+      "loss": 17.5388,
+      "step": 7989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.567870095807864e-05,
+      "loss": 18.4613,
+      "step": 7990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.56706841202661e-05,
+      "loss": 0.6816,
+      "step": 7991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.566266683568599e-05,
+      "loss": 17.7155,
+      "step": 7992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.565464910456685e-05,
+      "loss": 18.7022,
+      "step": 7993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.564663092713728e-05,
+      "loss": 19.5655,
+      "step": 7994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.563861230362585e-05,
+      "loss": 19.5307,
+      "step": 7995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.563059323426117e-05,
+      "loss": 0.7074,
+      "step": 7996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.56225737192719e-05,
+      "loss": 0.7023,
+      "step": 7997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.561455375888662e-05,
+      "loss": 17.6253,
+      "step": 7998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.560653335333403e-05,
+      "loss": 18.9435,
+      "step": 7999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.559851250284275e-05,
+      "loss": 19.4046,
+      "step": 8000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 19182,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 2000,
+  "total_flos": 2.1301236613845405e+20,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}