diff --git "a/lora-checkpoints-5/trainer_state.json" "b/lora-checkpoints-5/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/lora-checkpoints-5/trainer_state.json"
@@ -0,0 +1,26047 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 4337,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.053435114503817e-07,
+      "loss": 1.1546,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.106870229007634e-07,
+      "loss": 1.5863,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.160305343511451e-07,
+      "loss": 1.7071,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2213740458015268e-06,
+      "loss": 1.1316,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5267175572519086e-06,
+      "loss": 1.1343,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8320610687022903e-06,
+      "loss": 1.3009,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.137404580152672e-06,
+      "loss": 1.2085,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4427480916030536e-06,
+      "loss": 1.1013,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7480916030534356e-06,
+      "loss": 1.3494,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0534351145038173e-06,
+      "loss": 1.3692,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.358778625954199e-06,
+      "loss": 1.2544,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6641221374045806e-06,
+      "loss": 1.2842,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.969465648854962e-06,
+      "loss": 1.1179,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.274809160305344e-06,
+      "loss": 1.6752,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.580152671755725e-06,
+      "loss": 1.0938,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.885496183206107e-06,
+      "loss": 1.1507,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.190839694656488e-06,
+      "loss": 1.6561,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.496183206106871e-06,
+      "loss": 1.2087,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.801526717557252e-06,
+      "loss": 1.1695,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.1068702290076346e-06,
+      "loss": 1.1323,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.412213740458016e-06,
+      "loss": 1.1007,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.717557251908398e-06,
+      "loss": 1.117,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.022900763358779e-06,
+      "loss": 1.1185,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.328244274809161e-06,
+      "loss": 1.1348,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.633587786259543e-06,
+      "loss": 1.4719,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.938931297709924e-06,
+      "loss": 0.8632,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.244274809160306e-06,
+      "loss": 1.6482,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.549618320610688e-06,
+      "loss": 1.3893,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.85496183206107e-06,
+      "loss": 1.0095,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.16030534351145e-06,
+      "loss": 1.0069,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.465648854961834e-06,
+      "loss": 1.0634,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.770992366412214e-06,
+      "loss": 1.0788,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0076335877862595e-05,
+      "loss": 1.6855,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0381679389312977e-05,
+      "loss": 1.0268,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.068702290076336e-05,
+      "loss": 1.1358,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0992366412213743e-05,
+      "loss": 1.1602,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1297709923664125e-05,
+      "loss": 1.0996,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1603053435114503e-05,
+      "loss": 1.0077,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1908396946564887e-05,
+      "loss": 1.0953,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2213740458015269e-05,
+      "loss": 1.0876,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2519083969465651e-05,
+      "loss": 1.1644,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2824427480916032e-05,
+      "loss": 1.0992,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3129770992366414e-05,
+      "loss": 1.0186,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3435114503816796e-05,
+      "loss": 1.0906,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3740458015267178e-05,
+      "loss": 0.9245,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4045801526717558e-05,
+      "loss": 1.2479,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.435114503816794e-05,
+      "loss": 1.4689,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4656488549618322e-05,
+      "loss": 0.8777,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4961832061068704e-05,
+      "loss": 0.9467,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5267175572519086e-05,
+      "loss": 1.2606,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5572519083969465e-05,
+      "loss": 1.0356,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5877862595419847e-05,
+      "loss": 0.8046,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.618320610687023e-05,
+      "loss": 0.9158,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.648854961832061e-05,
+      "loss": 0.9861,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6793893129770993e-05,
+      "loss": 0.9692,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7099236641221375e-05,
+      "loss": 0.8896,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7404580152671757e-05,
+      "loss": 0.9864,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.770992366412214e-05,
+      "loss": 0.893,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8015267175572518e-05,
+      "loss": 1.0149,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.83206106870229e-05,
+      "loss": 0.8883,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8625954198473282e-05,
+      "loss": 0.8808,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8931297709923668e-05,
+      "loss": 1.0932,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.923664122137405e-05,
+      "loss": 0.7266,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.954198473282443e-05,
+      "loss": 0.8716,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.984732824427481e-05,
+      "loss": 0.8232,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.015267175572519e-05,
+      "loss": 0.9315,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.045801526717557e-05,
+      "loss": 0.9173,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0763358778625953e-05,
+      "loss": 0.9468,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1068702290076335e-05,
+      "loss": 0.8663,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.137404580152672e-05,
+      "loss": 0.9897,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1679389312977103e-05,
+      "loss": 0.8306,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1984732824427485e-05,
+      "loss": 1.3473,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2290076335877867e-05,
+      "loss": 1.0339,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.259541984732825e-05,
+      "loss": 1.3076,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2900763358778624e-05,
+      "loss": 1.0157,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3206106870229007e-05,
+      "loss": 0.9519,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3511450381679392e-05,
+      "loss": 1.2878,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3816793893129774e-05,
+      "loss": 0.8375,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4122137404580156e-05,
+      "loss": 0.8266,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4427480916030538e-05,
+      "loss": 0.9946,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.473282442748092e-05,
+      "loss": 0.9324,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5038167938931302e-05,
+      "loss": 0.9458,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5343511450381678e-05,
+      "loss": 0.9052,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5648854961832063e-05,
+      "loss": 1.1992,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5954198473282445e-05,
+      "loss": 0.8747,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6259541984732827e-05,
+      "loss": 0.9392,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.656488549618321e-05,
+      "loss": 0.9941,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.687022900763359e-05,
+      "loss": 0.9105,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7175572519083973e-05,
+      "loss": 0.9174,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7480916030534355e-05,
+      "loss": 0.8555,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7786259541984738e-05,
+      "loss": 0.9343,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8091603053435116e-05,
+      "loss": 0.7461,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8396946564885498e-05,
+      "loss": 0.8413,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.870229007633588e-05,
+      "loss": 0.9404,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9007633587786262e-05,
+      "loss": 0.8218,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9312977099236644e-05,
+      "loss": 1.0058,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9618320610687027e-05,
+      "loss": 0.8058,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.992366412213741e-05,
+      "loss": 1.059,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.022900763358779e-05,
+      "loss": 0.8625,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.053435114503817e-05,
+      "loss": 0.8711,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.083969465648855e-05,
+      "loss": 0.8226,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.114503816793893e-05,
+      "loss": 0.8473,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.145038167938931e-05,
+      "loss": 0.8201,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1755725190839694e-05,
+      "loss": 0.8585,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2061068702290076e-05,
+      "loss": 0.8062,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.236641221374046e-05,
+      "loss": 0.7672,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.267175572519084e-05,
+      "loss": 0.9464,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.297709923664122e-05,
+      "loss": 1.0316,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3282442748091605e-05,
+      "loss": 0.873,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.358778625954199e-05,
+      "loss": 0.9024,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.389312977099237e-05,
+      "loss": 0.7062,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.419847328244275e-05,
+      "loss": 0.9637,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.450381679389313e-05,
+      "loss": 1.1582,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4809160305343515e-05,
+      "loss": 0.7958,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.51145038167939e-05,
+      "loss": 0.8139,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.541984732824428e-05,
+      "loss": 1.0452,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.572519083969466e-05,
+      "loss": 0.9239,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6030534351145036e-05,
+      "loss": 0.8782,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.633587786259542e-05,
+      "loss": 1.2617,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.66412213740458e-05,
+      "loss": 0.8222,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.694656488549618e-05,
+      "loss": 0.9482,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7251908396946565e-05,
+      "loss": 1.2022,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.755725190839695e-05,
+      "loss": 0.9138,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7862595419847336e-05,
+      "loss": 0.7909,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.816793893129772e-05,
+      "loss": 0.9674,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.84732824427481e-05,
+      "loss": 0.8335,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8778625954198475e-05,
+      "loss": 1.1252,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.908396946564886e-05,
+      "loss": 0.848,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.938931297709924e-05,
+      "loss": 1.0336,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.969465648854962e-05,
+      "loss": 0.8172,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4e-05,
+      "loss": 0.9559,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999999442093799e-05,
+      "loss": 1.1124,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999997768375508e-05,
+      "loss": 0.9703,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999949788460596e-05,
+      "loss": 0.9623,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99999107350701e-05,
+      "loss": 0.9177,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999986052360539e-05,
+      "loss": 1.2945,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999979915409448e-05,
+      "loss": 0.918,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999726626571596e-05,
+      "loss": 0.8189,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999964294107722e-05,
+      "loss": 1.2568,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999954809765801e-05,
+      "loss": 0.7931,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999944209636692e-05,
+      "loss": 0.8432,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999324937263065e-05,
+      "loss": 1.1578,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999919662041181e-05,
+      "loss": 0.7347,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999905714588474e-05,
+      "loss": 0.886,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9998906513759686e-05,
+      "loss": 1.177,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999874472412066e-05,
+      "loss": 0.8972,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999857177705795e-05,
+      "loss": 0.8747,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9998387672668034e-05,
+      "loss": 0.788,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999819241105363e-05,
+      "loss": 0.7748,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9997985992323666e-05,
+      "loss": 0.9486,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999776841659332e-05,
+      "loss": 0.7394,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9997539683983955e-05,
+      "loss": 0.7844,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9997299794623205e-05,
+      "loss": 0.9398,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9997048748644895e-05,
+      "loss": 0.7815,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999678654618909e-05,
+      "loss": 0.7918,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9996513187402067e-05,
+      "loss": 0.782,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999622867243634e-05,
+      "loss": 0.9553,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999593300145064e-05,
+      "loss": 0.7623,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9995626174609925e-05,
+      "loss": 0.8832,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9995308192085376e-05,
+      "loss": 0.7415,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99949790540544e-05,
+      "loss": 0.8118,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9994638760700614e-05,
+      "loss": 0.8318,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999428731221388e-05,
+      "loss": 0.9223,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999392470879027e-05,
+      "loss": 0.8524,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999355095063209e-05,
+      "loss": 0.8378,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9993166037947844e-05,
+      "loss": 0.8666,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999276997095229e-05,
+      "loss": 1.1238,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99923627498664e-05,
+      "loss": 0.9479,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999194437491736e-05,
+      "loss": 1.1891,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999151484633858e-05,
+      "loss": 0.935,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999107416436971e-05,
+      "loss": 1.0666,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999062232925659e-05,
+      "loss": 1.0671,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9990159341251316e-05,
+      "loss": 1.1497,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998968520061218e-05,
+      "loss": 1.0648,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998919990760372e-05,
+      "loss": 0.7585,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998870346249668e-05,
+      "loss": 0.8616,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9988195865568025e-05,
+      "loss": 0.7019,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998767711710096e-05,
+      "loss": 0.9288,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998714721738488e-05,
+      "loss": 0.8247,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9986606166715434e-05,
+      "loss": 0.783,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998605396539446e-05,
+      "loss": 0.9183,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998549061373006e-05,
+      "loss": 0.8104,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998491611203651e-05,
+      "loss": 0.9681,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9984330460634336e-05,
+      "loss": 1.1243,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998373365985027e-05,
+      "loss": 0.7377,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9983125710017286e-05,
+      "loss": 0.7818,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998250661147455e-05,
+      "loss": 0.7724,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998187636456746e-05,
+      "loss": 0.6739,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998123496964764e-05,
+      "loss": 0.8481,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998058242707293e-05,
+      "loss": 0.9186,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9979918737207396e-05,
+      "loss": 0.904,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997924390042129e-05,
+      "loss": 0.8486,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997855791709112e-05,
+      "loss": 1.1047,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997786078759961e-05,
+      "loss": 1.0037,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997715251233568e-05,
+      "loss": 0.8073,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9976433091694484e-05,
+      "loss": 0.7282,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99757025260774e-05,
+      "loss": 0.7571,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9974960815892004e-05,
+      "loss": 0.7742,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997420796155211e-05,
+      "loss": 0.7731,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997344396347772e-05,
+      "loss": 0.8329,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99726688220951e-05,
+      "loss": 1.2642,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9971882537836696e-05,
+      "loss": 0.7201,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997108511114117e-05,
+      "loss": 0.6882,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9970276542453425e-05,
+      "loss": 0.8447,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996945683222456e-05,
+      "loss": 0.836,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99686259809119e-05,
+      "loss": 0.7898,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996778398897898e-05,
+      "loss": 0.67,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9966930856895556e-05,
+      "loss": 0.9155,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9966066585137594e-05,
+      "loss": 1.2127,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996519117418727e-05,
+      "loss": 0.7845,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996430462453299e-05,
+      "loss": 0.8269,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9963406936669356e-05,
+      "loss": 0.9106,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996249811109721e-05,
+      "loss": 0.7315,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996157814832357e-05,
+      "loss": 0.7742,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9960647048861706e-05,
+      "loss": 0.7678,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9959704813231075e-05,
+      "loss": 0.8196,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995875144195736e-05,
+      "loss": 0.6994,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9957786935572455e-05,
+      "loss": 1.0276,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995681129461446e-05,
+      "loss": 0.64,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995582451962768e-05,
+      "loss": 0.7977,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9954826611162674e-05,
+      "loss": 1.0735,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995381756977615e-05,
+      "loss": 1.2456,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995279739603108e-05,
+      "loss": 0.7582,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9951766090496613e-05,
+      "loss": 1.2178,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995072365374812e-05,
+      "loss": 1.0932,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994967008636719e-05,
+      "loss": 1.03,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994860538894161e-05,
+      "loss": 1.2021,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9947529562065384e-05,
+      "loss": 0.8199,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994644260633872e-05,
+      "loss": 0.7734,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9945344522368036e-05,
+      "loss": 1.1799,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9944235310765965e-05,
+      "loss": 0.7561,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994311497215135e-05,
+      "loss": 0.8045,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994198350714921e-05,
+      "loss": 0.8463,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994084091639082e-05,
+      "loss": 1.2276,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993968720051363e-05,
+      "loss": 0.7978,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99385223601613e-05,
+      "loss": 0.7693,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9937346395983716e-05,
+      "loss": 0.7971,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993615930863694e-05,
+      "loss": 0.9372,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993496109878327e-05,
+      "loss": 0.8889,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993375176709118e-05,
+      "loss": 0.8882,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993253131423537e-05,
+      "loss": 0.8368,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9931299740896736e-05,
+      "loss": 0.8738,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9930057047762394e-05,
+      "loss": 0.8411,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9928803235525634e-05,
+      "loss": 0.8553,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992753830488597e-05,
+      "loss": 0.9429,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9926262256549116e-05,
+      "loss": 1.2483,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992497509122698e-05,
+      "loss": 0.8963,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9923676809637696e-05,
+      "loss": 0.7942,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992236741250557e-05,
+      "loss": 1.2758,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992104690056113e-05,
+      "loss": 1.2297,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9919715274541096e-05,
+      "loss": 0.7905,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9918372535188387e-05,
+      "loss": 1.1766,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991701868325212e-05,
+      "loss": 0.6952,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9915653719487634e-05,
+      "loss": 0.8073,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9914277644656436e-05,
+      "loss": 0.7324,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991289045952625e-05,
+      "loss": 0.8888,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991149216487101e-05,
+      "loss": 1.0818,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991008276147081e-05,
+      "loss": 0.9616,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990866225011198e-05,
+      "loss": 0.7596,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990723063158702e-05,
+      "loss": 0.8463,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990578790669465e-05,
+      "loss": 0.7689,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990433407623978e-05,
+      "loss": 0.8412,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990286914103349e-05,
+      "loss": 0.6661,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.99013931018931e-05,
+      "loss": 0.8961,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989990595964209e-05,
+      "loss": 0.8148,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989840771511014e-05,
+      "loss": 1.2602,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989689836913313e-05,
+      "loss": 1.0659,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989537792255315e-05,
+      "loss": 0.7735,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989384637621846e-05,
+      "loss": 0.8704,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989230373098351e-05,
+      "loss": 0.8482,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989074998770896e-05,
+      "loss": 0.6844,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988918514726165e-05,
+      "loss": 1.0924,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988760921051461e-05,
+      "loss": 1.1108,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9886022178347074e-05,
+      "loss": 0.7382,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9884424051644445e-05,
+      "loss": 1.0437,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9882814831298337e-05,
+      "loss": 0.8406,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988119451820654e-05,
+      "loss": 0.8444,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987956311327305e-05,
+      "loss": 0.793,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987792061740802e-05,
+      "loss": 0.753,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987626703152781e-05,
+      "loss": 0.8923,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987460235655497e-05,
+      "loss": 0.7261,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9872926593418245e-05,
+      "loss": 0.9211,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9871239743052537e-05,
+      "loss": 0.7562,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9869541806398954e-05,
+      "loss": 0.8923,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986783278440479e-05,
+      "loss": 1.0365,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9866112678023515e-05,
+      "loss": 0.7093,
+      "step": 286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986438148821478e-05,
+      "loss": 0.6922,
+      "step": 287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9862639215944436e-05,
+      "loss": 1.1095,
+      "step": 288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986088586218451e-05,
+      "loss": 0.6919,
+      "step": 289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9859121427913215e-05,
+      "loss": 1.0003,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985734591411491e-05,
+      "loss": 0.8435,
+      "step": 291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98555593217802e-05,
+      "loss": 0.7917,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985376165190581e-05,
+      "loss": 0.8523,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9851952905494686e-05,
+      "loss": 0.7576,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9850133083555935e-05,
+      "loss": 1.0315,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984830218710485e-05,
+      "loss": 0.8461,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984646021716289e-05,
+      "loss": 0.8213,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984460717475771e-05,
+      "loss": 0.9744,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984274306092314e-05,
+      "loss": 0.8883,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984086787669916e-05,
+      "loss": 0.6365,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983898162313196e-05,
+      "loss": 0.8095,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983708430127389e-05,
+      "loss": 0.7402,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9835175912183495e-05,
+      "loss": 0.8299,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983325645692545e-05,
+      "loss": 0.9618,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983132593657064e-05,
+      "loss": 1.2089,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9829384352196115e-05,
+      "loss": 0.8044,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982743170488511e-05,
+      "loss": 0.8675,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982546799572699e-05,
+      "loss": 0.8224,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9823493225817354e-05,
+      "loss": 1.0417,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982150739625792e-05,
+      "loss": 1.1242,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9819510508156594e-05,
+      "loss": 1.2228,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981750256262745e-05,
+      "loss": 0.746,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9815483560790745e-05,
+      "loss": 0.708,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9813453503772885e-05,
+      "loss": 0.7826,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981141239270645e-05,
+      "loss": 1.0792,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98093602287302e-05,
+      "loss": 1.1573,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980729701298903e-05,
+      "loss": 0.7189,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980522274663404e-05,
+      "loss": 0.789,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9803137430822464e-05,
+      "loss": 0.9377,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9801041066717715e-05,
+      "loss": 1.0512,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9798933655489375e-05,
+      "loss": 1.2052,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9796815198313174e-05,
+      "loss": 0.7481,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979468569637101e-05,
+      "loss": 0.8171,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9792545150850956e-05,
+      "loss": 0.9811,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979039356294722e-05,
+      "loss": 0.8507,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978823093386021e-05,
+      "loss": 0.7904,
+      "step": 326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978605726479645e-05,
+      "loss": 0.7656,
+      "step": 327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978387255696864e-05,
+      "loss": 0.8159,
+      "step": 328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9781676811595664e-05,
+      "loss": 0.7576,
+      "step": 329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9779470029902525e-05,
+      "loss": 0.8647,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97772522131204e-05,
+      "loss": 0.7479,
+      "step": 331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977502336248663e-05,
+      "loss": 0.7308,
+      "step": 332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97727834792447e-05,
+      "loss": 0.7208,
+      "step": 333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977053256464426e-05,
+      "loss": 0.7275,
+      "step": 334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9768270619941105e-05,
+      "loss": 0.8567,
+      "step": 335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976599764639719e-05,
+      "loss": 0.7694,
+      "step": 336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976371364528062e-05,
+      "loss": 0.845,
+      "step": 337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9761418617865654e-05,
+      "loss": 1.1793,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975911256543269e-05,
+      "loss": 1.0907,
+      "step": 339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975679548926831e-05,
+      "loss": 0.9335,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975446739066522e-05,
+      "loss": 0.7224,
+      "step": 341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975212827092227e-05,
+      "loss": 1.0644,
+      "step": 342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974977813134447e-05,
+      "loss": 0.7566,
+      "step": 343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9747416973242986e-05,
+      "loss": 1.1586,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974504479793512e-05,
+      "loss": 0.6739,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974266160674433e-05,
+      "loss": 0.9749,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97402674010002e-05,
+      "loss": 0.6783,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973786218203848e-05,
+      "loss": 0.6852,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973544595120105e-05,
+      "loss": 1.1267,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9733018709835954e-05,
+      "loss": 0.7364,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973058045929735e-05,
+      "loss": 0.8857,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972813120094556e-05,
+      "loss": 1.1353,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972567093614704e-05,
+      "loss": 0.7513,
+      "step": 353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972319966627438e-05,
+      "loss": 1.0838,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972071739270633e-05,
+      "loss": 0.9505,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971822411682776e-05,
+      "loss": 0.7925,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9715719840029674e-05,
+      "loss": 0.7082,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971320456370924e-05,
+      "loss": 0.7304,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971067828926974e-05,
+      "loss": 0.704,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9708141018120595e-05,
+      "loss": 0.9576,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9705592751677366e-05,
+      "loss": 0.8581,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970303349136175e-05,
+      "loss": 0.7407,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970046323860157e-05,
+      "loss": 0.9785,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9697881994830784e-05,
+      "loss": 0.801,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969528976148949e-05,
+      "loss": 1.1321,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9692686540023906e-05,
+      "loss": 0.638,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969007233188639e-05,
+      "loss": 0.7584,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.968744713853542e-05,
+      "loss": 0.8489,
+      "step": 368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.968481096143561e-05,
+      "loss": 0.8772,
+      "step": 369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.968216380205771e-05,
+      "loss": 0.7375,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9679505661878564e-05,
+      "loss": 1.1966,
+      "step": 371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967683654238118e-05,
+      "loss": 0.702,
+      "step": 372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967415644505468e-05,
+      "loss": 1.2024,
+      "step": 373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967146537139429e-05,
+      "loss": 0.7579,
+      "step": 374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9668763322901395e-05,
+      "loss": 0.678,
+      "step": 375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966605030108348e-05,
+      "loss": 0.6954,
+      "step": 376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9663326307454144e-05,
+      "loss": 0.8739,
+      "step": 377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966059134353313e-05,
+      "loss": 0.6641,
+      "step": 378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96578454108463e-05,
+      "loss": 0.894,
+      "step": 379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965508851092561e-05,
+      "loss": 1.2794,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965232064530916e-05,
+      "loss": 0.7168,
+      "step": 381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9649541815541156e-05,
+      "loss": 0.9756,
+      "step": 382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964675202317193e-05,
+      "loss": 0.9371,
+      "step": 383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9643951269757925e-05,
+      "loss": 1.0387,
+      "step": 384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964113955686169e-05,
+      "loss": 0.7503,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9638316886051906e-05,
+      "loss": 0.6991,
+      "step": 386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963548325890335e-05,
+      "loss": 0.8351,
+      "step": 387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963263867699693e-05,
+      "loss": 0.7796,
+      "step": 388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962978314191965e-05,
+      "loss": 1.1948,
+      "step": 389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962691665526463e-05,
+      "loss": 0.8008,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96240392186311e-05,
+      "loss": 0.7528,
+      "step": 391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9621150833624405e-05,
+      "loss": 0.7397,
+      "step": 392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961825150185599e-05,
+      "loss": 0.6502,
+      "step": 393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961534122494341e-05,
+      "loss": 1.2561,
+      "step": 394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961242000451032e-05,
+      "loss": 1.0554,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9609487842186494e-05,
+      "loss": 0.7939,
+      "step": 396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960654473960781e-05,
+      "loss": 0.8415,
+      "step": 397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960359069841624e-05,
+      "loss": 0.7627,
+      "step": 398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960062572025985e-05,
+      "loss": 0.7592,
+      "step": 399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959764980679283e-05,
+      "loss": 0.9275,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959466295967546e-05,
+      "loss": 0.7159,
+      "step": 401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9591665180574116e-05,
+      "loss": 0.8314,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958865647116127e-05,
+      "loss": 0.8446,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958563683311553e-05,
+      "loss": 1.014,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958260626812154e-05,
+      "loss": 0.6973,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957956477787009e-05,
+      "loss": 0.6939,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957651236405803e-05,
+      "loss": 0.7058,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957344902838833e-05,
+      "loss": 0.7922,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957037477257005e-05,
+      "loss": 0.7745,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956728959831832e-05,
+      "loss": 0.7482,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956419350735439e-05,
+      "loss": 0.7583,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956108650140558e-05,
+      "loss": 0.802,
+      "step": 412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955796858220532e-05,
+      "loss": 0.8377,
+      "step": 413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955483975149311e-05,
+      "loss": 0.7578,
+      "step": 414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955170001101454e-05,
+      "loss": 0.7371,
+      "step": 415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9548549362521296e-05,
+      "loss": 0.9613,
+      "step": 416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954538780777115e-05,
+      "loss": 0.7619,
+      "step": 417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954221534852793e-05,
+      "loss": 1.0074,
+      "step": 418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9539031986561595e-05,
+      "loss": 0.6375,
+      "step": 419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953583772364816e-05,
+      "loss": 0.8817,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9532632561569714e-05,
+      "loss": 0.7715,
+      "step": 421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9529416502114435e-05,
+      "loss": 0.6409,
+      "step": 422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952618954707659e-05,
+      "loss": 0.9562,
+      "step": 423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952295169825652e-05,
+      "loss": 0.8943,
+      "step": 424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951970295746063e-05,
+      "loss": 0.7796,
+      "step": 425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9516443326501424e-05,
+      "loss": 0.771,
+      "step": 426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9513172807197465e-05,
+      "loss": 1.1852,
+      "step": 427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9509891401373394e-05,
+      "loss": 1.1499,
+      "step": 428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950659911085993e-05,
+      "loss": 0.7672,
+      "step": 429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950329593749386e-05,
+      "loss": 0.983,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9499981883118054e-05,
+      "loss": 1.1743,
+      "step": 431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9496656949581434e-05,
+      "loss": 0.7415,
+      "step": 432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9493321138739004e-05,
+      "loss": 1.1779,
+      "step": 433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948997445245183e-05,
+      "loss": 1.1951,
+      "step": 434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948661689258705e-05,
+      "loss": 0.7735,
+      "step": 435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9483248461017874e-05,
+      "loss": 0.7678,
+      "step": 436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947986915962357e-05,
+      "loss": 1.0,
+      "step": 437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947647899028947e-05,
+      "loss": 1.2016,
+      "step": 438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947307795490696e-05,
+      "loss": 0.9109,
+      "step": 439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946966605537351e-05,
+      "loss": 0.857,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9466243293592634e-05,
+      "loss": 0.8636,
+      "step": 441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946280967147392e-05,
+      "loss": 0.7635,
+      "step": 442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9459365190933e-05,
+      "loss": 0.7999,
+      "step": 443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945590985389157e-05,
+      "loss": 0.7965,
+      "step": 444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945244366227739e-05,
+      "loss": 0.7626,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944896661802426e-05,
+      "loss": 1.0065,
+      "step": 446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944547872307206e-05,
+      "loss": 0.7009,
+      "step": 447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94419799793667e-05,
+      "loss": 0.6759,
+      "step": 448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9438470388860144e-05,
+      "loss": 1.0455,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943494995351043e-05,
+      "loss": 0.9868,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9431418675281614e-05,
+      "loss": 0.7762,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9427876556143825e-05,
+      "loss": 1.2246,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942432359807324e-05,
+      "loss": 0.8461,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9420759803052064e-05,
+      "loss": 0.7096,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.941718517306857e-05,
+      "loss": 0.8244,
+      "step": 455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.941359971011706e-05,
+      "loss": 1.1067,
+      "step": 456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.941000341619789e-05,
+      "loss": 0.9489,
+      "step": 457
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940639629331745e-05,
+      "loss": 0.7568,
+      "step": 458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9402778343488186e-05,
+      "loss": 0.989,
+      "step": 459
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939914956872857e-05,
+      "loss": 0.7111,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9395509971063105e-05,
+      "loss": 0.7415,
+      "step": 461
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939185955252236e-05,
+      "loss": 0.7131,
+      "step": 462
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938819831514293e-05,
+      "loss": 0.9228,
+      "step": 463
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938452626096743e-05,
+      "loss": 0.8157,
+      "step": 464
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938084339204452e-05,
+      "loss": 1.0536,
+      "step": 465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937714971042891e-05,
+      "loss": 0.8086,
+      "step": 466
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9373445218181316e-05,
+      "loss": 0.7136,
+      "step": 467
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93697299173685e-05,
+      "loss": 0.8678,
+      "step": 468
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936600381006325e-05,
+      "loss": 1.1396,
+      "step": 469
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9362266898344395e-05,
+      "loss": 0.8313,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935851918429676e-05,
+      "loss": 0.9932,
+      "step": 471
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9354760670011235e-05,
+      "loss": 1.1286,
+      "step": 472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935099135758472e-05,
+      "loss": 1.1632,
+      "step": 473
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934721124912013e-05,
+      "loss": 0.6974,
+      "step": 474
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934342034672641e-05,
+      "loss": 0.7362,
+      "step": 475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933961865251853e-05,
+      "loss": 0.8608,
+      "step": 476
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933580616861748e-05,
+      "loss": 0.8188,
+      "step": 477
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933198289715027e-05,
+      "loss": 0.8043,
+      "step": 478
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932814884024992e-05,
+      "loss": 0.837,
+      "step": 479
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932430400005548e-05,
+      "loss": 0.8336,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932044837871202e-05,
+      "loss": 0.9949,
+      "step": 481
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9316581978370587e-05,
+      "loss": 0.678,
+      "step": 482
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93127048011883e-05,
+      "loss": 0.7027,
+      "step": 483
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930881684932824e-05,
+      "loss": 0.7401,
+      "step": 484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9304918124959524e-05,
+      "loss": 0.8944,
+      "step": 485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930100863025728e-05,
+      "loss": 0.6923,
+      "step": 486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9297088367402635e-05,
+      "loss": 1.1972,
+      "step": 487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929315733858274e-05,
+      "loss": 0.9591,
+      "step": 488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928921554599072e-05,
+      "loss": 0.774,
+      "step": 489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928526299182573e-05,
+      "loss": 0.8115,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9281299678292936e-05,
+      "loss": 0.9692,
+      "step": 491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.927732560760349e-05,
+      "loss": 0.7751,
+      "step": 492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9273340781974554e-05,
+      "loss": 0.7453,
+      "step": 493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926934520362927e-05,
+      "loss": 0.8602,
+      "step": 494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926533887479681e-05,
+      "loss": 0.8922,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926132179771234e-05,
+      "loss": 0.813,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925729397461699e-05,
+      "loss": 1.0802,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925325540775792e-05,
+      "loss": 0.6788,
+      "step": 498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924920609938827e-05,
+      "loss": 1.0805,
+      "step": 499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924514605176717e-05,
+      "loss": 0.8728,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924107526715975e-05,
+      "loss": 0.7306,
+      "step": 501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923699374783713e-05,
+      "loss": 0.9238,
+      "step": 502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923290149607641e-05,
+      "loss": 0.822,
+      "step": 503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9228798514160675e-05,
+      "loss": 1.1279,
+      "step": 504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922468480437902e-05,
+      "loss": 0.9832,
+      "step": 505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.92205603690265e-05,
+      "loss": 0.8258,
+      "step": 506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921642521040416e-05,
+      "loss": 0.6503,
+      "step": 507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921227933081904e-05,
+      "loss": 1.1277,
+      "step": 508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.920812273258414e-05,
+      "loss": 1.0405,
+      "step": 509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.920395541801847e-05,
+      "loss": 0.7332,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919977738944697e-05,
+      "loss": 0.8753,
+      "step": 511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919558864920062e-05,
+      "loss": 1.1527,
+      "step": 512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9191389199616327e-05,
+      "loss": 0.7707,
+      "step": 513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9187179043037e-05,
+      "loss": 0.7274,
+      "step": 514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91829581818115e-05,
+      "loss": 0.821,
+      "step": 515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917872661829468e-05,
+      "loss": 0.7613,
+      "step": 516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9174484354847355e-05,
+      "loss": 0.9418,
+      "step": 517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917023139383631e-05,
+      "loss": 0.832,
+      "step": 518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916596773763429e-05,
+      "loss": 0.8123,
+      "step": 519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9161693388620026e-05,
+      "loss": 0.7925,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915740834917819e-05,
+      "loss": 0.7605,
+      "step": 521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915311262169946e-05,
+      "loss": 0.8519,
+      "step": 522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914880620858041e-05,
+      "loss": 0.7093,
+      "step": 523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914448911222365e-05,
+      "loss": 0.8672,
+      "step": 524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91401613350377e-05,
+      "loss": 1.0254,
+      "step": 525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913582287943704e-05,
+      "loss": 1.2612,
+      "step": 526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913147374784214e-05,
+      "loss": 0.8671,
+      "step": 527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912711394267941e-05,
+      "loss": 0.6682,
+      "step": 528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91227434663812e-05,
+      "loss": 0.916,
+      "step": 529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911836232138583e-05,
+      "loss": 0.8071,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911397051013757e-05,
+      "loss": 0.8469,
+      "step": 531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9109568035086634e-05,
+      "loss": 0.7319,
+      "step": 532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91051548986892e-05,
+      "loss": 1.2647,
+      "step": 533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9100731103407376e-05,
+      "loss": 0.7983,
+      "step": 534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909629665170922e-05,
+      "loss": 0.9925,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909185154606875e-05,
+      "loss": 1.143,
+      "step": 536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908739578896592e-05,
+      "loss": 0.7329,
+      "step": 537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9082929382886605e-05,
+      "loss": 0.704,
+      "step": 538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.907845233032267e-05,
+      "loss": 0.6861,
+      "step": 539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9073964633771864e-05,
+      "loss": 1.0396,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9069466295737924e-05,
+      "loss": 0.8201,
+      "step": 541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906495731873048e-05,
+      "loss": 0.7657,
+      "step": 542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.906043770526513e-05,
+      "loss": 0.8369,
+      "step": 543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.905590745786338e-05,
+      "loss": 0.7608,
+      "step": 544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9051366579052706e-05,
+      "loss": 0.7521,
+      "step": 545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.904681507136648e-05,
+      "loss": 0.8141,
+      "step": 546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.904225293734401e-05,
+      "loss": 0.8095,
+      "step": 547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.903768017953055e-05,
+      "loss": 0.8272,
+      "step": 548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.903309680047727e-05,
+      "loss": 0.7812,
+      "step": 549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9028502802741255e-05,
+      "loss": 0.851,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9023898188885534e-05,
+      "loss": 0.7514,
+      "step": 551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901928296147904e-05,
+      "loss": 1.2589,
+      "step": 552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9014657123096656e-05,
+      "loss": 1.0628,
+      "step": 553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9010020676319144e-05,
+      "loss": 1.2465,
+      "step": 554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9005373623733226e-05,
+      "loss": 0.7334,
+      "step": 555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.900071596793151e-05,
+      "loss": 0.753,
+      "step": 556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899604771151253e-05,
+      "loss": 0.8231,
+      "step": 557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899136885708073e-05,
+      "loss": 0.6414,
+      "step": 558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.898667940724648e-05,
+      "loss": 0.7237,
+      "step": 559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8981979364626064e-05,
+      "loss": 0.9983,
+      "step": 560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8977268731841646e-05,
+      "loss": 0.7865,
+      "step": 561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897254751152133e-05,
+      "loss": 0.6834,
+      "step": 562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896781570629911e-05,
+      "loss": 0.6804,
+      "step": 563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8963073318814886e-05,
+      "loss": 0.7447,
+      "step": 564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.895832035171447e-05,
+      "loss": 0.7022,
+      "step": 565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.895355680764956e-05,
+      "loss": 0.7511,
+      "step": 566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.894878268927779e-05,
+      "loss": 0.8347,
+      "step": 567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.894399799926265e-05,
+      "loss": 0.6429,
+      "step": 568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893920274027355e-05,
+      "loss": 0.7558,
+      "step": 569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893439691498581e-05,
+      "loss": 1.1442,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8929580526080616e-05,
+      "loss": 1.2056,
+      "step": 571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.892475357624507e-05,
+      "loss": 0.8003,
+      "step": 572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891991606817214e-05,
+      "loss": 0.831,
+      "step": 573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891506800456072e-05,
+      "loss": 0.8595,
+      "step": 574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891020938811557e-05,
+      "loss": 1.0985,
+      "step": 575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890534022154734e-05,
+      "loss": 0.6513,
+      "step": 576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890046050757257e-05,
+      "loss": 1.1339,
+      "step": 577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889557024891368e-05,
+      "loss": 1.1246,
+      "step": 578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8890669448298976e-05,
+      "loss": 0.9101,
+      "step": 579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.888575810846265e-05,
+      "loss": 0.8023,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8880836232144755e-05,
+      "loss": 1.0996,
+      "step": 581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887590382209126e-05,
+      "loss": 0.6933,
+      "step": 582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887096088105397e-05,
+      "loss": 0.8283,
+      "step": 583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8866007411790586e-05,
+      "loss": 0.8736,
+      "step": 584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8861043417064675e-05,
+      "loss": 0.7262,
+      "step": 585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.885606889964569e-05,
+      "loss": 0.8731,
+      "step": 586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8851083862308934e-05,
+      "loss": 0.8822,
+      "step": 587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8846088307835606e-05,
+      "loss": 0.759,
+      "step": 588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8841082239012735e-05,
+      "loss": 0.8123,
+      "step": 589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883606565863326e-05,
+      "loss": 0.8231,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8831038569495946e-05,
+      "loss": 0.9288,
+      "step": 591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.882600097440545e-05,
+      "loss": 0.9152,
+      "step": 592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8820952876172265e-05,
+      "loss": 0.7177,
+      "step": 593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8815894277612764e-05,
+      "loss": 0.7207,
+      "step": 594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8810825181549165e-05,
+      "loss": 0.9929,
+      "step": 595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8805745590809557e-05,
+      "loss": 0.6831,
+      "step": 596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.880065550822787e-05,
+      "loss": 0.7306,
+      "step": 597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879555493664389e-05,
+      "loss": 1.1546,
+      "step": 598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8790443878903265e-05,
+      "loss": 0.6693,
+      "step": 599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.878532233785747e-05,
+      "loss": 0.8717,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.878019031636387e-05,
+      "loss": 0.6722,
+      "step": 601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8775047817285634e-05,
+      "loss": 0.8082,
+      "step": 602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876989484349179e-05,
+      "loss": 0.8695,
+      "step": 603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876473139785722e-05,
+      "loss": 0.9017,
+      "step": 604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875955748326265e-05,
+      "loss": 0.7512,
+      "step": 605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875437310259464e-05,
+      "loss": 0.5728,
+      "step": 606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8749178258745565e-05,
+      "loss": 0.7281,
+      "step": 607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8743972954613684e-05,
+      "loss": 0.7759,
+      "step": 608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.873875719310306e-05,
+      "loss": 0.8913,
+      "step": 609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.87335309771236e-05,
+      "loss": 0.7114,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872829430959104e-05,
+      "loss": 0.769,
+      "step": 611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872304719342695e-05,
+      "loss": 0.8624,
+      "step": 612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.871778963155874e-05,
+      "loss": 0.8659,
+      "step": 613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8712521626919615e-05,
+      "loss": 1.1681,
+      "step": 614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870724318244864e-05,
+      "loss": 0.805,
+      "step": 615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870195430109069e-05,
+      "loss": 0.9285,
+      "step": 616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.869665498579647e-05,
+      "loss": 0.8052,
+      "step": 617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8691345239522496e-05,
+      "loss": 0.7284,
+      "step": 618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.868602506523111e-05,
+      "loss": 0.7299,
+      "step": 619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.868069446589046e-05,
+      "loss": 0.9373,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.867535344447453e-05,
+      "loss": 0.6675,
+      "step": 621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.867000200396311e-05,
+      "loss": 0.8777,
+      "step": 622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8664640147341807e-05,
+      "loss": 0.8212,
+      "step": 623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865926787760202e-05,
+      "loss": 0.8034,
+      "step": 624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8653885197740985e-05,
+      "loss": 0.7639,
+      "step": 625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.864849211076173e-05,
+      "loss": 0.8351,
+      "step": 626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.864308861967307e-05,
+      "loss": 0.8353,
+      "step": 627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8637674727489676e-05,
+      "loss": 0.8472,
+      "step": 628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.863225043723198e-05,
+      "loss": 1.0043,
+      "step": 629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.862681575192622e-05,
+      "loss": 0.6828,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.862137067460445e-05,
+      "loss": 1.1467,
+      "step": 631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8615915208304516e-05,
+      "loss": 1.0765,
+      "step": 632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8610449356070045e-05,
+      "loss": 0.8581,
+      "step": 633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.860497312095048e-05,
+      "loss": 0.795,
+      "step": 634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8599486506001034e-05,
+      "loss": 0.9148,
+      "step": 635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.859398951428274e-05,
+      "loss": 0.7575,
+      "step": 636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.858848214886239e-05,
+      "loss": 0.7171,
+      "step": 637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.858296441281258e-05,
+      "loss": 1.1899,
+      "step": 638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85774363092117e-05,
+      "loss": 1.0768,
+      "step": 639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8571897841143894e-05,
+      "loss": 0.806,
+      "step": 640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.856634901169913e-05,
+      "loss": 0.7402,
+      "step": 641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.856078982397312e-05,
+      "loss": 1.0508,
+      "step": 642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.855522028106737e-05,
+      "loss": 0.8431,
+      "step": 643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854964038608917e-05,
+      "loss": 0.5879,
+      "step": 644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854405014215156e-05,
+      "loss": 0.7273,
+      "step": 645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85384495523734e-05,
+      "loss": 0.6569,
+      "step": 646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.853283861987927e-05,
+      "loss": 0.8492,
+      "step": 647
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8527217347799554e-05,
+      "loss": 0.8228,
+      "step": 648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85215857392704e-05,
+      "loss": 0.6914,
+      "step": 649
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85159437974337e-05,
+      "loss": 0.7809,
+      "step": 650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851029152543714e-05,
+      "loss": 0.9363,
+      "step": 651
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8504628926434155e-05,
+      "loss": 0.8381,
+      "step": 652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.849895600358396e-05,
+      "loss": 0.7986,
+      "step": 653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.849327276005148e-05,
+      "loss": 0.7765,
+      "step": 654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.848757919900746e-05,
+      "loss": 1.1054,
+      "step": 655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.848187532362836e-05,
+      "loss": 0.6864,
+      "step": 656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.847616113709641e-05,
+      "loss": 0.9609,
+      "step": 657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.84704366425996e-05,
+      "loss": 0.6852,
+      "step": 658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.846470184333165e-05,
+      "loss": 0.8321,
+      "step": 659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.845895674249204e-05,
+      "loss": 1.109,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8453201343286e-05,
+      "loss": 0.7189,
+      "step": 661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.844743564892451e-05,
+      "loss": 1.0585,
+      "step": 662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.844165966262427e-05,
+      "loss": 0.6844,
+      "step": 663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8435873387607754e-05,
+      "loss": 1.1216,
+      "step": 664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.843007682710315e-05,
+      "loss": 0.7669,
+      "step": 665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.84242699843444e-05,
+      "loss": 0.8888,
+      "step": 666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841845286257118e-05,
+      "loss": 0.8028,
+      "step": 667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8412625465028895e-05,
+      "loss": 1.0542,
+      "step": 668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840678779496869e-05,
+      "loss": 0.7995,
+      "step": 669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840093985564743e-05,
+      "loss": 0.8253,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.839508165032772e-05,
+      "loss": 0.8693,
+      "step": 671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8389213182277896e-05,
+      "loss": 0.6903,
+      "step": 672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8383334454772e-05,
+      "loss": 0.6701,
+      "step": 673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837744547108982e-05,
+      "loss": 0.7893,
+      "step": 674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837154623451685e-05,
+      "loss": 0.5461,
+      "step": 675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.836563674834432e-05,
+      "loss": 0.8946,
+      "step": 676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835971701586917e-05,
+      "loss": 0.7192,
+      "step": 677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8353787040394035e-05,
+      "loss": 0.7692,
+      "step": 678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.83478468252273e-05,
+      "loss": 0.7021,
+      "step": 679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8341896373683054e-05,
+      "loss": 0.8666,
+      "step": 680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.833593568908108e-05,
+      "loss": 0.6848,
+      "step": 681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.832996477474689e-05,
+      "loss": 0.8159,
+      "step": 682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.832398363401168e-05,
+      "loss": 1.0001,
+      "step": 683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.831799227021238e-05,
+      "loss": 1.2082,
+      "step": 684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.83119906866916e-05,
+      "loss": 1.1204,
+      "step": 685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.830597888679766e-05,
+      "loss": 0.8361,
+      "step": 686
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.82999568738846e-05,
+      "loss": 1.141,
+      "step": 687
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.829392465131211e-05,
+      "loss": 0.8591,
+      "step": 688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.828788222244561e-05,
+      "loss": 0.8671,
+      "step": 689
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.828182959065623e-05,
+      "loss": 0.8172,
+      "step": 690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8275766759320745e-05,
+      "loss": 0.9499,
+      "step": 691
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.826969373182167e-05,
+      "loss": 0.7069,
+      "step": 692
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.826361051154716e-05,
+      "loss": 0.75,
+      "step": 693
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.82575171018911e-05,
+      "loss": 0.7533,
+      "step": 694
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.825141350625304e-05,
+      "loss": 1.1379,
+      "step": 695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8245299728038196e-05,
+      "loss": 0.9344,
+      "step": 696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.82391757706575e-05,
+      "loss": 0.7611,
+      "step": 697
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8233041637527544e-05,
+      "loss": 0.8373,
+      "step": 698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.82268973320706e-05,
+      "loss": 0.8411,
+      "step": 699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.82207428577146e-05,
+      "loss": 0.7149,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.821457821789317e-05,
+      "loss": 0.7209,
+      "step": 701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8208403416045616e-05,
+      "loss": 1.1321,
+      "step": 702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.820221845561689e-05,
+      "loss": 0.7816,
+      "step": 703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.819602334005761e-05,
+      "loss": 0.6827,
+      "step": 704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8189818072824065e-05,
+      "loss": 0.7707,
+      "step": 705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8183602657378235e-05,
+      "loss": 1.0546,
+      "step": 706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.817737709718772e-05,
+      "loss": 1.014,
+      "step": 707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.817114139572581e-05,
+      "loss": 0.6676,
+      "step": 708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.816489555647144e-05,
+      "loss": 1.1034,
+      "step": 709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815863958290919e-05,
+      "loss": 0.5872,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815237347852932e-05,
+      "loss": 1.009,
+      "step": 711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.814609724682773e-05,
+      "loss": 0.8149,
+      "step": 712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.813981089130595e-05,
+      "loss": 1.1806,
+      "step": 713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8133514415471194e-05,
+      "loss": 0.7089,
+      "step": 714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.812720782283631e-05,
+      "loss": 0.7829,
+      "step": 715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8120891116919764e-05,
+      "loss": 1.0816,
+      "step": 716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.81145643012457e-05,
+      "loss": 1.1378,
+      "step": 717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8108227379343884e-05,
+      "loss": 0.9636,
+      "step": 718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8101880354749725e-05,
+      "loss": 0.8098,
+      "step": 719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8095523231004266e-05,
+      "loss": 0.7409,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.808915601165419e-05,
+      "loss": 0.7868,
+      "step": 721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.80827787002518e-05,
+      "loss": 0.8111,
+      "step": 722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8076391300355046e-05,
+      "loss": 1.0249,
+      "step": 723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.806999381552749e-05,
+      "loss": 1.0879,
+      "step": 724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.806358624933834e-05,
+      "loss": 0.6969,
+      "step": 725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8057168605362404e-05,
+      "loss": 0.8687,
+      "step": 726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8050740887180135e-05,
+      "loss": 0.6966,
+      "step": 727
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.804430309837759e-05,
+      "loss": 0.9125,
+      "step": 728
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8037855242546455e-05,
+      "loss": 0.757,
+      "step": 729
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.803139732328403e-05,
+      "loss": 0.6715,
+      "step": 730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.802492934419323e-05,
+      "loss": 0.7926,
+      "step": 731
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.801845130888256e-05,
+      "loss": 0.8716,
+      "step": 732
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.801196322096619e-05,
+      "loss": 0.7426,
+      "step": 733
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8005465084063844e-05,
+      "loss": 0.8224,
+      "step": 734
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.799895690180087e-05,
+      "loss": 0.8872,
+      "step": 735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7992438677808236e-05,
+      "loss": 0.8171,
+      "step": 736
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.798591041572249e-05,
+      "loss": 0.7499,
+      "step": 737
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.79793721191858e-05,
+      "loss": 0.7214,
+      "step": 738
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7972823791845904e-05,
+      "loss": 0.9491,
+      "step": 739
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.796626543735617e-05,
+      "loss": 1.0601,
+      "step": 740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7959697059375546e-05,
+      "loss": 0.872,
+      "step": 741
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7953118661568555e-05,
+      "loss": 1.0625,
+      "step": 742
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.794653024760534e-05,
+      "loss": 0.6394,
+      "step": 743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.793993182116162e-05,
+      "loss": 0.6572,
+      "step": 744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.793332338591868e-05,
+      "loss": 0.7361,
+      "step": 745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.792670494556343e-05,
+      "loss": 0.7743,
+      "step": 746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7920076503788326e-05,
+      "loss": 0.8341,
+      "step": 747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.791343806429141e-05,
+      "loss": 0.739,
+      "step": 748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7906789630776326e-05,
+      "loss": 0.7142,
+      "step": 749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7900131206952265e-05,
+      "loss": 0.6689,
+      "step": 750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.789346279653401e-05,
+      "loss": 0.8974,
+      "step": 751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.78867844032419e-05,
+      "loss": 1.0784,
+      "step": 752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.788009603080185e-05,
+      "loss": 0.8529,
+      "step": 753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.787339768294536e-05,
+      "loss": 0.6943,
+      "step": 754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.786668936340947e-05,
+      "loss": 0.8946,
+      "step": 755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.785997107593679e-05,
+      "loss": 1.0535,
+      "step": 756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7853242824275496e-05,
+      "loss": 0.8095,
+      "step": 757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.784650461217933e-05,
+      "loss": 0.8102,
+      "step": 758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.783975644340757e-05,
+      "loss": 0.7124,
+      "step": 759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7832998321725065e-05,
+      "loss": 0.6978,
+      "step": 760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7826230250902214e-05,
+      "loss": 0.7923,
+      "step": 761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.781945223471497e-05,
+      "loss": 0.6693,
+      "step": 762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.781266427694482e-05,
+      "loss": 0.9915,
+      "step": 763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.780586638137882e-05,
+      "loss": 0.7634,
+      "step": 764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.779905855180955e-05,
+      "loss": 1.051,
+      "step": 765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.779224079203514e-05,
+      "loss": 1.0469,
+      "step": 766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7785413105859265e-05,
+      "loss": 0.791,
+      "step": 767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.777857549709114e-05,
+      "loss": 0.7424,
+      "step": 768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.777172796954549e-05,
+      "loss": 0.7737,
+      "step": 769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7764870527042615e-05,
+      "loss": 1.026,
+      "step": 770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.77580031734083e-05,
+      "loss": 0.8956,
+      "step": 771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7751125912473907e-05,
+      "loss": 0.7385,
+      "step": 772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.774423874807629e-05,
+      "loss": 0.7072,
+      "step": 773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7737341684057846e-05,
+      "loss": 1.0304,
+      "step": 774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7730434724266486e-05,
+      "loss": 0.7942,
+      "step": 775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.772351787255565e-05,
+      "loss": 0.9107,
+      "step": 776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7716591132784286e-05,
+      "loss": 0.6541,
+      "step": 777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.770965450881686e-05,
+      "loss": 0.6701,
+      "step": 778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.770270800452338e-05,
+      "loss": 0.6099,
+      "step": 779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.769575162377933e-05,
+      "loss": 0.812,
+      "step": 780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.768878537046571e-05,
+      "loss": 0.6676,
+      "step": 781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.768180924846905e-05,
+      "loss": 0.7824,
+      "step": 782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.767482326168137e-05,
+      "loss": 1.1085,
+      "step": 783
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.766782741400019e-05,
+      "loss": 1.1414,
+      "step": 784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.766082170932853e-05,
+      "loss": 0.7086,
+      "step": 785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.765380615157493e-05,
+      "loss": 0.8268,
+      "step": 786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.764678074465341e-05,
+      "loss": 0.7627,
+      "step": 787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.763974549248348e-05,
+      "loss": 0.8206,
+      "step": 788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7632700398990157e-05,
+      "loss": 0.7591,
+      "step": 789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7625645468103947e-05,
+      "loss": 0.7173,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.761858070376083e-05,
+      "loss": 0.7043,
+      "step": 791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.761150610990228e-05,
+      "loss": 0.9266,
+      "step": 792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.760442169047528e-05,
+      "loss": 1.0499,
+      "step": 793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.759732744943223e-05,
+      "loss": 0.8057,
+      "step": 794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7590223390731095e-05,
+      "loss": 0.7924,
+      "step": 795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.758310951833525e-05,
+      "loss": 0.9388,
+      "step": 796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.757598583621357e-05,
+      "loss": 1.0048,
+      "step": 797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.75688523483404e-05,
+      "loss": 0.6718,
+      "step": 798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.756170905869557e-05,
+      "loss": 0.7078,
+      "step": 799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7554555971264354e-05,
+      "loss": 0.8083,
+      "step": 800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.75473930900375e-05,
+      "loss": 0.8051,
+      "step": 801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.754022041901124e-05,
+      "loss": 0.6813,
+      "step": 802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.753303796218723e-05,
+      "loss": 0.7988,
+      "step": 803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.752584572357263e-05,
+      "loss": 0.5521,
+      "step": 804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7518643707180016e-05,
+      "loss": 0.6978,
+      "step": 805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7511431917027447e-05,
+      "loss": 0.686,
+      "step": 806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7504210357138425e-05,
+      "loss": 0.7306,
+      "step": 807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.74969790315419e-05,
+      "loss": 0.6755,
+      "step": 808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.748973794427228e-05,
+      "loss": 0.8734,
+      "step": 809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.74824870993694e-05,
+      "loss": 0.6852,
+      "step": 810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.747522650087857e-05,
+      "loss": 0.9966,
+      "step": 811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7467956152850505e-05,
+      "loss": 0.9462,
+      "step": 812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.746067605934138e-05,
+      "loss": 1.0863,
+      "step": 813
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.745338622441281e-05,
+      "loss": 0.9726,
+      "step": 814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.744608665213184e-05,
+      "loss": 0.6809,
+      "step": 815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.743877734657095e-05,
+      "loss": 0.7368,
+      "step": 816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7431458311808035e-05,
+      "loss": 0.8844,
+      "step": 817
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.742412955192643e-05,
+      "loss": 0.5659,
+      "step": 818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.741679107101491e-05,
+      "loss": 0.8185,
+      "step": 819
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.740944287316765e-05,
+      "loss": 0.8946,
+      "step": 820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.740208496248425e-05,
+      "loss": 0.7391,
+      "step": 821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7394717343069746e-05,
+      "loss": 0.6164,
+      "step": 822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7387340019034575e-05,
+      "loss": 0.7245,
+      "step": 823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.737995299449458e-05,
+      "loss": 0.6826,
+      "step": 824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.737255627357104e-05,
+      "loss": 0.8208,
+      "step": 825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.736514986039063e-05,
+      "loss": 0.7613,
+      "step": 826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.735773375908543e-05,
+      "loss": 0.7369,
+      "step": 827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.735030797379293e-05,
+      "loss": 0.7551,
+      "step": 828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7342872508656025e-05,
+      "loss": 1.0647,
+      "step": 829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7335427367823e-05,
+      "loss": 0.7638,
+      "step": 830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.732797255544756e-05,
+      "loss": 0.7137,
+      "step": 831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.732050807568878e-05,
+      "loss": 0.8472,
+      "step": 832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.731303393271113e-05,
+      "loss": 0.8118,
+      "step": 833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.73055501306845e-05,
+      "loss": 1.2148,
+      "step": 834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.729805667378414e-05,
+      "loss": 0.9633,
+      "step": 835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.729055356619069e-05,
+      "loss": 0.7507,
+      "step": 836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.72830408120902e-05,
+      "loss": 1.173,
+      "step": 837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.727551841567406e-05,
+      "loss": 0.8176,
+      "step": 838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7267986381139074e-05,
+      "loss": 0.8658,
+      "step": 839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.726044471268741e-05,
+      "loss": 0.8833,
+      "step": 840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7252893414526606e-05,
+      "loss": 0.7542,
+      "step": 841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7245332490869574e-05,
+      "loss": 0.6891,
+      "step": 842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.723776194593462e-05,
+      "loss": 0.8273,
+      "step": 843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.723018178394538e-05,
+      "loss": 0.8044,
+      "step": 844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.722259200913088e-05,
+      "loss": 0.7032,
+      "step": 845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7214992625725505e-05,
+      "loss": 0.7578,
+      "step": 846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7207383637968986e-05,
+      "loss": 1.0171,
+      "step": 847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.719976505010644e-05,
+      "loss": 0.729,
+      "step": 848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.719213686638831e-05,
+      "loss": 0.7432,
+      "step": 849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.718449909107042e-05,
+      "loss": 0.7565,
+      "step": 850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.717685172841393e-05,
+      "loss": 1.3074,
+      "step": 851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7169194782685345e-05,
+      "loss": 1.0627,
+      "step": 852
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7161528258156516e-05,
+      "loss": 0.7862,
+      "step": 853
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.715385215910467e-05,
+      "loss": 0.736,
+      "step": 854
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.714616648981232e-05,
+      "loss": 0.7423,
+      "step": 855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.713847125456737e-05,
+      "loss": 0.7825,
+      "step": 856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.713076645766303e-05,
+      "loss": 0.8324,
+      "step": 857
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.712305210339785e-05,
+      "loss": 1.0583,
+      "step": 858
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.711532819607573e-05,
+      "loss": 0.7793,
+      "step": 859
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7107594740005876e-05,
+      "loss": 0.806,
+      "step": 860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.709985173950283e-05,
+      "loss": 1.1024,
+      "step": 861
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.709209919888647e-05,
+      "loss": 1.0043,
+      "step": 862
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.708433712248197e-05,
+      "loss": 0.7856,
+      "step": 863
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7076565514619855e-05,
+      "loss": 0.8291,
+      "step": 864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.706878437963595e-05,
+      "loss": 0.7351,
+      "step": 865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.706099372187139e-05,
+      "loss": 0.8985,
+      "step": 866
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7053193545672645e-05,
+      "loss": 0.8592,
+      "step": 867
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.704538385539147e-05,
+      "loss": 1.0924,
+      "step": 868
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.703756465538495e-05,
+      "loss": 0.6843,
+      "step": 869
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7029735950015455e-05,
+      "loss": 1.0675,
+      "step": 870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.702189774365067e-05,
+      "loss": 0.8457,
+      "step": 871
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7014050040663584e-05,
+      "loss": 1.1539,
+      "step": 872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.700619284543248e-05,
+      "loss": 0.856,
+      "step": 873
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.699832616234093e-05,
+      "loss": 0.982,
+      "step": 874
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6990449995777806e-05,
+      "loss": 1.0576,
+      "step": 875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.698256435013727e-05,
+      "loss": 1.1297,
+      "step": 876
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.697466922981878e-05,
+      "loss": 0.6644,
+      "step": 877
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.696676463922707e-05,
+      "loss": 0.7258,
+      "step": 878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.695885058277215e-05,
+      "loss": 0.8344,
+      "step": 879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.695092706486933e-05,
+      "loss": 0.7792,
+      "step": 880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.69429940899392e-05,
+      "loss": 0.8191,
+      "step": 881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6935051662407585e-05,
+      "loss": 0.6633,
+      "step": 882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6927099786705646e-05,
+      "loss": 0.627,
+      "step": 883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.691913846726977e-05,
+      "loss": 0.6247,
+      "step": 884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6911167708541634e-05,
+      "loss": 0.891,
+      "step": 885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6903187514968155e-05,
+      "loss": 0.7662,
+      "step": 886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.689519789100155e-05,
+      "loss": 0.7623,
+      "step": 887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6887198841099274e-05,
+      "loss": 0.8247,
+      "step": 888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.687919036972405e-05,
+      "loss": 0.8865,
+      "step": 889
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.687117248134385e-05,
+      "loss": 1.0773,
+      "step": 890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6863145180431915e-05,
+      "loss": 0.6775,
+      "step": 891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6855108471466705e-05,
+      "loss": 0.7472,
+      "step": 892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6847062358931964e-05,
+      "loss": 0.6373,
+      "step": 893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6839006847316656e-05,
+      "loss": 0.6739,
+      "step": 894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6830941941115016e-05,
+      "loss": 1.151,
+      "step": 895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6822867644826504e-05,
+      "loss": 0.8789,
+      "step": 896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6814783962955815e-05,
+      "loss": 0.603,
+      "step": 897
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.680669090001287e-05,
+      "loss": 0.9165,
+      "step": 898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.679858846051287e-05,
+      "loss": 1.1505,
+      "step": 899
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6790476648976194e-05,
+      "loss": 0.6503,
+      "step": 900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6782355469928484e-05,
+      "loss": 0.729,
+      "step": 901
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.677422492790059e-05,
+      "loss": 0.609,
+      "step": 902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.676608502742859e-05,
+      "loss": 1.1451,
+      "step": 903
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6757935773053786e-05,
+      "loss": 0.6474,
+      "step": 904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6749777169322695e-05,
+      "loss": 0.7043,
+      "step": 905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6741609220787065e-05,
+      "loss": 0.7203,
+      "step": 906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.673343193200383e-05,
+      "loss": 0.6975,
+      "step": 907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.672524530753516e-05,
+      "loss": 0.7692,
+      "step": 908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.671704935194842e-05,
+      "loss": 0.8544,
+      "step": 909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.670884406981618e-05,
+      "loss": 0.8007,
+      "step": 910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.670062946571623e-05,
+      "loss": 1.1231,
+      "step": 911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.669240554423153e-05,
+      "loss": 1.0024,
+      "step": 912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.668417230995028e-05,
+      "loss": 0.6157,
+      "step": 913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6675929767465835e-05,
+      "loss": 0.6128,
+      "step": 914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.666767792137677e-05,
+      "loss": 0.8447,
+      "step": 915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.665941677628683e-05,
+      "loss": 0.8865,
+      "step": 916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.665114633680497e-05,
+      "loss": 1.0453,
+      "step": 917
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.664286660754531e-05,
+      "loss": 1.0768,
+      "step": 918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.663457759312717e-05,
+      "loss": 0.9529,
+      "step": 919
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6626279298175047e-05,
+      "loss": 1.0018,
+      "step": 920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6617971727318604e-05,
+      "loss": 1.0931,
+      "step": 921
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.660965488519268e-05,
+      "loss": 0.6579,
+      "step": 922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.66013287764373e-05,
+      "loss": 1.2193,
+      "step": 923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.659299340569765e-05,
+      "loss": 0.7105,
+      "step": 924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.658464877762409e-05,
+      "loss": 0.7918,
+      "step": 925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6576294896872137e-05,
+      "loss": 0.9447,
+      "step": 926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.656793176810247e-05,
+      "loss": 0.901,
+      "step": 927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.655955939598093e-05,
+      "loss": 0.8722,
+      "step": 928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.655117778517853e-05,
+      "loss": 0.7148,
+      "step": 929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.65427869403714e-05,
+      "loss": 0.8761,
+      "step": 930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.653438686624085e-05,
+      "loss": 0.7489,
+      "step": 931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.652597756747335e-05,
+      "loss": 0.9324,
+      "step": 932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.651755904876048e-05,
+      "loss": 0.7866,
+      "step": 933
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.650913131479899e-05,
+      "loss": 1.1894,
+      "step": 934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.650069437029078e-05,
+      "loss": 0.703,
+      "step": 935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6492248219942846e-05,
+      "loss": 1.1731,
+      "step": 936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.648379286846737e-05,
+      "loss": 0.7237,
+      "step": 937
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.647532832058163e-05,
+      "loss": 0.7111,
+      "step": 938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.646685458100806e-05,
+      "loss": 1.0908,
+      "step": 939
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6458371654474214e-05,
+      "loss": 0.8396,
+      "step": 940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.644987954571276e-05,
+      "loss": 0.7447,
+      "step": 941
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6441378259461506e-05,
+      "loss": 0.7574,
+      "step": 942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6432867800463365e-05,
+      "loss": 0.9466,
+      "step": 943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.642434817346638e-05,
+      "loss": 0.8235,
+      "step": 944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6415819383223696e-05,
+      "loss": 0.7647,
+      "step": 945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6407281434493595e-05,
+      "loss": 0.7556,
+      "step": 946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.639873433203944e-05,
+      "loss": 1.2369,
+      "step": 947
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6390178080629706e-05,
+      "loss": 0.6519,
+      "step": 948
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.638161268503798e-05,
+      "loss": 0.5847,
+      "step": 949
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6373038150042965e-05,
+      "loss": 0.715,
+      "step": 950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6364454480428436e-05,
+      "loss": 0.9983,
+      "step": 951
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.635586168098327e-05,
+      "loss": 1.1565,
+      "step": 952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.634725975650146e-05,
+      "loss": 0.678,
+      "step": 953
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.633864871178206e-05,
+      "loss": 0.7303,
+      "step": 954
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6330028551629234e-05,
+      "loss": 0.9164,
+      "step": 955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.632139928085221e-05,
+      "loss": 0.9997,
+      "step": 956
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.631276090426532e-05,
+      "loss": 0.8745,
+      "step": 957
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.630411342668796e-05,
+      "loss": 0.794,
+      "step": 958
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6295456852944625e-05,
+      "loss": 0.8718,
+      "step": 959
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.628679118786487e-05,
+      "loss": 1.2222,
+      "step": 960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.627811643628331e-05,
+      "loss": 0.6931,
+      "step": 961
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.626943260303965e-05,
+      "loss": 0.644,
+      "step": 962
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.626073969297865e-05,
+      "loss": 1.2269,
+      "step": 963
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.625203771095015e-05,
+      "loss": 1.0086,
+      "step": 964
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.624332666180904e-05,
+      "loss": 0.8402,
+      "step": 965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.623460655041525e-05,
+      "loss": 0.8145,
+      "step": 966
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.622587738163379e-05,
+      "loss": 0.8606,
+      "step": 967
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6217139160334734e-05,
+      "loss": 0.7257,
+      "step": 968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.620839189139317e-05,
+      "loss": 0.8258,
+      "step": 969
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6199635579689265e-05,
+      "loss": 0.7389,
+      "step": 970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.619087023010823e-05,
+      "loss": 0.5409,
+      "step": 971
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.618209584754027e-05,
+      "loss": 1.0466,
+      "step": 972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.617331243688071e-05,
+      "loss": 0.6168,
+      "step": 973
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.616452000302984e-05,
+      "loss": 0.7352,
+      "step": 974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.615571855089304e-05,
+      "loss": 0.6756,
+      "step": 975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.614690808538066e-05,
+      "loss": 1.074,
+      "step": 976
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.613808861140815e-05,
+      "loss": 0.6643,
+      "step": 977
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6129260133895916e-05,
+      "loss": 0.6748,
+      "step": 978
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6120422657769447e-05,
+      "loss": 0.6798,
+      "step": 979
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6111576187959216e-05,
+      "loss": 0.8383,
+      "step": 980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6102720729400715e-05,
+      "loss": 0.8103,
+      "step": 981
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.609385628703447e-05,
+      "loss": 0.7578,
+      "step": 982
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.608498286580601e-05,
+      "loss": 0.7,
+      "step": 983
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.607610047066586e-05,
+      "loss": 0.9994,
+      "step": 984
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6067209106569576e-05,
+      "loss": 0.6515,
+      "step": 985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6058308778477694e-05,
+      "loss": 0.666,
+      "step": 986
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.604939949135577e-05,
+      "loss": 0.695,
+      "step": 987
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.604048125017435e-05,
+      "loss": 0.7406,
+      "step": 988
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.603155405990896e-05,
+      "loss": 0.7007,
+      "step": 989
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.602261792554017e-05,
+      "loss": 0.9423,
+      "step": 990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.601367285205347e-05,
+      "loss": 0.6814,
+      "step": 991
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6004718844439394e-05,
+      "loss": 0.8667,
+      "step": 992
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.599575590769342e-05,
+      "loss": 0.8132,
+      "step": 993
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.598678404681604e-05,
+      "loss": 0.7928,
+      "step": 994
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.597780326681271e-05,
+      "loss": 1.0479,
+      "step": 995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.596881357269386e-05,
+      "loss": 0.7639,
+      "step": 996
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5959814969474886e-05,
+      "loss": 0.6335,
+      "step": 997
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.595080746217618e-05,
+      "loss": 0.7284,
+      "step": 998
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.594179105582309e-05,
+      "loss": 0.9829,
+      "step": 999
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5932765755445904e-05,
+      "loss": 0.5889,
+      "step": 1000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5923731566079906e-05,
+      "loss": 0.6771,
+      "step": 1001
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.591468849276533e-05,
+      "loss": 1.0482,
+      "step": 1002
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.590563654054735e-05,
+      "loss": 0.626,
+      "step": 1003
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5896575714476116e-05,
+      "loss": 0.7776,
+      "step": 1004
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.588750601960672e-05,
+      "loss": 0.657,
+      "step": 1005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.587842746099919e-05,
+      "loss": 0.7554,
+      "step": 1006
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.586934004371853e-05,
+      "loss": 0.7322,
+      "step": 1007
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.586024377283465e-05,
+      "loss": 0.8075,
+      "step": 1008
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5851138653422415e-05,
+      "loss": 0.7103,
+      "step": 1009
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5842024690561634e-05,
+      "loss": 0.7471,
+      "step": 1010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.583290188933704e-05,
+      "loss": 0.7767,
+      "step": 1011
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.582377025483832e-05,
+      "loss": 0.7958,
+      "step": 1012
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.581462979216003e-05,
+      "loss": 0.732,
+      "step": 1013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5805480506401724e-05,
+      "loss": 0.7572,
+      "step": 1014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.579632240266783e-05,
+      "loss": 0.7649,
+      "step": 1015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5787155486067714e-05,
+      "loss": 1.0056,
+      "step": 1016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.577797976171567e-05,
+      "loss": 0.8356,
+      "step": 1017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.576879523473086e-05,
+      "loss": 0.6457,
+      "step": 1018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.575960191023742e-05,
+      "loss": 0.6407,
+      "step": 1019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5750399793364336e-05,
+      "loss": 0.7213,
+      "step": 1020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.574118888924555e-05,
+      "loss": 0.6245,
+      "step": 1021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.573196920301987e-05,
+      "loss": 0.86,
+      "step": 1022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5722740739831026e-05,
+      "loss": 1.1242,
+      "step": 1023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5713503504827616e-05,
+      "loss": 0.7746,
+      "step": 1024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.570425750316316e-05,
+      "loss": 0.7762,
+      "step": 1025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.569500273999607e-05,
+      "loss": 0.9427,
+      "step": 1026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5685739220489625e-05,
+      "loss": 0.6593,
+      "step": 1027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5676466949811996e-05,
+      "loss": 1.0041,
+      "step": 1028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5667185933136254e-05,
+      "loss": 0.8795,
+      "step": 1029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.565789617564033e-05,
+      "loss": 0.9278,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5648597682507024e-05,
+      "loss": 0.7517,
+      "step": 1031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5639290458924045e-05,
+      "loss": 0.6339,
+      "step": 1032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.562997451008393e-05,
+      "loss": 0.8688,
+      "step": 1033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.562064984118412e-05,
+      "loss": 0.7143,
+      "step": 1034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.561131645742691e-05,
+      "loss": 0.6629,
+      "step": 1035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5601974364019434e-05,
+      "loss": 0.8645,
+      "step": 1036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.559262356617371e-05,
+      "loss": 0.6922,
+      "step": 1037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5583264069106616e-05,
+      "loss": 0.6807,
+      "step": 1038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.557389587803986e-05,
+      "loss": 0.8535,
+      "step": 1039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.556451899820003e-05,
+      "loss": 0.6903,
+      "step": 1040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.555513343481853e-05,
+      "loss": 0.7104,
+      "step": 1041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5545739193131624e-05,
+      "loss": 0.6744,
+      "step": 1042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5536336278380426e-05,
+      "loss": 0.7786,
+      "step": 1043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.552692469581089e-05,
+      "loss": 0.8443,
+      "step": 1044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.551750445067377e-05,
+      "loss": 0.8345,
+      "step": 1045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.55080755482247e-05,
+      "loss": 1.0507,
+      "step": 1046
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.549863799372411e-05,
+      "loss": 0.7887,
+      "step": 1047
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.548919179243729e-05,
+      "loss": 0.7966,
+      "step": 1048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.547973694963431e-05,
+      "loss": 0.6511,
+      "step": 1049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.54702734705901e-05,
+      "loss": 0.7394,
+      "step": 1050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5460801360584386e-05,
+      "loss": 0.7288,
+      "step": 1051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.545132062490173e-05,
+      "loss": 0.775,
+      "step": 1052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5441831268831474e-05,
+      "loss": 0.7291,
+      "step": 1053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.54323332976678e-05,
+      "loss": 0.8454,
+      "step": 1054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.542282671670969e-05,
+      "loss": 0.7071,
+      "step": 1055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.541331153126092e-05,
+      "loss": 1.0028,
+      "step": 1056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.540378774663006e-05,
+      "loss": 0.9742,
+      "step": 1057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5394255368130505e-05,
+      "loss": 0.7345,
+      "step": 1058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.538471440108042e-05,
+      "loss": 0.652,
+      "step": 1059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.537516485080277e-05,
+      "loss": 0.8088,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5365606722625304e-05,
+      "loss": 0.8229,
+      "step": 1061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.535604002188057e-05,
+      "loss": 0.8723,
+      "step": 1062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.534646475390588e-05,
+      "loss": 0.6933,
+      "step": 1063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.533688092404335e-05,
+      "loss": 0.915,
+      "step": 1064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.532728853763985e-05,
+      "loss": 0.9514,
+      "step": 1065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5317687600047024e-05,
+      "loss": 0.6635,
+      "step": 1066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.53080781166213e-05,
+      "loss": 0.6535,
+      "step": 1067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5298460092723875e-05,
+      "loss": 0.7537,
+      "step": 1068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5288833533720695e-05,
+      "loss": 0.6852,
+      "step": 1069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5279198444982485e-05,
+      "loss": 0.6549,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.526955483188471e-05,
+      "loss": 0.7855,
+      "step": 1071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.525990269980762e-05,
+      "loss": 0.8338,
+      "step": 1072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.525024205413618e-05,
+      "loss": 0.8957,
+      "step": 1073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.524057290026013e-05,
+      "loss": 0.6001,
+      "step": 1074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5230895243573954e-05,
+      "loss": 0.7557,
+      "step": 1075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.522120908947687e-05,
+      "loss": 0.9061,
+      "step": 1076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.521151444337286e-05,
+      "loss": 0.6976,
+      "step": 1077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.520181131067061e-05,
+      "loss": 0.8216,
+      "step": 1078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.519209969678357e-05,
+      "loss": 0.719,
+      "step": 1079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.518237960712989e-05,
+      "loss": 0.7128,
+      "step": 1080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.517265104713249e-05,
+      "loss": 0.856,
+      "step": 1081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.516291402221899e-05,
+      "loss": 0.6607,
+      "step": 1082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.515316853782173e-05,
+      "loss": 0.9037,
+      "step": 1083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.514341459937778e-05,
+      "loss": 0.6348,
+      "step": 1084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5133652212328914e-05,
+      "loss": 0.7625,
+      "step": 1085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5123881382121634e-05,
+      "loss": 0.9449,
+      "step": 1086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5114102114207155e-05,
+      "loss": 0.6113,
+      "step": 1087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.510431441404138e-05,
+      "loss": 0.6657,
+      "step": 1088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5094518287084934e-05,
+      "loss": 0.8001,
+      "step": 1089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.508471373880314e-05,
+      "loss": 1.1248,
+      "step": 1090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5074900774666e-05,
+      "loss": 0.8445,
+      "step": 1091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.506507940014825e-05,
+      "loss": 0.8143,
+      "step": 1092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.505524962072928e-05,
+      "loss": 0.8381,
+      "step": 1093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.504541144189319e-05,
+      "loss": 1.1816,
+      "step": 1094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.503556486912876e-05,
+      "loss": 0.8296,
+      "step": 1095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5025709907929456e-05,
+      "loss": 0.6911,
+      "step": 1096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5015846563793414e-05,
+      "loss": 0.7294,
+      "step": 1097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5005974842223466e-05,
+      "loss": 0.747,
+      "step": 1098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.499609474872711e-05,
+      "loss": 0.7977,
+      "step": 1099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4986206288816494e-05,
+      "loss": 0.7424,
+      "step": 1100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.497630946800846e-05,
+      "loss": 0.7461,
+      "step": 1101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.496640429182451e-05,
+      "loss": 0.7875,
+      "step": 1102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4956490765790794e-05,
+      "loss": 0.5636,
+      "step": 1103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.494656889543814e-05,
+      "loss": 0.6019,
+      "step": 1104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.493663868630202e-05,
+      "loss": 1.1541,
+      "step": 1105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4926700143922545e-05,
+      "loss": 0.7124,
+      "step": 1106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.49167532738445e-05,
+      "loss": 0.7164,
+      "step": 1107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4906798081617314e-05,
+      "loss": 0.6971,
+      "step": 1108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.489683457279504e-05,
+      "loss": 0.7163,
+      "step": 1109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4886862752936385e-05,
+      "loss": 0.8227,
+      "step": 1110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.487688262760468e-05,
+      "loss": 0.8309,
+      "step": 1111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.486689420236791e-05,
+      "loss": 0.5682,
+      "step": 1112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4856897482798677e-05,
+      "loss": 1.201,
+      "step": 1113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4846892474474204e-05,
+      "loss": 0.6772,
+      "step": 1114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4836879182976356e-05,
+      "loss": 1.0813,
+      "step": 1115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4826857613891614e-05,
+      "loss": 0.7355,
+      "step": 1116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4816827772811064e-05,
+      "loss": 0.7488,
+      "step": 1117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4806789665330424e-05,
+      "loss": 0.6329,
+      "step": 1118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.479674329705001e-05,
+      "loss": 0.5683,
+      "step": 1119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.478668867357476e-05,
+      "loss": 0.6101,
+      "step": 1120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.477662580051421e-05,
+      "loss": 0.7273,
+      "step": 1121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.476655468348249e-05,
+      "loss": 0.6944,
+      "step": 1122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4756475328098344e-05,
+      "loss": 0.991,
+      "step": 1123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4746387739985114e-05,
+      "loss": 0.632,
+      "step": 1124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.473629192477072e-05,
+      "loss": 0.7627,
+      "step": 1125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.472618788808768e-05,
+      "loss": 0.6557,
+      "step": 1126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.47160756355731e-05,
+      "loss": 0.7329,
+      "step": 1127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.470595517286867e-05,
+      "loss": 0.9925,
+      "step": 1128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.469582650562067e-05,
+      "loss": 0.6156,
+      "step": 1129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.468568963947991e-05,
+      "loss": 0.8435,
+      "step": 1130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4675544580101855e-05,
+      "loss": 0.7152,
+      "step": 1131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4665391333146465e-05,
+      "loss": 0.5943,
+      "step": 1132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.465522990427831e-05,
+      "loss": 0.7202,
+      "step": 1133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.464506029916652e-05,
+      "loss": 0.7839,
+      "step": 1134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4634882523484786e-05,
+      "loss": 0.6392,
+      "step": 1135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4624696582911327e-05,
+      "loss": 0.6884,
+      "step": 1136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.461450248312896e-05,
+      "loss": 0.7077,
+      "step": 1137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.460430022982503e-05,
+      "loss": 1.0972,
+      "step": 1138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.459408982869145e-05,
+      "loss": 0.9367,
+      "step": 1139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.458387128542465e-05,
+      "loss": 0.9916,
+      "step": 1140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4573644605725625e-05,
+      "loss": 0.6911,
+      "step": 1141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.456340979529991e-05,
+      "loss": 0.6894,
+      "step": 1142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.455316685985756e-05,
+      "loss": 0.7271,
+      "step": 1143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.454291580511318e-05,
+      "loss": 1.0735,
+      "step": 1144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.453265663678588e-05,
+      "loss": 0.7903,
+      "step": 1145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.452238936059933e-05,
+      "loss": 0.7613,
+      "step": 1146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.451211398228171e-05,
+      "loss": 0.7736,
+      "step": 1147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.450183050756571e-05,
+      "loss": 0.7594,
+      "step": 1148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.449153894218854e-05,
+      "loss": 0.7454,
+      "step": 1149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.448123929189193e-05,
+      "loss": 0.8851,
+      "step": 1150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4470931562422125e-05,
+      "loss": 0.7667,
+      "step": 1151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4460615759529864e-05,
+      "loss": 0.6583,
+      "step": 1152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.44502918889704e-05,
+      "loss": 0.7573,
+      "step": 1153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4439959956503496e-05,
+      "loss": 1.0746,
+      "step": 1154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.442961996789338e-05,
+      "loss": 0.7976,
+      "step": 1155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.441927192890881e-05,
+      "loss": 1.0986,
+      "step": 1156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4408915845323014e-05,
+      "loss": 0.8129,
+      "step": 1157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4398551722913725e-05,
+      "loss": 0.7964,
+      "step": 1158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.438817956746315e-05,
+      "loss": 0.6855,
+      "step": 1159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.437779938475796e-05,
+      "loss": 0.7267,
+      "step": 1160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.436741118058934e-05,
+      "loss": 0.7813,
+      "step": 1161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4357014960752936e-05,
+      "loss": 0.6957,
+      "step": 1162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4346610731048856e-05,
+      "loss": 0.9782,
+      "step": 1163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4336198497281686e-05,
+      "loss": 0.7556,
+      "step": 1164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.432577826526047e-05,
+      "loss": 0.704,
+      "step": 1165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4315350040798734e-05,
+      "loss": 0.8005,
+      "step": 1166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.430491382971444e-05,
+      "loss": 0.6316,
+      "step": 1167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4294469637830016e-05,
+      "loss": 0.7744,
+      "step": 1168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.428401747097234e-05,
+      "loss": 0.9301,
+      "step": 1169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4273557334972744e-05,
+      "loss": 0.7563,
+      "step": 1170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.426308923566699e-05,
+      "loss": 0.7814,
+      "step": 1171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.425261317889532e-05,
+      "loss": 0.851,
+      "step": 1172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4242129170502374e-05,
+      "loss": 0.9059,
+      "step": 1173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.423163721633725e-05,
+      "loss": 0.8634,
+      "step": 1174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4221137322253466e-05,
+      "loss": 1.051,
+      "step": 1175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.421062949410899e-05,
+      "loss": 1.0706,
+      "step": 1176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4200113737766206e-05,
+      "loss": 0.6677,
+      "step": 1177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4189590059091907e-05,
+      "loss": 0.6101,
+      "step": 1178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.417905846395732e-05,
+      "loss": 0.6283,
+      "step": 1179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.416851895823809e-05,
+      "loss": 1.1027,
+      "step": 1180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.415797154781429e-05,
+      "loss": 0.9146,
+      "step": 1181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.414741623857036e-05,
+      "loss": 0.7584,
+      "step": 1182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.413685303639519e-05,
+      "loss": 0.6942,
+      "step": 1183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4126281947182044e-05,
+      "loss": 0.8166,
+      "step": 1184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4115702976828605e-05,
+      "loss": 0.907,
+      "step": 1185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.410511613123694e-05,
+      "loss": 0.8033,
+      "step": 1186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4094521416313517e-05,
+      "loss": 0.6722,
+      "step": 1187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4083918837969203e-05,
+      "loss": 0.6174,
+      "step": 1188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4073308402119235e-05,
+      "loss": 0.6958,
+      "step": 1189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4062690114683234e-05,
+      "loss": 0.708,
+      "step": 1190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.405206398158522e-05,
+      "loss": 0.8507,
+      "step": 1191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4041430008753575e-05,
+      "loss": 0.7431,
+      "step": 1192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.403078820212106e-05,
+      "loss": 0.6281,
+      "step": 1193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.40201385676248e-05,
+      "loss": 1.0745,
+      "step": 1194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.40094811112063e-05,
+      "loss": 0.5958,
+      "step": 1195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.399881583881141e-05,
+      "loss": 0.7448,
+      "step": 1196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.398814275639036e-05,
+      "loss": 0.7797,
+      "step": 1197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3977461869897725e-05,
+      "loss": 0.7706,
+      "step": 1198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3966773185292436e-05,
+      "loss": 1.1766,
+      "step": 1199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.395607670853778e-05,
+      "loss": 0.664,
+      "step": 1200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.394537244560139e-05,
+      "loss": 0.6104,
+      "step": 1201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.393466040245524e-05,
+      "loss": 0.7527,
+      "step": 1202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.392394058507564e-05,
+      "loss": 0.8523,
+      "step": 1203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.391321299944324e-05,
+      "loss": 0.7553,
+      "step": 1204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.390247765154304e-05,
+      "loss": 0.8412,
+      "step": 1205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.389173454736435e-05,
+      "loss": 0.74,
+      "step": 1206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.388098369290082e-05,
+      "loss": 0.6681,
+      "step": 1207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3870225094150404e-05,
+      "loss": 1.1707,
+      "step": 1208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3859458757115394e-05,
+      "loss": 1.053,
+      "step": 1209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3848684687802405e-05,
+      "loss": 0.875,
+      "step": 1210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3837902892222354e-05,
+      "loss": 0.817,
+      "step": 1211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.382711337639047e-05,
+      "loss": 0.7097,
+      "step": 1212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.381631614632629e-05,
+      "loss": 1.1383,
+      "step": 1213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3805511208053664e-05,
+      "loss": 0.7716,
+      "step": 1214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3794698567600727e-05,
+      "loss": 0.5899,
+      "step": 1215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.378387823099991e-05,
+      "loss": 0.8511,
+      "step": 1216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.377305020428796e-05,
+      "loss": 1.021,
+      "step": 1217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.37622144935059e-05,
+      "loss": 0.6996,
+      "step": 1218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3751371104699036e-05,
+      "loss": 0.665,
+      "step": 1219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.374052004391695e-05,
+      "loss": 0.7393,
+      "step": 1220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.372966131721353e-05,
+      "loss": 0.86,
+      "step": 1221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.371879493064692e-05,
+      "loss": 0.7103,
+      "step": 1222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.370792089027956e-05,
+      "loss": 0.7971,
+      "step": 1223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.369703920217813e-05,
+      "loss": 0.965,
+      "step": 1224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3686149872413595e-05,
+      "loss": 0.7887,
+      "step": 1225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.367525290706117e-05,
+      "loss": 0.668,
+      "step": 1226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.366434831220035e-05,
+      "loss": 0.7946,
+      "step": 1227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.365343609391488e-05,
+      "loss": 0.7352,
+      "step": 1228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3642516258292755e-05,
+      "loss": 1.0594,
+      "step": 1229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.36315888114262e-05,
+      "loss": 0.7405,
+      "step": 1230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.362065375941172e-05,
+      "loss": 0.8637,
+      "step": 1231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.360971110835005e-05,
+      "loss": 0.677,
+      "step": 1232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.359876086434616e-05,
+      "loss": 0.7089,
+      "step": 1233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.358780303350925e-05,
+      "loss": 0.6899,
+      "step": 1234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.357683762195278e-05,
+      "loss": 0.6768,
+      "step": 1235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.356586463579441e-05,
+      "loss": 0.6409,
+      "step": 1236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.355488408115603e-05,
+      "loss": 0.9447,
+      "step": 1237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3543895964163776e-05,
+      "loss": 0.7915,
+      "step": 1238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3532900290947976e-05,
+      "loss": 0.7007,
+      "step": 1239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3521897067643184e-05,
+      "loss": 0.6007,
+      "step": 1240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.351088630038818e-05,
+      "loss": 0.9388,
+      "step": 1241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3499867995325916e-05,
+      "loss": 0.6838,
+      "step": 1242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.348884215860359e-05,
+      "loss": 1.0271,
+      "step": 1243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.347780879637258e-05,
+      "loss": 0.7633,
+      "step": 1244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.346676791478847e-05,
+      "loss": 1.1331,
+      "step": 1245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.345571952001102e-05,
+      "loss": 0.7056,
+      "step": 1246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3444663618204226e-05,
+      "loss": 0.7355,
+      "step": 1247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.343360021553622e-05,
+      "loss": 0.679,
+      "step": 1248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.342252931817936e-05,
+      "loss": 0.7146,
+      "step": 1249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.341145093231016e-05,
+      "loss": 0.7153,
+      "step": 1250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.340036506410932e-05,
+      "loss": 0.8351,
+      "step": 1251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.33892717197617e-05,
+      "loss": 0.6645,
+      "step": 1252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3378170905456384e-05,
+      "loss": 1.158,
+      "step": 1253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3367062627386554e-05,
+      "loss": 0.6057,
+      "step": 1254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3355946891749595e-05,
+      "loss": 0.644,
+      "step": 1255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.334482370474705e-05,
+      "loss": 0.6995,
+      "step": 1256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.33336930725846e-05,
+      "loss": 0.7378,
+      "step": 1257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3322555001472105e-05,
+      "loss": 1.0347,
+      "step": 1258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.331140949762357e-05,
+      "loss": 0.7632,
+      "step": 1259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.330025656725712e-05,
+      "loss": 0.7953,
+      "step": 1260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.328909621659506e-05,
+      "loss": 0.7228,
+      "step": 1261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.327792845186382e-05,
+      "loss": 0.75,
+      "step": 1262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.326675327929396e-05,
+      "loss": 0.6857,
+      "step": 1263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.325557070512018e-05,
+      "loss": 0.6447,
+      "step": 1264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.32443807355813e-05,
+      "loss": 0.765,
+      "step": 1265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.323318337692029e-05,
+      "loss": 1.0329,
+      "step": 1266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.322197863538421e-05,
+      "loss": 0.7464,
+      "step": 1267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3210766517224255e-05,
+      "loss": 0.7159,
+      "step": 1268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.319954702869575e-05,
+      "loss": 0.7242,
+      "step": 1269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.31883201760581e-05,
+      "loss": 0.7215,
+      "step": 1270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3177085965574846e-05,
+      "loss": 1.1457,
+      "step": 1271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.316584440351363e-05,
+      "loss": 0.6814,
+      "step": 1272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.315459549614616e-05,
+      "loss": 1.1683,
+      "step": 1273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3143339249748307e-05,
+      "loss": 0.7642,
+      "step": 1274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.313207567059998e-05,
+      "loss": 0.7931,
+      "step": 1275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.312080476498521e-05,
+      "loss": 0.8169,
+      "step": 1276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.31095265391921e-05,
+      "loss": 0.6468,
+      "step": 1277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.309824099951284e-05,
+      "loss": 0.6602,
+      "step": 1278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.30869481522437e-05,
+      "loss": 0.6881,
+      "step": 1279
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.307564800368504e-05,
+      "loss": 0.63,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3064340560141266e-05,
+      "loss": 0.6495,
+      "step": 1281
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3053025827920894e-05,
+      "loss": 0.6178,
+      "step": 1282
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3041703813336466e-05,
+      "loss": 0.6973,
+      "step": 1283
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.303037452270461e-05,
+      "loss": 0.8021,
+      "step": 1284
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3019037962346005e-05,
+      "loss": 0.6777,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3007694138585394e-05,
+      "loss": 0.8201,
+      "step": 1286
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.299634305775156e-05,
+      "loss": 0.7041,
+      "step": 1287
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.298498472617735e-05,
+      "loss": 0.9735,
+      "step": 1288
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2973619150199634e-05,
+      "loss": 1.1432,
+      "step": 1289
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.296224633615935e-05,
+      "loss": 0.6962,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2950866290401455e-05,
+      "loss": 0.6889,
+      "step": 1291
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.293947901927495e-05,
+      "loss": 0.8499,
+      "step": 1292
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.292808452913286e-05,
+      "loss": 1.1809,
+      "step": 1293
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.291668282633225e-05,
+      "loss": 0.6297,
+      "step": 1294
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.290527391723419e-05,
+      "loss": 0.8224,
+      "step": 1295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.289385780820379e-05,
+      "loss": 0.6512,
+      "step": 1296
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2882434505610157e-05,
+      "loss": 0.8382,
+      "step": 1297
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.287100401582643e-05,
+      "loss": 0.8477,
+      "step": 1298
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2859566345229765e-05,
+      "loss": 1.0049,
+      "step": 1299
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.284812150020128e-05,
+      "loss": 0.6966,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.283666948712614e-05,
+      "loss": 0.6407,
+      "step": 1301
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.28252103123935e-05,
+      "loss": 0.6457,
+      "step": 1302
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2813743982396495e-05,
+      "loss": 0.7559,
+      "step": 1303
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.280227050353227e-05,
+      "loss": 0.6865,
+      "step": 1304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.279078988220193e-05,
+      "loss": 0.7577,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.277930212481061e-05,
+      "loss": 0.6573,
+      "step": 1306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.276780723776739e-05,
+      "loss": 0.7182,
+      "step": 1307
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.275630522748534e-05,
+      "loss": 0.9924,
+      "step": 1308
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2744796100381496e-05,
+      "loss": 1.0576,
+      "step": 1309
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2733279862876885e-05,
+      "loss": 0.7352,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.272175652139647e-05,
+      "loss": 0.7553,
+      "step": 1311
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.271022608236922e-05,
+      "loss": 0.7569,
+      "step": 1312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.269868855222801e-05,
+      "loss": 0.7128,
+      "step": 1313
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2687143937409715e-05,
+      "loss": 0.8804,
+      "step": 1314
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.267559224435515e-05,
+      "loss": 0.8217,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2664033479509066e-05,
+      "loss": 0.7999,
+      "step": 1316
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.265246764932017e-05,
+      "loss": 0.6945,
+      "step": 1317
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2640894760241125e-05,
+      "loss": 0.7913,
+      "step": 1318
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.26293148187285e-05,
+      "loss": 0.6657,
+      "step": 1319
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.261772783124283e-05,
+      "loss": 0.7478,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.260613380424856e-05,
+      "loss": 0.7013,
+      "step": 1321
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2594532744214066e-05,
+      "loss": 0.7195,
+      "step": 1322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2582924657611655e-05,
+      "loss": 0.7716,
+      "step": 1323
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2571309550917556e-05,
+      "loss": 0.7816,
+      "step": 1324
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.25596874306119e-05,
+      "loss": 0.7089,
+      "step": 1325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2548058303178746e-05,
+      "loss": 0.6955,
+      "step": 1326
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.253642217510606e-05,
+      "loss": 0.8031,
+      "step": 1327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2524779052885695e-05,
+      "loss": 0.7665,
+      "step": 1328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.251312894301344e-05,
+      "loss": 0.8269,
+      "step": 1329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2501471851988945e-05,
+      "loss": 0.7332,
+      "step": 1330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2489807786315783e-05,
+      "loss": 0.9061,
+      "step": 1331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2478136752501416e-05,
+      "loss": 0.9878,
+      "step": 1332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2466458757057174e-05,
+      "loss": 0.7397,
+      "step": 1333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.245477380649829e-05,
+      "loss": 0.7912,
+      "step": 1334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.244308190734387e-05,
+      "loss": 0.5952,
+      "step": 1335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.243138306611688e-05,
+      "loss": 0.688,
+      "step": 1336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2419677289344206e-05,
+      "loss": 1.0705,
+      "step": 1337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.240796458355655e-05,
+      "loss": 0.9856,
+      "step": 1338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.239624495528852e-05,
+      "loss": 0.9735,
+      "step": 1339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.238451841107855e-05,
+      "loss": 0.7259,
+      "step": 1340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.237278495746897e-05,
+      "loss": 0.826,
+      "step": 1341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.236104460100594e-05,
+      "loss": 0.711,
+      "step": 1342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.234929734823947e-05,
+      "loss": 0.662,
+      "step": 1343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2337543205723436e-05,
+      "loss": 0.9488,
+      "step": 1344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.232578218001554e-05,
+      "loss": 0.749,
+      "step": 1345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.231401427767734e-05,
+      "loss": 0.8008,
+      "step": 1346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.230223950527421e-05,
+      "loss": 0.5977,
+      "step": 1347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.229045786937538e-05,
+      "loss": 0.725,
+      "step": 1348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.227866937655388e-05,
+      "loss": 1.0634,
+      "step": 1349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2266874033386606e-05,
+      "loss": 0.6753,
+      "step": 1350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.225507184645424e-05,
+      "loss": 0.8622,
+      "step": 1351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.224326282234129e-05,
+      "loss": 0.8085,
+      "step": 1352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2231446967636104e-05,
+      "loss": 0.77,
+      "step": 1353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.22196242889308e-05,
+      "loss": 0.7898,
+      "step": 1354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.220779479282134e-05,
+      "loss": 0.738,
+      "step": 1355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.219595848590746e-05,
+      "loss": 0.6879,
+      "step": 1356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.218411537479271e-05,
+      "loss": 0.5408,
+      "step": 1357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.217226546608444e-05,
+      "loss": 0.7985,
+      "step": 1358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2160408766393784e-05,
+      "loss": 0.7817,
+      "step": 1359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.214854528233568e-05,
+      "loss": 0.6825,
+      "step": 1360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2136675020528824e-05,
+      "loss": 0.6275,
+      "step": 1361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.212479798759572e-05,
+      "loss": 1.1094,
+      "step": 1362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2112914190162624e-05,
+      "loss": 0.728,
+      "step": 1363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2101023634859596e-05,
+      "loss": 0.7625,
+      "step": 1364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.208912632832045e-05,
+      "loss": 1.0116,
+      "step": 1365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.207722227718275e-05,
+      "loss": 0.7348,
+      "step": 1366
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.206531148808785e-05,
+      "loss": 0.8411,
+      "step": 1367
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.205339396768086e-05,
+      "loss": 0.664,
+      "step": 1368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2041469722610634e-05,
+      "loss": 0.6549,
+      "step": 1369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.202953875952978e-05,
+      "loss": 0.7653,
+      "step": 1370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.201760108509466e-05,
+      "loss": 0.7429,
+      "step": 1371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.200565670596537e-05,
+      "loss": 0.6349,
+      "step": 1372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1993705628805756e-05,
+      "loss": 0.6594,
+      "step": 1373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.19817478602834e-05,
+      "loss": 0.7084,
+      "step": 1374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1969783407069615e-05,
+      "loss": 0.648,
+      "step": 1375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1957812275839444e-05,
+      "loss": 0.6161,
+      "step": 1376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.194583447327166e-05,
+      "loss": 0.7177,
+      "step": 1377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.193385000604873e-05,
+      "loss": 0.7779,
+      "step": 1378
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.192185888085689e-05,
+      "loss": 0.6376,
+      "step": 1379
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1909861104386054e-05,
+      "loss": 0.7742,
+      "step": 1380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.189785668332986e-05,
+      "loss": 0.8217,
+      "step": 1381
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.188584562438564e-05,
+      "loss": 0.7664,
+      "step": 1382
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1873827934254444e-05,
+      "loss": 0.7844,
+      "step": 1383
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.186180361964101e-05,
+      "loss": 0.8697,
+      "step": 1384
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1849772687253786e-05,
+      "loss": 0.6985,
+      "step": 1385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.18377351438049e-05,
+      "loss": 0.8088,
+      "step": 1386
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.182569099601017e-05,
+      "loss": 1.1183,
+      "step": 1387
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.18136402505891e-05,
+      "loss": 0.8955,
+      "step": 1388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.180158291426488e-05,
+      "loss": 0.725,
+      "step": 1389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.178951899376437e-05,
+      "loss": 0.5945,
+      "step": 1390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.177744849581811e-05,
+      "loss": 0.7712,
+      "step": 1391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.17653714271603e-05,
+      "loss": 0.7652,
+      "step": 1392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1753287794528815e-05,
+      "loss": 0.8642,
+      "step": 1393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.174119760466519e-05,
+      "loss": 1.1065,
+      "step": 1394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1729100864314616e-05,
+      "loss": 0.6008,
+      "step": 1395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.171699758022593e-05,
+      "loss": 1.0693,
+      "step": 1396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.170488775915164e-05,
+      "loss": 0.7269,
+      "step": 1397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1692771407847885e-05,
+      "loss": 0.7473,
+      "step": 1398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.168064853307447e-05,
+      "loss": 0.7224,
+      "step": 1399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1668519141594787e-05,
+      "loss": 0.6513,
+      "step": 1400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.165638324017593e-05,
+      "loss": 0.994,
+      "step": 1401
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1644240835588584e-05,
+      "loss": 0.6005,
+      "step": 1402
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.163209193460707e-05,
+      "loss": 0.7351,
+      "step": 1403
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1619936544009327e-05,
+      "loss": 0.9977,
+      "step": 1404
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.160777467057694e-05,
+      "loss": 0.7588,
+      "step": 1405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.159560632109508e-05,
+      "loss": 0.6009,
+      "step": 1406
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.158343150235255e-05,
+      "loss": 1.0755,
+      "step": 1407
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.157125022114175e-05,
+      "loss": 0.7129,
+      "step": 1408
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1559062484258714e-05,
+      "loss": 0.6198,
+      "step": 1409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.154686829850303e-05,
+      "loss": 0.8365,
+      "step": 1410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.153466767067793e-05,
+      "loss": 0.7924,
+      "step": 1411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1522460607590197e-05,
+      "loss": 0.8323,
+      "step": 1412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1510247116050247e-05,
+      "loss": 0.7146,
+      "step": 1413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.149802720287206e-05,
+      "loss": 0.6634,
+      "step": 1414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.148580087487319e-05,
+      "loss": 0.6058,
+      "step": 1415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1473568138874794e-05,
+      "loss": 0.6885,
+      "step": 1416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.146132900170159e-05,
+      "loss": 0.7474,
+      "step": 1417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1449083470181853e-05,
+      "loss": 0.5932,
+      "step": 1418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.143683155114746e-05,
+      "loss": 0.6295,
+      "step": 1419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1424573251433826e-05,
+      "loss": 0.8051,
+      "step": 1420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1412308577879933e-05,
+      "loss": 1.0397,
+      "step": 1421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.140003753732831e-05,
+      "loss": 0.7419,
+      "step": 1422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1387760136625054e-05,
+      "loss": 0.6348,
+      "step": 1423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.137547638261981e-05,
+      "loss": 0.7779,
+      "step": 1424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.136318628216575e-05,
+      "loss": 0.818,
+      "step": 1425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1350889842119595e-05,
+      "loss": 0.7481,
+      "step": 1426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.133858706934162e-05,
+      "loss": 0.7812,
+      "step": 1427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.13262779706956e-05,
+      "loss": 0.6222,
+      "step": 1428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.131396255304887e-05,
+      "loss": 0.7932,
+      "step": 1429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.130164082327228e-05,
+      "loss": 1.1398,
+      "step": 1430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1289312788240185e-05,
+      "loss": 0.6852,
+      "step": 1431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.127697845483048e-05,
+      "loss": 0.7048,
+      "step": 1432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.126463782992457e-05,
+      "loss": 0.7096,
+      "step": 1433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1252290920407364e-05,
+      "loss": 1.0845,
+      "step": 1434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.123993773316728e-05,
+      "loss": 0.7298,
+      "step": 1435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1227578275096236e-05,
+      "loss": 0.7155,
+      "step": 1436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.121521255308965e-05,
+      "loss": 1.1025,
+      "step": 1437
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1202840574046434e-05,
+      "loss": 0.9538,
+      "step": 1438
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.119046234486899e-05,
+      "loss": 0.6291,
+      "step": 1439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.117807787246322e-05,
+      "loss": 1.0259,
+      "step": 1440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1165687163738485e-05,
+      "loss": 0.8972,
+      "step": 1441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1153290225607644e-05,
+      "loss": 0.7896,
+      "step": 1442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.114088706498702e-05,
+      "loss": 0.6589,
+      "step": 1443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.112847768879643e-05,
+      "loss": 0.7497,
+      "step": 1444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.111606210395912e-05,
+      "loss": 0.6583,
+      "step": 1445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.110364031740184e-05,
+      "loss": 0.8205,
+      "step": 1446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.109121233605476e-05,
+      "loss": 0.7532,
+      "step": 1447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1078778166851556e-05,
+      "loss": 0.8571,
+      "step": 1448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1066337816729306e-05,
+      "loss": 0.5367,
+      "step": 1449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.105389129262857e-05,
+      "loss": 0.8651,
+      "step": 1450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.104143860149334e-05,
+      "loss": 0.6618,
+      "step": 1451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1028979750271034e-05,
+      "loss": 0.7262,
+      "step": 1452
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.101651474591254e-05,
+      "loss": 0.7338,
+      "step": 1453
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.100404359537216e-05,
+      "loss": 0.7364,
+      "step": 1454
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.099156630560762e-05,
+      "loss": 0.7131,
+      "step": 1455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.097908288358008e-05,
+      "loss": 0.6316,
+      "step": 1456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.096659333625412e-05,
+      "loss": 0.6971,
+      "step": 1457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0954097670597733e-05,
+      "loss": 1.0466,
+      "step": 1458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.094159589358233e-05,
+      "loss": 1.0524,
+      "step": 1459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.092908801218273e-05,
+      "loss": 0.703,
+      "step": 1460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0916574033377154e-05,
+      "loss": 0.6598,
+      "step": 1461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.090405396414724e-05,
+      "loss": 1.141,
+      "step": 1462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0891527811477995e-05,
+      "loss": 1.0527,
+      "step": 1463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.087899558235785e-05,
+      "loss": 0.7808,
+      "step": 1464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0866457283778605e-05,
+      "loss": 0.8794,
+      "step": 1465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0853912922735466e-05,
+      "loss": 0.6197,
+      "step": 1466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0841362506227e-05,
+      "loss": 1.0461,
+      "step": 1467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0828806041255155e-05,
+      "loss": 0.6557,
+      "step": 1468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.081624353482528e-05,
+      "loss": 1.1616,
+      "step": 1469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.080367499394606e-05,
+      "loss": 1.0537,
+      "step": 1470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.079110042562958e-05,
+      "loss": 0.664,
+      "step": 1471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.077851983689124e-05,
+      "loss": 0.8133,
+      "step": 1472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.076593323474986e-05,
+      "loss": 0.7172,
+      "step": 1473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.075334062622757e-05,
+      "loss": 1.0504,
+      "step": 1474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0740742018349854e-05,
+      "loss": 0.6311,
+      "step": 1475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.072813741814557e-05,
+      "loss": 0.591,
+      "step": 1476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.071552683264689e-05,
+      "loss": 0.7117,
+      "step": 1477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.070291026888934e-05,
+      "loss": 0.6399,
+      "step": 1478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.069028773391179e-05,
+      "loss": 0.6334,
+      "step": 1479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.067765923475642e-05,
+      "loss": 0.7182,
+      "step": 1480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.066502477846875e-05,
+      "loss": 0.8982,
+      "step": 1481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.065238437209763e-05,
+      "loss": 0.6964,
+      "step": 1482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.063973802269521e-05,
+      "loss": 0.7953,
+      "step": 1483
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.062708573731696e-05,
+      "loss": 0.8483,
+      "step": 1484
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.06144275230217e-05,
+      "loss": 0.6944,
+      "step": 1485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.060176338687149e-05,
+      "loss": 1.1698,
+      "step": 1486
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0589093335931745e-05,
+      "loss": 1.1081,
+      "step": 1487
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.057641737727117e-05,
+      "loss": 0.7581,
+      "step": 1488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.056373551796176e-05,
+      "loss": 0.8995,
+      "step": 1489
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0551047765078784e-05,
+      "loss": 0.7516,
+      "step": 1490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0538354125700855e-05,
+      "loss": 0.9009,
+      "step": 1491
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.052565460690979e-05,
+      "loss": 0.7284,
+      "step": 1492
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0512949215790762e-05,
+      "loss": 1.0562,
+      "step": 1493
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0500237959432177e-05,
+      "loss": 0.9646,
+      "step": 1494
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0487520844925716e-05,
+      "loss": 0.6823,
+      "step": 1495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0474797879366343e-05,
+      "loss": 0.7399,
+      "step": 1496
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0462069069852293e-05,
+      "loss": 0.9257,
+      "step": 1497
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0449334423485025e-05,
+      "loss": 0.7549,
+      "step": 1498
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.043659394736928e-05,
+      "loss": 0.6237,
+      "step": 1499
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0423847648613064e-05,
+      "loss": 0.5873,
+      "step": 1500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0411095534327598e-05,
+      "loss": 0.8546,
+      "step": 1501
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.039833761162738e-05,
+      "loss": 0.9765,
+      "step": 1502
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0385573887630127e-05,
+      "loss": 0.719,
+      "step": 1503
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0372804369456804e-05,
+      "loss": 0.612,
+      "step": 1504
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0360029064231596e-05,
+      "loss": 0.7087,
+      "step": 1505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0347247979081936e-05,
+      "loss": 0.6157,
+      "step": 1506
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.033446112113846e-05,
+      "loss": 0.7816,
+      "step": 1507
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.032166849753505e-05,
+      "loss": 0.7217,
+      "step": 1508
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0308870115408776e-05,
+      "loss": 0.8964,
+      "step": 1509
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0296065981899946e-05,
+      "loss": 0.6955,
+      "step": 1510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.028325610415205e-05,
+      "loss": 1.0143,
+      "step": 1511
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.027044048931181e-05,
+      "loss": 0.731,
+      "step": 1512
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0257619144529132e-05,
+      "loss": 0.938,
+      "step": 1513
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.024479207695713e-05,
+      "loss": 1.0682,
+      "step": 1514
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0231959293752103e-05,
+      "loss": 0.6291,
+      "step": 1515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0219120802073534e-05,
+      "loss": 0.5977,
+      "step": 1516
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0206276609084103e-05,
+      "loss": 0.7166,
+      "step": 1517
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0193426721949656e-05,
+      "loss": 0.676,
+      "step": 1518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0180571147839238e-05,
+      "loss": 0.8028,
+      "step": 1519
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.016770989392505e-05,
+      "loss": 0.964,
+      "step": 1520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0154842967382458e-05,
+      "loss": 0.8181,
+      "step": 1521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0141970375390008e-05,
+      "loss": 0.7944,
+      "step": 1522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.012909212512939e-05,
+      "loss": 0.6417,
+      "step": 1523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0116208223785474e-05,
+      "loss": 0.6571,
+      "step": 1524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.010331867854625e-05,
+      "loss": 1.0602,
+      "step": 1525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.009042349660289e-05,
+      "loss": 0.6767,
+      "step": 1526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0077522685149686e-05,
+      "loss": 0.8087,
+      "step": 1527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0064616251384095e-05,
+      "loss": 0.8652,
+      "step": 1528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.005170420250668e-05,
+      "loss": 0.7755,
+      "step": 1529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0038786545721162e-05,
+      "loss": 0.7544,
+      "step": 1530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0025863288234377e-05,
+      "loss": 0.7929,
+      "step": 1531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0012934437256295e-05,
+      "loss": 0.7817,
+      "step": 1532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6813,
+      "step": 1533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.99870599836817e-05,
+      "loss": 0.7164,
+      "step": 1534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9974114395520703e-05,
+      "loss": 0.6085,
+      "step": 1535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.996116324273943e-05,
+      "loss": 0.9148,
+      "step": 1536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9948206532563423e-05,
+      "loss": 1.1268,
+      "step": 1537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9935244272221296e-05,
+      "loss": 0.8121,
+      "step": 1538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9922276468944793e-05,
+      "loss": 1.281,
+      "step": 1539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9909303129968708e-05,
+      "loss": 0.6723,
+      "step": 1540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9896324262530963e-05,
+      "loss": 0.892,
+      "step": 1541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9883339873872547e-05,
+      "loss": 0.7493,
+      "step": 1542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9870349971237537e-05,
+      "loss": 0.7347,
+      "step": 1543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.985735456187306e-05,
+      "loss": 0.765,
+      "step": 1544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.984435365302935e-05,
+      "loss": 0.705,
+      "step": 1545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9831347251959693e-05,
+      "loss": 0.8225,
+      "step": 1546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9818335365920443e-05,
+      "loss": 0.6626,
+      "step": 1547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9805318002171003e-05,
+      "loss": 1.0392,
+      "step": 1548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9792295167973856e-05,
+      "loss": 1.1625,
+      "step": 1549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9779266870594505e-05,
+      "loss": 0.6425,
+      "step": 1550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.976623311730153e-05,
+      "loss": 1.0906,
+      "step": 1551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975319391536654e-05,
+      "loss": 0.8609,
+      "step": 1552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9740149272064185e-05,
+      "loss": 0.7263,
+      "step": 1553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9727099194672145e-05,
+      "loss": 0.8275,
+      "step": 1554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9714043690471163e-05,
+      "loss": 0.6066,
+      "step": 1555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9700982766744957e-05,
+      "loss": 1.0091,
+      "step": 1556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9687916430780314e-05,
+      "loss": 1.0649,
+      "step": 1557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.967484468986702e-05,
+      "loss": 0.6803,
+      "step": 1558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9661767551297876e-05,
+      "loss": 0.9885,
+      "step": 1559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9648685022368705e-05,
+      "loss": 0.7413,
+      "step": 1560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9635597110378334e-05,
+      "loss": 0.7895,
+      "step": 1561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9622503822628577e-05,
+      "loss": 0.9538,
+      "step": 1562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9609405166424273e-05,
+      "loss": 0.6455,
+      "step": 1563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9596301149073238e-05,
+      "loss": 0.8332,
+      "step": 1564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9583191777886284e-05,
+      "loss": 0.7645,
+      "step": 1565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.957007706017722e-05,
+      "loss": 0.6992,
+      "step": 1566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9556957003262812e-05,
+      "loss": 0.7034,
+      "step": 1567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9543831614462828e-05,
+      "loss": 0.7867,
+      "step": 1568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.953070090110001e-05,
+      "loss": 0.9598,
+      "step": 1569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.951756487050006e-05,
+      "loss": 0.8402,
+      "step": 1570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.950442352999165e-05,
+      "loss": 1.0769,
+      "step": 1571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.949127688690641e-05,
+      "loss": 0.6498,
+      "step": 1572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9478124948578947e-05,
+      "loss": 0.6375,
+      "step": 1573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9464967722346797e-05,
+      "loss": 1.0612,
+      "step": 1574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.945180521555047e-05,
+      "loss": 0.7638,
+      "step": 1575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9438637435533395e-05,
+      "loss": 0.7007,
+      "step": 1576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.942546438964197e-05,
+      "loss": 0.7293,
+      "step": 1577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9412286085225516e-05,
+      "loss": 0.9674,
+      "step": 1578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9399102529636288e-05,
+      "loss": 0.9455,
+      "step": 1579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9385913730229477e-05,
+      "loss": 0.5525,
+      "step": 1580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.937271969436319e-05,
+      "loss": 1.0862,
+      "step": 1581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.935952042939847e-05,
+      "loss": 0.7177,
+      "step": 1582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.934631594269926e-05,
+      "loss": 0.9176,
+      "step": 1583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9333106241632433e-05,
+      "loss": 0.7112,
+      "step": 1584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9319891333567754e-05,
+      "loss": 0.6893,
+      "step": 1585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9306671225877913e-05,
+      "loss": 0.9194,
+      "step": 1586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9293445925938484e-05,
+      "loss": 1.1536,
+      "step": 1587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9280215441127944e-05,
+      "loss": 0.9121,
+      "step": 1588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.926697977882766e-05,
+      "loss": 0.842,
+      "step": 1589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9253738946421904e-05,
+      "loss": 0.7735,
+      "step": 1590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9240492951297797e-05,
+      "loss": 0.6418,
+      "step": 1591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9227241800845375e-05,
+      "loss": 0.841,
+      "step": 1592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9213985502457533e-05,
+      "loss": 0.9013,
+      "step": 1593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.920072406353005e-05,
+      "loss": 0.6557,
+      "step": 1594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.918745749146155e-05,
+      "loss": 0.6216,
+      "step": 1595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9174185793653553e-05,
+      "loss": 0.8605,
+      "step": 1596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9160908977510406e-05,
+      "loss": 0.9948,
+      "step": 1597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9147627050439334e-05,
+      "loss": 0.9313,
+      "step": 1598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9134340019850406e-05,
+      "loss": 0.9603,
+      "step": 1599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.912104789315654e-05,
+      "loss": 0.8794,
+      "step": 1600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9107750677773504e-05,
+      "loss": 0.7087,
+      "step": 1601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.909444838111988e-05,
+      "loss": 0.798,
+      "step": 1602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.908114101061711e-05,
+      "loss": 1.0346,
+      "step": 1603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9067828573689462e-05,
+      "loss": 0.6933,
+      "step": 1604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9054511077764016e-05,
+      "loss": 0.6355,
+      "step": 1605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9041188530270702e-05,
+      "loss": 0.6882,
+      "step": 1606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.902786093864223e-05,
+      "loss": 0.675,
+      "step": 1607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.901452831031417e-05,
+      "loss": 0.7363,
+      "step": 1608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9001190652724865e-05,
+      "loss": 0.8575,
+      "step": 1609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8987847973315476e-05,
+      "loss": 0.6394,
+      "step": 1610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.897450027952997e-05,
+      "loss": 0.8482,
+      "step": 1611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8961147578815114e-05,
+      "loss": 0.6495,
+      "step": 1612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.894778987862045e-05,
+      "loss": 0.5968,
+      "step": 1613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8934427186398327e-05,
+      "loss": 0.62,
+      "step": 1614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8921059509603876e-05,
+      "loss": 0.6722,
+      "step": 1615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8907686855695004e-05,
+      "loss": 0.7494,
+      "step": 1616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.88943092321324e-05,
+      "loss": 0.9896,
+      "step": 1617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8880926646379523e-05,
+      "loss": 0.7802,
+      "step": 1618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.88675391059026e-05,
+      "loss": 0.7525,
+      "step": 1619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8854146618170616e-05,
+      "loss": 0.6921,
+      "step": 1620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8840749190655336e-05,
+      "loss": 1.0283,
+      "step": 1621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.882734683083126e-05,
+      "loss": 0.5651,
+      "step": 1622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.881393954617565e-05,
+      "loss": 0.9311,
+      "step": 1623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.880052734416851e-05,
+      "loss": 0.7351,
+      "step": 1624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.87871102322926e-05,
+      "loss": 0.683,
+      "step": 1625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8773688218033395e-05,
+      "loss": 0.8169,
+      "step": 1626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8760261308879133e-05,
+      "loss": 0.78,
+      "step": 1627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8746829512320756e-05,
+      "loss": 0.7907,
+      "step": 1628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.873339283585196e-05,
+      "loss": 0.7281,
+      "step": 1629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8719951286969147e-05,
+      "loss": 0.6254,
+      "step": 1630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.870650487317144e-05,
+      "loss": 0.6826,
+      "step": 1631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.869305360196067e-05,
+      "loss": 0.7325,
+      "step": 1632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8679597480841398e-05,
+      "loss": 0.9267,
+      "step": 1633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8666136517320865e-05,
+      "loss": 0.7408,
+      "step": 1634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.865267071890903e-05,
+      "loss": 0.9326,
+      "step": 1635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.863920009311855e-05,
+      "loss": 0.686,
+      "step": 1636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8625724647464763e-05,
+      "loss": 0.9178,
+      "step": 1637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8612244389465706e-05,
+      "loss": 0.6672,
+      "step": 1638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8598759326642104e-05,
+      "loss": 0.7693,
+      "step": 1639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8585269466517345e-05,
+      "loss": 0.9183,
+      "step": 1640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.857177481661752e-05,
+      "loss": 0.7509,
+      "step": 1641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.855827538447137e-05,
+      "loss": 0.6984,
+      "step": 1642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8544771177610316e-05,
+      "loss": 0.9437,
+      "step": 1643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8531262203568433e-05,
+      "loss": 0.7693,
+      "step": 1644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8517748469882465e-05,
+      "loss": 0.7343,
+      "step": 1645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.850422998409181e-05,
+      "loss": 0.791,
+      "step": 1646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.849070675373851e-05,
+      "loss": 0.7427,
+      "step": 1647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8477178786367264e-05,
+      "loss": 1.066,
+      "step": 1648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8463646089525405e-05,
+      "loss": 0.6385,
+      "step": 1649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8450108670762907e-05,
+      "loss": 0.8161,
+      "step": 1650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8436566537632386e-05,
+      "loss": 0.8127,
+      "step": 1651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.842301969768908e-05,
+      "loss": 0.6284,
+      "step": 1652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.840946815849085e-05,
+      "loss": 0.6637,
+      "step": 1653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8395911927598185e-05,
+      "loss": 0.8784,
+      "step": 1654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.83823510125742e-05,
+      "loss": 0.6489,
+      "step": 1655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.83687854209846e-05,
+      "loss": 0.7367,
+      "step": 1656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.835521516039773e-05,
+      "loss": 0.8105,
+      "step": 1657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.83416402383845e-05,
+      "loss": 1.1166,
+      "step": 1658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8328060662518456e-05,
+      "loss": 0.6037,
+      "step": 1659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8314476440375735e-05,
+      "loss": 0.8629,
+      "step": 1660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.830088757953504e-05,
+      "loss": 1.0056,
+      "step": 1661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8287294087577697e-05,
+      "loss": 0.8877,
+      "step": 1662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.827369597208759e-05,
+      "loss": 0.7187,
+      "step": 1663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.82600932406512e-05,
+      "loss": 0.814,
+      "step": 1664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8246485900857565e-05,
+      "loss": 0.9152,
+      "step": 1665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.823287396029831e-05,
+      "loss": 0.6842,
+      "step": 1666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8219257426567624e-05,
+      "loss": 0.7724,
+      "step": 1667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8205636307262254e-05,
+      "loss": 0.7338,
+      "step": 1668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8192010609981495e-05,
+      "loss": 0.9829,
+      "step": 1669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8178380342327227e-05,
+      "loss": 0.8025,
+      "step": 1670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8164745511903855e-05,
+      "loss": 0.5831,
+      "step": 1671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8151106126318327e-05,
+      "loss": 0.6002,
+      "step": 1672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8137462193180147e-05,
+      "loss": 0.7989,
+      "step": 1673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8123813720101352e-05,
+      "loss": 0.6015,
+      "step": 1674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8110160714696505e-05,
+      "loss": 0.6992,
+      "step": 1675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8096503184582702e-05,
+      "loss": 1.0015,
+      "step": 1676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8082841137379566e-05,
+      "loss": 1.0514,
+      "step": 1677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8069174580709245e-05,
+      "loss": 0.5945,
+      "step": 1678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.805550352219638e-05,
+      "loss": 0.9168,
+      "step": 1679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8041827969468155e-05,
+      "loss": 0.806,
+      "step": 1680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8028147930154234e-05,
+      "loss": 0.7373,
+      "step": 1681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8014463411886806e-05,
+      "loss": 0.5218,
+      "step": 1682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8000774422300538e-05,
+      "loss": 0.6979,
+      "step": 1683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.798708096903261e-05,
+      "loss": 0.6804,
+      "step": 1684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.797338305972268e-05,
+      "loss": 1.009,
+      "step": 1685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7959680702012902e-05,
+      "loss": 0.8986,
+      "step": 1686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7945973903547896e-05,
+      "loss": 0.5937,
+      "step": 1687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7932262671974784e-05,
+      "loss": 0.7595,
+      "step": 1688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.791854701494313e-05,
+      "loss": 0.6924,
+      "step": 1689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7904826940104997e-05,
+      "loss": 0.7534,
+      "step": 1690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.78911024551149e-05,
+      "loss": 0.7929,
+      "step": 1691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.787737356762981e-05,
+      "loss": 0.6512,
+      "step": 1692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.786364028530916e-05,
+      "loss": 0.6568,
+      "step": 1693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7849902615814834e-05,
+      "loss": 0.7475,
+      "step": 1694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.783616056681116e-05,
+      "loss": 0.5876,
+      "step": 1695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.782241414596491e-05,
+      "loss": 0.7586,
+      "step": 1696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.780866336094531e-05,
+      "loss": 0.8848,
+      "step": 1697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.779490821942399e-05,
+      "loss": 0.6859,
+      "step": 1698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7781148729075043e-05,
+      "loss": 0.6921,
+      "step": 1699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7767384897574963e-05,
+      "loss": 0.6053,
+      "step": 1700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7753616732602694e-05,
+      "loss": 0.7149,
+      "step": 1701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7739844241839564e-05,
+      "loss": 1.1688,
+      "step": 1702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7726067432969343e-05,
+      "loss": 0.7474,
+      "step": 1703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7712286313678185e-05,
+      "loss": 0.7393,
+      "step": 1704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7698500891654673e-05,
+      "loss": 0.7571,
+      "step": 1705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.768471117458977e-05,
+      "loss": 0.8,
+      "step": 1706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7670917170176857e-05,
+      "loss": 0.6238,
+      "step": 1707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7657118886111686e-05,
+      "loss": 0.6614,
+      "step": 1708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7643316330092403e-05,
+      "loss": 0.7397,
+      "step": 1709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7629509509819544e-05,
+      "loss": 0.7311,
+      "step": 1710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.761569843299602e-05,
+      "loss": 0.7226,
+      "step": 1711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7601883107327115e-05,
+      "loss": 0.7359,
+      "step": 1712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7588063540520492e-05,
+      "loss": 1.083,
+      "step": 1713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7574239740286156e-05,
+      "loss": 0.6652,
+      "step": 1714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.756041171433651e-05,
+      "loss": 0.663,
+      "step": 1715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.754657947038628e-05,
+      "loss": 0.6852,
+      "step": 1716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7532743016152575e-05,
+      "loss": 0.6914,
+      "step": 1717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7518902359354823e-05,
+      "loss": 0.6602,
+      "step": 1718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.750505750771483e-05,
+      "loss": 0.8484,
+      "step": 1719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7491208468956702e-05,
+      "loss": 0.6906,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7477355250806926e-05,
+      "loss": 0.6139,
+      "step": 1721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7463497860994284e-05,
+      "loss": 0.8755,
+      "step": 1722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.744963630724991e-05,
+      "loss": 0.8837,
+      "step": 1723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7435770597307238e-05,
+      "loss": 0.7529,
+      "step": 1724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7421900738902042e-05,
+      "loss": 0.6608,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7408026739772403e-05,
+      "loss": 0.6641,
+      "step": 1726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7394148607658713e-05,
+      "loss": 0.6127,
+      "step": 1727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7380266350303655e-05,
+      "loss": 0.9989,
+      "step": 1728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7366379975452245e-05,
+      "loss": 0.7486,
+      "step": 1729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7352489490851764e-05,
+      "loss": 0.9056,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7338594904251805e-05,
+      "loss": 0.613,
+      "step": 1731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7324696223404234e-05,
+      "loss": 0.7128,
+      "step": 1732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7310793456063227e-05,
+      "loss": 0.7357,
+      "step": 1733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7296886609985215e-05,
+      "loss": 0.6687,
+      "step": 1734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7282975692928917e-05,
+      "loss": 0.8435,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.726906071265532e-05,
+      "loss": 0.6258,
+      "step": 1736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7255141676927676e-05,
+      "loss": 0.7492,
+      "step": 1737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7241218593511492e-05,
+      "loss": 0.731,
+      "step": 1738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7227291470174563e-05,
+      "loss": 0.7251,
+      "step": 1739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7213360314686902e-05,
+      "loss": 0.6959,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7199425134820795e-05,
+      "loss": 0.8352,
+      "step": 1741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.718548593835076e-05,
+      "loss": 0.7599,
+      "step": 1742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7171542733053562e-05,
+      "loss": 0.7929,
+      "step": 1743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7157595526708206e-05,
+      "loss": 0.7519,
+      "step": 1744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7143644327095917e-05,
+      "loss": 0.7223,
+      "step": 1745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7129689142000163e-05,
+      "loss": 0.5734,
+      "step": 1746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7115729979206623e-05,
+      "loss": 0.5958,
+      "step": 1747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.710176684650321e-05,
+      "loss": 0.9751,
+      "step": 1748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7087799751680025e-05,
+      "loss": 0.7214,
+      "step": 1749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7073828702529416e-05,
+      "loss": 0.7014,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7059853706845904e-05,
+      "loss": 0.8312,
+      "step": 1751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7045874772426238e-05,
+      "loss": 0.6478,
+      "step": 1752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.703189190706934e-05,
+      "loss": 0.7994,
+      "step": 1753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7017905118576347e-05,
+      "loss": 1.0383,
+      "step": 1754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.700391441475056e-05,
+      "loss": 0.7043,
+      "step": 1755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.69899198033975e-05,
+      "loss": 0.7486,
+      "step": 1756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.697592129232483e-05,
+      "loss": 0.6584,
+      "step": 1757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6961918889342422e-05,
+      "loss": 0.7644,
+      "step": 1758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.694791260226229e-05,
+      "loss": 0.7212,
+      "step": 1759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.693390243889863e-05,
+      "loss": 0.7169,
+      "step": 1760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6919888407067805e-05,
+      "loss": 0.7351,
+      "step": 1761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6905870514588332e-05,
+      "loss": 0.831,
+      "step": 1762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6891848769280873e-05,
+      "loss": 1.0213,
+      "step": 1763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6877823178968255e-05,
+      "loss": 0.8348,
+      "step": 1764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.686379375147543e-05,
+      "loss": 0.695,
+      "step": 1765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.684976049462951e-05,
+      "loss": 0.5867,
+      "step": 1766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6835723416259733e-05,
+      "loss": 0.6287,
+      "step": 1767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6821682524197475e-05,
+      "loss": 0.9808,
+      "step": 1768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6807637826276237e-05,
+      "loss": 0.7412,
+      "step": 1769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.679358933033164e-05,
+      "loss": 0.7328,
+      "step": 1770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.677953704420143e-05,
+      "loss": 0.7897,
+      "step": 1771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6765480975725463e-05,
+      "loss": 0.7583,
+      "step": 1772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6751421132745706e-05,
+      "loss": 0.7024,
+      "step": 1773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6737357523106235e-05,
+      "loss": 0.5827,
+      "step": 1774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6723290154653222e-05,
+      "loss": 0.664,
+      "step": 1775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6709219035234945e-05,
+      "loss": 0.6783,
+      "step": 1776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6695144172701763e-05,
+      "loss": 0.6973,
+      "step": 1777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.668106557490613e-05,
+      "loss": 0.7457,
+      "step": 1778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6666983249702583e-05,
+      "loss": 0.9843,
+      "step": 1779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6652897204947738e-05,
+      "loss": 0.6485,
+      "step": 1780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6638807448500292e-05,
+      "loss": 0.6778,
+      "step": 1781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6624713988221e-05,
+      "loss": 0.7837,
+      "step": 1782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6610616831972697e-05,
+      "loss": 0.7377,
+      "step": 1783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.659651598762027e-05,
+      "loss": 0.6369,
+      "step": 1784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6582411463030667e-05,
+      "loss": 0.7657,
+      "step": 1785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6568303266072892e-05,
+      "loss": 1.0286,
+      "step": 1786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6554191404618e-05,
+      "loss": 0.8959,
+      "step": 1787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.654007588653907e-05,
+      "loss": 0.7416,
+      "step": 1788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.652595671971126e-05,
+      "loss": 0.6462,
+      "step": 1789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.651183391201172e-05,
+      "loss": 0.9152,
+      "step": 1790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.649770747131967e-05,
+      "loss": 1.0635,
+      "step": 1791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.648357740551632e-05,
+      "loss": 1.0206,
+      "step": 1792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.646944372248494e-05,
+      "loss": 0.9966,
+      "step": 1793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6455306430110785e-05,
+      "loss": 0.6395,
+      "step": 1794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.644116553628114e-05,
+      "loss": 1.0537,
+      "step": 1795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6427021048885304e-05,
+      "loss": 0.8148,
+      "step": 1796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6412872975814575e-05,
+      "loss": 0.7373,
+      "step": 1797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.639872132496224e-05,
+      "loss": 0.9947,
+      "step": 1798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.638456610422361e-05,
+      "loss": 0.7592,
+      "step": 1799
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6370407321495953e-05,
+      "loss": 0.923,
+      "step": 1800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6356244984678546e-05,
+      "loss": 0.7826,
+      "step": 1801
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6342079101672653e-05,
+      "loss": 1.1023,
+      "step": 1802
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.63279096803815e-05,
+      "loss": 0.6461,
+      "step": 1803
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6313736728710292e-05,
+      "loss": 0.7278,
+      "step": 1804
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.629956025456622e-05,
+      "loss": 0.5801,
+      "step": 1805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.628538026585841e-05,
+      "loss": 1.013,
+      "step": 1806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6271196770497983e-05,
+      "loss": 0.646,
+      "step": 1807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6257009776397983e-05,
+      "loss": 0.9355,
+      "step": 1808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6242819291473437e-05,
+      "loss": 0.8936,
+      "step": 1809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6228625323641294e-05,
+      "loss": 0.8048,
+      "step": 1810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6214427880820456e-05,
+      "loss": 1.0403,
+      "step": 1811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6200226970931768e-05,
+      "loss": 0.7505,
+      "step": 1812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.618602260189801e-05,
+      "loss": 0.7091,
+      "step": 1813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6171814781643878e-05,
+      "loss": 0.7605,
+      "step": 1814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.615760351809601e-05,
+      "loss": 0.6671,
+      "step": 1815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6143388819182956e-05,
+      "loss": 0.6977,
+      "step": 1816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6129170692835185e-05,
+      "loss": 0.7968,
+      "step": 1817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.611494914698508e-05,
+      "loss": 0.8169,
+      "step": 1818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6100724189566924e-05,
+      "loss": 1.0422,
+      "step": 1819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6086495828516913e-05,
+      "loss": 0.7618,
+      "step": 1820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6072264071773136e-05,
+      "loss": 0.725,
+      "step": 1821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6058028927275584e-05,
+      "loss": 0.993,
+      "step": 1822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.604379040296612e-05,
+      "loss": 0.7008,
+      "step": 1823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6029548506788523e-05,
+      "loss": 0.674,
+      "step": 1824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6015303246688418e-05,
+      "loss": 0.7161,
+      "step": 1825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6001054630613337e-05,
+      "loss": 0.7408,
+      "step": 1826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5986802666512657e-05,
+      "loss": 0.6363,
+      "step": 1827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5972547362337646e-05,
+      "loss": 0.8871,
+      "step": 1828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.595828872604143e-05,
+      "loss": 0.9922,
+      "step": 1829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5944026765578988e-05,
+      "loss": 0.837,
+      "step": 1830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5929761488907153e-05,
+      "loss": 0.8167,
+      "step": 1831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.591549290398461e-05,
+      "loss": 0.9511,
+      "step": 1832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5901221018771896e-05,
+      "loss": 0.6309,
+      "step": 1833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.588694584123139e-05,
+      "loss": 1.126,
+      "step": 1834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5872667379327287e-05,
+      "loss": 0.904,
+      "step": 1835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.585838564102564e-05,
+      "loss": 0.8989,
+      "step": 1836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.584410063429431e-05,
+      "loss": 0.6698,
+      "step": 1837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5829812367103e-05,
+      "loss": 0.7799,
+      "step": 1838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.581552084742322e-05,
+      "loss": 0.7412,
+      "step": 1839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.58012260832283e-05,
+      "loss": 1.0364,
+      "step": 1840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5786928082493367e-05,
+      "loss": 0.7041,
+      "step": 1841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5772626853195385e-05,
+      "loss": 1.0555,
+      "step": 1842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.575832240331307e-05,
+      "loss": 0.7258,
+      "step": 1843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5744014740826994e-05,
+      "loss": 0.7898,
+      "step": 1844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.572970387371947e-05,
+      "loss": 0.6427,
+      "step": 1845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5715389809974623e-05,
+      "loss": 0.6486,
+      "step": 1846
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.570107255757836e-05,
+      "loss": 0.6893,
+      "step": 1847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.568675212451837e-05,
+      "loss": 0.6735,
+      "step": 1848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.56724285187841e-05,
+      "loss": 0.6404,
+      "step": 1849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.565810174836679e-05,
+      "loss": 0.7147,
+      "step": 1850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5643771821259426e-05,
+      "loss": 1.0749,
+      "step": 1851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5629438745456766e-05,
+      "loss": 0.7436,
+      "step": 1852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5615102528955325e-05,
+      "loss": 0.6422,
+      "step": 1853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.560076317975336e-05,
+      "loss": 0.8822,
+      "step": 1854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.558642070585089e-05,
+      "loss": 0.6807,
+      "step": 1855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5572075115249665e-05,
+      "loss": 0.7904,
+      "step": 1856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.555772641595318e-05,
+      "loss": 0.8258,
+      "step": 1857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5543374615966668e-05,
+      "loss": 0.6749,
+      "step": 1858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5529019723297078e-05,
+      "loss": 0.5542,
+      "step": 1859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5514661745953105e-05,
+      "loss": 0.6935,
+      "step": 1860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5500300691945142e-05,
+      "loss": 1.0441,
+      "step": 1861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5485936569285326e-05,
+      "loss": 0.7458,
+      "step": 1862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5471569385987466e-05,
+      "loss": 0.7336,
+      "step": 1863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5457199150067124e-05,
+      "loss": 1.1127,
+      "step": 1864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5442825869541535e-05,
+      "loss": 0.6198,
+      "step": 1865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5428449552429646e-05,
+      "loss": 0.8219,
+      "step": 1866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5414070206752085e-05,
+      "loss": 0.6159,
+      "step": 1867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5399687840531188e-05,
+      "loss": 0.6464,
+      "step": 1868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5385302461790956e-05,
+      "loss": 0.6277,
+      "step": 1869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5370914078557088e-05,
+      "loss": 1.0132,
+      "step": 1870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5356522698856954e-05,
+      "loss": 0.7824,
+      "step": 1871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.534212833071959e-05,
+      "loss": 0.6294,
+      "step": 1872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.53277309821757e-05,
+      "loss": 0.7923,
+      "step": 1873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.531333066125766e-05,
+      "loss": 0.7346,
+      "step": 1874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.52989273759995e-05,
+      "loss": 0.8912,
+      "step": 1875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.528452113443689e-05,
+      "loss": 1.1671,
+      "step": 1876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.527011194460718e-05,
+      "loss": 0.8345,
+      "step": 1877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5255699814549325e-05,
+      "loss": 0.7015,
+      "step": 1878
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5241284752303962e-05,
+      "loss": 0.7594,
+      "step": 1879
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5226866765913335e-05,
+      "loss": 0.8321,
+      "step": 1880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5212445863421326e-05,
+      "loss": 0.6095,
+      "step": 1881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5198022052873445e-05,
+      "loss": 0.655,
+      "step": 1882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5183595342316833e-05,
+      "loss": 0.7122,
+      "step": 1883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.516916573980023e-05,
+      "loss": 0.7439,
+      "step": 1884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5154733253374008e-05,
+      "loss": 0.6922,
+      "step": 1885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5140297891090142e-05,
+      "loss": 1.1394,
+      "step": 1886
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.512585966100221e-05,
+      "loss": 0.6404,
+      "step": 1887
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5111418571165383e-05,
+      "loss": 0.6092,
+      "step": 1888
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5096974629636442e-05,
+      "loss": 0.6435,
+      "step": 1889
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5082527844473745e-05,
+      "loss": 0.6948,
+      "step": 1890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5068078223737252e-05,
+      "loss": 0.6693,
+      "step": 1891
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5053625775488487e-05,
+      "loss": 0.5437,
+      "step": 1892
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.503917050779057e-05,
+      "loss": 0.7068,
+      "step": 1893
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5024712428708173e-05,
+      "loss": 0.752,
+      "step": 1894
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5010251546307556e-05,
+      "loss": 0.8768,
+      "step": 1895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4995787868656533e-05,
+      "loss": 0.6546,
+      "step": 1896
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.498132140382448e-05,
+      "loss": 0.6908,
+      "step": 1897
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4966852159882326e-05,
+      "loss": 1.0705,
+      "step": 1898
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4952380144902554e-05,
+      "loss": 0.8287,
+      "step": 1899
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4937905366959187e-05,
+      "loss": 0.7141,
+      "step": 1900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.49234278341278e-05,
+      "loss": 0.7482,
+      "step": 1901
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.490894755448549e-05,
+      "loss": 1.1042,
+      "step": 1902
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.48944645361109e-05,
+      "loss": 0.6807,
+      "step": 1903
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.48799787870842e-05,
+      "loss": 0.6803,
+      "step": 1904
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.486549031548707e-05,
+      "loss": 1.1119,
+      "step": 1905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4850999129402725e-05,
+      "loss": 0.5245,
+      "step": 1906
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.483650523691588e-05,
+      "loss": 0.6481,
+      "step": 1907
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4822008646112777e-05,
+      "loss": 0.8023,
+      "step": 1908
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4807509365081145e-05,
+      "loss": 0.7715,
+      "step": 1909
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.479300740191023e-05,
+      "loss": 0.6095,
+      "step": 1910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.477850276469076e-05,
+      "loss": 0.9835,
+      "step": 1911
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.476399546151497e-05,
+      "loss": 0.6546,
+      "step": 1912
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.474948550047657e-05,
+      "loss": 0.7071,
+      "step": 1913
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4734972889670756e-05,
+      "loss": 1.084,
+      "step": 1914
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.472045763719421e-05,
+      "loss": 0.5729,
+      "step": 1915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4705939751145077e-05,
+      "loss": 0.9302,
+      "step": 1916
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4691419239622965e-05,
+      "loss": 0.6746,
+      "step": 1917
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4676896110728978e-05,
+      "loss": 0.6823,
+      "step": 1918
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4662370372565647e-05,
+      "loss": 0.9494,
+      "step": 1919
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4647842033236976e-05,
+      "loss": 0.5577,
+      "step": 1920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4633311100848407e-05,
+      "loss": 0.9292,
+      "step": 1921
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.461877758350685e-05,
+      "loss": 0.8701,
+      "step": 1922
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4604241489320637e-05,
+      "loss": 0.651,
+      "step": 1923
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4589702826399545e-05,
+      "loss": 0.7567,
+      "step": 1924
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4575161602854783e-05,
+      "loss": 0.8153,
+      "step": 1925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4560617826798997e-05,
+      "loss": 1.1796,
+      "step": 1926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4546071506346246e-05,
+      "loss": 0.6022,
+      "step": 1927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4531522649612014e-05,
+      "loss": 0.6215,
+      "step": 1928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4516971264713188e-05,
+      "loss": 0.6425,
+      "step": 1929
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4502417359768087e-05,
+      "loss": 0.7604,
+      "step": 1930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4487860942896423e-05,
+      "loss": 0.6832,
+      "step": 1931
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.447330202221931e-05,
+      "loss": 0.5967,
+      "step": 1932
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.445874060585925e-05,
+      "loss": 0.8789,
+      "step": 1933
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4444176701940174e-05,
+      "loss": 0.7744,
+      "step": 1934
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4429610318587345e-05,
+      "loss": 0.8603,
+      "step": 1935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4415041463927458e-05,
+      "loss": 1.0406,
+      "step": 1936
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4400470146088554e-05,
+      "loss": 0.8753,
+      "step": 1937
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.438589637320008e-05,
+      "loss": 1.0448,
+      "step": 1938
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4371320153392818e-05,
+      "loss": 1.1408,
+      "step": 1939
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4356741494798927e-05,
+      "loss": 0.7318,
+      "step": 1940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.434216040555195e-05,
+      "loss": 0.5748,
+      "step": 1941
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.432757689378676e-05,
+      "loss": 0.6759,
+      "step": 1942
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4312990967639575e-05,
+      "loss": 0.6698,
+      "step": 1943
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.429840263524799e-05,
+      "loss": 0.5729,
+      "step": 1944
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4283811904750917e-05,
+      "loss": 0.949,
+      "step": 1945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4269218784288623e-05,
+      "loss": 0.7886,
+      "step": 1946
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4254623282002695e-05,
+      "loss": 0.7453,
+      "step": 1947
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4240025406036055e-05,
+      "loss": 0.8024,
+      "step": 1948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4225425164532947e-05,
+      "loss": 0.6798,
+      "step": 1949
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.421082256563894e-05,
+      "loss": 0.6681,
+      "step": 1950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4196217617500914e-05,
+      "loss": 0.6863,
+      "step": 1951
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4181610328267056e-05,
+      "loss": 0.9167,
+      "step": 1952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4167000706086866e-05,
+      "loss": 0.7319,
+      "step": 1953
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.415238875911114e-05,
+      "loss": 0.701,
+      "step": 1954
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.413777449549198e-05,
+      "loss": 0.7006,
+      "step": 1955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4123157923382768e-05,
+      "loss": 0.7387,
+      "step": 1956
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4108539050938188e-05,
+      "loss": 0.6856,
+      "step": 1957
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4093917886314188e-05,
+      "loss": 0.5888,
+      "step": 1958
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4079294437668018e-05,
+      "loss": 0.9136,
+      "step": 1959
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4064668713158183e-05,
+      "loss": 1.1708,
+      "step": 1960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4050040720944474e-05,
+      "loss": 0.6695,
+      "step": 1961
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.403541046918793e-05,
+      "loss": 0.6932,
+      "step": 1962
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.402077796605086e-05,
+      "loss": 0.6529,
+      "step": 1963
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4006143219696828e-05,
+      "loss": 0.65,
+      "step": 1964
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.399150623829065e-05,
+      "loss": 0.7067,
+      "step": 1965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.397686702999839e-05,
+      "loss": 0.7212,
+      "step": 1966
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3962225602987364e-05,
+      "loss": 0.7201,
+      "step": 1967
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3947581965426095e-05,
+      "loss": 0.7767,
+      "step": 1968
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3932936125484375e-05,
+      "loss": 0.6585,
+      "step": 1969
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3918288091333196e-05,
+      "loss": 0.7889,
+      "step": 1970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3903637871144798e-05,
+      "loss": 0.7527,
+      "step": 1971
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3888985473092624e-05,
+      "loss": 0.7596,
+      "step": 1972
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3874330905351343e-05,
+      "loss": 0.6523,
+      "step": 1973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3859674176096822e-05,
+      "loss": 0.7689,
+      "step": 1974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3845015293506143e-05,
+      "loss": 0.7438,
+      "step": 1975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3830354265757586e-05,
+      "loss": 0.6701,
+      "step": 1976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.381569110103064e-05,
+      "loss": 0.5839,
+      "step": 1977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.380102580750596e-05,
+      "loss": 0.7905,
+      "step": 1978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.378635839336542e-05,
+      "loss": 0.9291,
+      "step": 1979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3771688866792043e-05,
+      "loss": 0.8529,
+      "step": 1980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3757017235970067e-05,
+      "loss": 0.7753,
+      "step": 1981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3742343509084876e-05,
+      "loss": 0.5922,
+      "step": 1982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3727667694323037e-05,
+      "loss": 0.7497,
+      "step": 1983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3712989799872272e-05,
+      "loss": 0.7235,
+      "step": 1984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3698309833921478e-05,
+      "loss": 0.6308,
+      "step": 1985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3683627804660692e-05,
+      "loss": 0.6889,
+      "step": 1986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.366894372028111e-05,
+      "loss": 0.7333,
+      "step": 1987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3654257588975077e-05,
+      "loss": 0.9017,
+      "step": 1988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3639569418936077e-05,
+      "loss": 1.0138,
+      "step": 1989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3624879218358728e-05,
+      "loss": 0.6943,
+      "step": 1990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3610186995438786e-05,
+      "loss": 0.6931,
+      "step": 1991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3595492758373134e-05,
+      "loss": 0.6536,
+      "step": 1992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.358079651535977e-05,
+      "loss": 0.8966,
+      "step": 1993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.356609827459783e-05,
+      "loss": 0.7332,
+      "step": 1994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3551398044287552e-05,
+      "loss": 0.7282,
+      "step": 1995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3536695832630284e-05,
+      "loss": 0.6522,
+      "step": 1996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3521991647828475e-05,
+      "loss": 0.7365,
+      "step": 1997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3507285498085687e-05,
+      "loss": 0.648,
+      "step": 1998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.349257739160657e-05,
+      "loss": 0.7689,
+      "step": 1999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3477867336596877e-05,
+      "loss": 0.6281,
+      "step": 2000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3463155341263418e-05,
+      "loss": 0.6477,
+      "step": 2001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3448441413814127e-05,
+      "loss": 0.759,
+      "step": 2002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3433725562457986e-05,
+      "loss": 0.9804,
+      "step": 2003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.341900779540506e-05,
+      "loss": 0.661,
+      "step": 2004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3404288120866485e-05,
+      "loss": 0.6177,
+      "step": 2005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.338956654705446e-05,
+      "loss": 0.8184,
+      "step": 2006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3374843082182236e-05,
+      "loss": 0.9817,
+      "step": 2007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3360117734464135e-05,
+      "loss": 0.6737,
+      "step": 2008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3345390512115508e-05,
+      "loss": 0.8552,
+      "step": 2009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3330661423352773e-05,
+      "loss": 0.6839,
+      "step": 2010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.331593047639337e-05,
+      "loss": 0.608,
+      "step": 2011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.33011976794558e-05,
+      "loss": 0.8236,
+      "step": 2012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3286463040759564e-05,
+      "loss": 0.8559,
+      "step": 2013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3271726568525232e-05,
+      "loss": 0.5812,
+      "step": 2014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.325698827097435e-05,
+      "loss": 0.6587,
+      "step": 2015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3242248156329523e-05,
+      "loss": 1.0617,
+      "step": 2016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3227506232814335e-05,
+      "loss": 0.789,
+      "step": 2017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3212762508653414e-05,
+      "loss": 0.5445,
+      "step": 2018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3198016992072358e-05,
+      "loss": 0.7065,
+      "step": 2019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3183269691297806e-05,
+      "loss": 0.7375,
+      "step": 2020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.316852061455734e-05,
+      "loss": 0.7668,
+      "step": 2021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.315376977007958e-05,
+      "loss": 0.6465,
+      "step": 2022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3139017166094108e-05,
+      "loss": 0.6766,
+      "step": 2023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.31242628108315e-05,
+      "loss": 0.812,
+      "step": 2024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3109506712523288e-05,
+      "loss": 0.9338,
+      "step": 2025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3094748879402005e-05,
+      "loss": 0.6775,
+      "step": 2026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3079989319701123e-05,
+      "loss": 0.6554,
+      "step": 2027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.306522804165511e-05,
+      "loss": 0.706,
+      "step": 2028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3050465053499356e-05,
+      "loss": 1.0419,
+      "step": 2029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3035700363470244e-05,
+      "loss": 1.1246,
+      "step": 2030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3020933979805072e-05,
+      "loss": 0.9424,
+      "step": 2031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3006165910742094e-05,
+      "loss": 0.9498,
+      "step": 2032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.299139616452052e-05,
+      "loss": 0.7175,
+      "step": 2033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2976624749380472e-05,
+      "loss": 0.783,
+      "step": 2034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2961851673563023e-05,
+      "loss": 0.7589,
+      "step": 2035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.294707694531015e-05,
+      "loss": 1.0769,
+      "step": 2036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.293230057286478e-05,
+      "loss": 0.6499,
+      "step": 2037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2917522564470737e-05,
+      "loss": 0.524,
+      "step": 2038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2902742928372764e-05,
+      "loss": 0.9697,
+      "step": 2039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2887961672816513e-05,
+      "loss": 0.5965,
+      "step": 2040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2873178806048536e-05,
+      "loss": 0.6263,
+      "step": 2041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2858394336316283e-05,
+      "loss": 1.0699,
+      "step": 2042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2843608271868117e-05,
+      "loss": 0.74,
+      "step": 2043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.282882062095325e-05,
+      "loss": 1.0046,
+      "step": 2044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2814031391821826e-05,
+      "loss": 0.6145,
+      "step": 2045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.279924059272483e-05,
+      "loss": 0.697,
+      "step": 2046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.278444823191415e-05,
+      "loss": 0.7344,
+      "step": 2047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.276965431764254e-05,
+      "loss": 0.7618,
+      "step": 2048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.275485885816361e-05,
+      "loss": 0.6393,
+      "step": 2049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2740061861731838e-05,
+      "loss": 0.8563,
+      "step": 2050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2725263336602565e-05,
+      "loss": 0.738,
+      "step": 2051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2710463291031973e-05,
+      "loss": 0.7339,
+      "step": 2052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2695661733277114e-05,
+      "loss": 0.7115,
+      "step": 2053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2680858671595853e-05,
+      "loss": 0.6573,
+      "step": 2054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.266605411424691e-05,
+      "loss": 1.0228,
+      "step": 2055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.265124806948985e-05,
+      "loss": 0.9001,
+      "step": 2056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2636440545585053e-05,
+      "loss": 1.0218,
+      "step": 2057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.262163155079372e-05,
+      "loss": 0.739,
+      "step": 2058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2606821093377894e-05,
+      "loss": 0.7466,
+      "step": 2059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.259200918160041e-05,
+      "loss": 0.9179,
+      "step": 2060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.257719582372494e-05,
+      "loss": 0.6438,
+      "step": 2061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2562381028015924e-05,
+      "loss": 0.6259,
+      "step": 2062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2547564802738656e-05,
+      "loss": 0.6544,
+      "step": 2063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.253274715615918e-05,
+      "loss": 0.8923,
+      "step": 2064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2517928096544365e-05,
+      "loss": 1.1472,
+      "step": 2065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2503107632161855e-05,
+      "loss": 0.5958,
+      "step": 2066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2488285771280065e-05,
+      "loss": 1.0203,
+      "step": 2067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2473462522168223e-05,
+      "loss": 0.5272,
+      "step": 2068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2458637893096293e-05,
+      "loss": 0.6129,
+      "step": 2069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2443811892335044e-05,
+      "loss": 0.9115,
+      "step": 2070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.242898452815598e-05,
+      "loss": 0.7687,
+      "step": 2071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2414155808831392e-05,
+      "loss": 0.6575,
+      "step": 2072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.23993257426343e-05,
+      "loss": 1.0064,
+      "step": 2073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2384494337838504e-05,
+      "loss": 0.8397,
+      "step": 2074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2369661602718532e-05,
+      "loss": 0.5507,
+      "step": 2075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2354827545549654e-05,
+      "loss": 0.775,
+      "step": 2076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.233999217460789e-05,
+      "loss": 0.5859,
+      "step": 2077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.232515549816998e-05,
+      "loss": 0.6273,
+      "step": 2078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2310317524513398e-05,
+      "loss": 0.6775,
+      "step": 2079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2295478261916347e-05,
+      "loss": 0.822,
+      "step": 2080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.228063771865773e-05,
+      "loss": 0.9221,
+      "step": 2081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.22657959030172e-05,
+      "loss": 0.738,
+      "step": 2082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2250952823275075e-05,
+      "loss": 0.7838,
+      "step": 2083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2236108487712414e-05,
+      "loss": 0.8125,
+      "step": 2084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.222126290461096e-05,
+      "loss": 0.6851,
+      "step": 2085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2206416082253166e-05,
+      "loss": 0.6681,
+      "step": 2086
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.219156802892215e-05,
+      "loss": 0.9314,
+      "step": 2087
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2176718752901745e-05,
+      "loss": 0.6485,
+      "step": 2088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.216186826247645e-05,
+      "loss": 0.7091,
+      "step": 2089
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2147016565931447e-05,
+      "loss": 0.7785,
+      "step": 2090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2132163671552586e-05,
+      "loss": 0.7019,
+      "step": 2091
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.211730958762639e-05,
+      "loss": 0.6197,
+      "step": 2092
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2102454322440044e-05,
+      "loss": 0.6201,
+      "step": 2093
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.20875978842814e-05,
+      "loss": 0.7278,
+      "step": 2094
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2072740281438944e-05,
+      "loss": 0.7687,
+      "step": 2095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.205788152220184e-05,
+      "loss": 1.039,
+      "step": 2096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2043021614859868e-05,
+      "loss": 0.9368,
+      "step": 2097
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.202816056770347e-05,
+      "loss": 0.6613,
+      "step": 2098
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2013298389023712e-05,
+      "loss": 0.714,
+      "step": 2099
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1998435087112302e-05,
+      "loss": 0.7814,
+      "step": 2100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1983570670261568e-05,
+      "loss": 0.6866,
+      "step": 2101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1968705146764452e-05,
+      "loss": 0.59,
+      "step": 2102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1953838524914527e-05,
+      "loss": 0.8425,
+      "step": 2103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1938970813005976e-05,
+      "loss": 0.7386,
+      "step": 2104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.192410201933358e-05,
+      "loss": 1.0482,
+      "step": 2105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1909232152192743e-05,
+      "loss": 1.0363,
+      "step": 2106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.189436121987944e-05,
+      "loss": 0.6131,
+      "step": 2107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1879489230690276e-05,
+      "loss": 0.7958,
+      "step": 2108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1864616192922407e-05,
+      "loss": 0.8903,
+      "step": 2109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1849742114873606e-05,
+      "loss": 0.7306,
+      "step": 2110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1834867004842202e-05,
+      "loss": 0.6529,
+      "step": 2111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.181999087112713e-05,
+      "loss": 0.7813,
+      "step": 2112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1805113722027863e-05,
+      "loss": 0.687,
+      "step": 2113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.179023556584445e-05,
+      "loss": 1.0282,
+      "step": 2114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1775356410877517e-05,
+      "loss": 0.7573,
+      "step": 2115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1760476265428234e-05,
+      "loss": 0.7288,
+      "step": 2116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1745595137798325e-05,
+      "loss": 1.0272,
+      "step": 2117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.173071303629006e-05,
+      "loss": 0.604,
+      "step": 2118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1715829969206265e-05,
+      "loss": 0.7084,
+      "step": 2119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1700945944850288e-05,
+      "loss": 0.6279,
+      "step": 2120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1686060971526028e-05,
+      "loss": 0.9133,
+      "step": 2121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1671175057537886e-05,
+      "loss": 1.1239,
+      "step": 2122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.165628821119082e-05,
+      "loss": 1.0976,
+      "step": 2123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1641400440790297e-05,
+      "loss": 1.0191,
+      "step": 2124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1626511754642287e-05,
+      "loss": 0.7737,
+      "step": 2125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.161162216105328e-05,
+      "loss": 0.6576,
+      "step": 2126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1596731668330277e-05,
+      "loss": 0.8396,
+      "step": 2127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1581840284780774e-05,
+      "loss": 0.6007,
+      "step": 2128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.156694801871277e-05,
+      "loss": 0.6033,
+      "step": 2129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1552054878434743e-05,
+      "loss": 0.6515,
+      "step": 2130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1537160872255686e-05,
+      "loss": 0.6997,
+      "step": 2131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1522266008485033e-05,
+      "loss": 1.0456,
+      "step": 2132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1507370295432747e-05,
+      "loss": 0.7889,
+      "step": 2133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.149247374140922e-05,
+      "loss": 0.6807,
+      "step": 2134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1477576354725337e-05,
+      "loss": 0.7283,
+      "step": 2135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1462678143692445e-05,
+      "loss": 1.0345,
+      "step": 2136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1447779116622353e-05,
+      "loss": 0.6955,
+      "step": 2137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1432879281827305e-05,
+      "loss": 0.7385,
+      "step": 2138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1417978647620023e-05,
+      "loss": 0.6491,
+      "step": 2139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.140307722231366e-05,
+      "loss": 0.703,
+      "step": 2140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1388175014221817e-05,
+      "loss": 0.7023,
+      "step": 2141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1373272031658518e-05,
+      "loss": 0.7136,
+      "step": 2142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1358368282938245e-05,
+      "loss": 0.7113,
+      "step": 2143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.134346377637588e-05,
+      "loss": 0.6808,
+      "step": 2144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1328558520286745e-05,
+      "loss": 0.7406,
+      "step": 2145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.131365252298657e-05,
+      "loss": 0.7657,
+      "step": 2146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.129874579279151e-05,
+      "loss": 0.7642,
+      "step": 2147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.128383833801812e-05,
+      "loss": 0.6976,
+      "step": 2148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1268930166983358e-05,
+      "loss": 0.7426,
+      "step": 2149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1254021288004593e-05,
+      "loss": 0.6466,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.123911170939956e-05,
+      "loss": 0.6893,
+      "step": 2151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1224201439486435e-05,
+      "loss": 0.966,
+      "step": 2152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1209290486583728e-05,
+      "loss": 0.7811,
+      "step": 2153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1194378859010367e-05,
+      "loss": 0.6281,
+      "step": 2154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.117946656508563e-05,
+      "loss": 1.0585,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1164553613129182e-05,
+      "loss": 0.6062,
+      "step": 2156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.114964001146105e-05,
+      "loss": 0.7023,
+      "step": 2157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1134725768401633e-05,
+      "loss": 0.5973,
+      "step": 2158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1119810892271667e-05,
+      "loss": 0.7627,
+      "step": 2159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1104895391392266e-05,
+      "loss": 0.6049,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.108997927408487e-05,
+      "loss": 0.8133,
+      "step": 2161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1075062548671284e-05,
+      "loss": 0.63,
+      "step": 2162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1060145223473626e-05,
+      "loss": 0.643,
+      "step": 2163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.104522730681438e-05,
+      "loss": 0.7685,
+      "step": 2164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1030308807016335e-05,
+      "loss": 0.6326,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1015389732402625e-05,
+      "loss": 0.7138,
+      "step": 2166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.100047009129668e-05,
+      "loss": 0.6656,
+      "step": 2167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.098554989202227e-05,
+      "loss": 0.6555,
+      "step": 2168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0970629142903462e-05,
+      "loss": 0.6271,
+      "step": 2169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0955707852264635e-05,
+      "loss": 0.7535,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0940786028430477e-05,
+      "loss": 0.6662,
+      "step": 2171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0925863679725954e-05,
+      "loss": 0.7289,
+      "step": 2172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0910940814476348e-05,
+      "loss": 0.6879,
+      "step": 2173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0896017441007207e-05,
+      "loss": 0.8308,
+      "step": 2174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0881093567644384e-05,
+      "loss": 0.8327,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0866169202713994e-05,
+      "loss": 0.6557,
+      "step": 2176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.085124435454243e-05,
+      "loss": 0.7245,
+      "step": 2177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0836319031456372e-05,
+      "loss": 0.6172,
+      "step": 2178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0821393241782734e-05,
+      "loss": 0.6915,
+      "step": 2179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0806466993848712e-05,
+      "loss": 1.1474,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0791540295981757e-05,
+      "loss": 0.6128,
+      "step": 2181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0776613156509567e-05,
+      "loss": 0.7131,
+      "step": 2182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0761685583760075e-05,
+      "loss": 0.8482,
+      "step": 2183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0746757586061474e-05,
+      "loss": 0.5895,
+      "step": 2184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.073182917174218e-05,
+      "loss": 0.5846,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0716900349130858e-05,
+      "loss": 0.6866,
+      "step": 2186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0701971126556382e-05,
+      "loss": 0.7894,
+      "step": 2187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.068704151234787e-05,
+      "loss": 0.6469,
+      "step": 2188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0672111514834626e-05,
+      "loss": 0.6014,
+      "step": 2189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.065718114234621e-05,
+      "loss": 1.0763,
+      "step": 2190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0642250403212352e-05,
+      "loss": 0.6873,
+      "step": 2191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0627319305763017e-05,
+      "loss": 1.1275,
+      "step": 2192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.061238785832835e-05,
+      "loss": 0.7339,
+      "step": 2193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0597456069238703e-05,
+      "loss": 0.6686,
+      "step": 2194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0582523946824607e-05,
+      "loss": 0.7488,
+      "step": 2195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0567591499416787e-05,
+      "loss": 0.5955,
+      "step": 2196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0552658735346153e-05,
+      "loss": 0.7848,
+      "step": 2197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0537725662943777e-05,
+      "loss": 0.7212,
+      "step": 2198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0522792290540928e-05,
+      "loss": 0.6811,
+      "step": 2199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.050785862646901e-05,
+      "loss": 0.6387,
+      "step": 2200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0492924679059626e-05,
+      "loss": 0.6594,
+      "step": 2201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0477990456644497e-05,
+      "loss": 0.6829,
+      "step": 2202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0463055967555528e-05,
+      "loss": 0.883,
+      "step": 2203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0448121220124767e-05,
+      "loss": 0.7103,
+      "step": 2204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0433186222684407e-05,
+      "loss": 0.8278,
+      "step": 2205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0418250983566758e-05,
+      "loss": 0.683,
+      "step": 2206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.040331551110429e-05,
+      "loss": 0.7287,
+      "step": 2207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.03883798136296e-05,
+      "loss": 0.657,
+      "step": 2208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.03734438994754e-05,
+      "loss": 0.9886,
+      "step": 2209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0358507776974538e-05,
+      "loss": 0.8537,
+      "step": 2210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0343571454459963e-05,
+      "loss": 0.5787,
+      "step": 2211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.032863494026474e-05,
+      "loss": 0.5884,
+      "step": 2212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.031369824272205e-05,
+      "loss": 0.5378,
+      "step": 2213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0298761370165164e-05,
+      "loss": 0.6203,
+      "step": 2214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0283824330927454e-05,
+      "loss": 0.7115,
+      "step": 2215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.026888713334239e-05,
+      "loss": 0.6455,
+      "step": 2216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0253949785743537e-05,
+      "loss": 1.0783,
+      "step": 2217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0239012296464513e-05,
+      "loss": 0.6383,
+      "step": 2218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0224074673839054e-05,
+      "loss": 0.7532,
+      "step": 2219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0209136926200937e-05,
+      "loss": 0.7648,
+      "step": 2220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0194199061884035e-05,
+      "loss": 0.6499,
+      "step": 2221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.017926108922227e-05,
+      "loss": 1.0159,
+      "step": 2222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0164323016549637e-05,
+      "loss": 0.6863,
+      "step": 2223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.014938485220017e-05,
+      "loss": 0.6427,
+      "step": 2224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.013444660450797e-05,
+      "loss": 1.0156,
+      "step": 2225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0119508281807175e-05,
+      "loss": 1.0019,
+      "step": 2226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0104569892431968e-05,
+      "loss": 0.7512,
+      "step": 2227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.008963144471657e-05,
+      "loss": 0.9816,
+      "step": 2228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0074692946995236e-05,
+      "loss": 0.7106,
+      "step": 2229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0059754407602244e-05,
+      "loss": 0.6017,
+      "step": 2230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0044815834871888e-05,
+      "loss": 0.8055,
+      "step": 2231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0029877237138508e-05,
+      "loss": 0.8041,
+      "step": 2232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.001493862273643e-05,
+      "loss": 0.8707,
+      "step": 2233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.5379,
+      "step": 2234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9985061377263578e-05,
+      "loss": 0.6667,
+      "step": 2235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.99701227628615e-05,
+      "loss": 0.9033,
+      "step": 2236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9955184165128115e-05,
+      "loss": 0.9132,
+      "step": 2237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9940245592397766e-05,
+      "loss": 0.6525,
+      "step": 2238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9925307053004774e-05,
+      "loss": 0.7201,
+      "step": 2239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9910368555283434e-05,
+      "loss": 0.7643,
+      "step": 2240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9895430107568036e-05,
+      "loss": 0.8716,
+      "step": 2241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9880491718192828e-05,
+      "loss": 0.6675,
+      "step": 2242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9865553395492036e-05,
+      "loss": 0.7982,
+      "step": 2243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.985061514779983e-05,
+      "loss": 0.5922,
+      "step": 2244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9835676983450366e-05,
+      "loss": 0.6677,
+      "step": 2245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982073891077774e-05,
+      "loss": 0.7208,
+      "step": 2246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9805800938115968e-05,
+      "loss": 0.6174,
+      "step": 2247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9790863073799073e-05,
+      "loss": 0.6632,
+      "step": 2248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9775925326160953e-05,
+      "loss": 0.6317,
+      "step": 2249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9760987703535497e-05,
+      "loss": 0.7089,
+      "step": 2250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.974605021425647e-05,
+      "loss": 0.648,
+      "step": 2251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9731112866657614e-05,
+      "loss": 0.8277,
+      "step": 2252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716175669072546e-05,
+      "loss": 0.7856,
+      "step": 2253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9701238629834846e-05,
+      "loss": 0.8033,
+      "step": 2254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.968630175727795e-05,
+      "loss": 0.6381,
+      "step": 2255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9671365059735264e-05,
+      "loss": 0.7218,
+      "step": 2256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9656428545540037e-05,
+      "loss": 0.88,
+      "step": 2257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.964149222302547e-05,
+      "loss": 1.0765,
+      "step": 2258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.962655610052461e-05,
+      "loss": 0.8268,
+      "step": 2259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9611620186370408e-05,
+      "loss": 0.7031,
+      "step": 2260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.959668448889572e-05,
+      "loss": 0.6277,
+      "step": 2261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9581749016433252e-05,
+      "loss": 0.9077,
+      "step": 2262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9566813777315606e-05,
+      "loss": 0.6783,
+      "step": 2263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9551878779875233e-05,
+      "loss": 0.7009,
+      "step": 2264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9536944032444475e-05,
+      "loss": 0.7947,
+      "step": 2265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9522009543355507e-05,
+      "loss": 0.6947,
+      "step": 2266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9507075320940384e-05,
+      "loss": 0.7616,
+      "step": 2267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.949214137353099e-05,
+      "loss": 0.8526,
+      "step": 2268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.947720770945908e-05,
+      "loss": 0.708,
+      "step": 2269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9462274337056226e-05,
+      "loss": 0.8598,
+      "step": 2270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9447341264653853e-05,
+      "loss": 0.824,
+      "step": 2271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.943240850058322e-05,
+      "loss": 0.9886,
+      "step": 2272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.94174760531754e-05,
+      "loss": 0.6544,
+      "step": 2273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9402543930761304e-05,
+      "loss": 0.6154,
+      "step": 2274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.938761214167165e-05,
+      "loss": 0.6779,
+      "step": 2275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9372680694236986e-05,
+      "loss": 0.6865,
+      "step": 2276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9357749596787648e-05,
+      "loss": 0.9711,
+      "step": 2277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9342818857653795e-05,
+      "loss": 0.5797,
+      "step": 2278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9327888485165374e-05,
+      "loss": 0.6835,
+      "step": 2279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.931295848765214e-05,
+      "loss": 0.7297,
+      "step": 2280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9298028873443625e-05,
+      "loss": 0.7386,
+      "step": 2281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9283099650869145e-05,
+      "loss": 0.7721,
+      "step": 2282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.926817082825783e-05,
+      "loss": 0.7155,
+      "step": 2283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9253242413938532e-05,
+      "loss": 0.6676,
+      "step": 2284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.923831441623994e-05,
+      "loss": 0.7546,
+      "step": 2285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.922338684349044e-05,
+      "loss": 0.9437,
+      "step": 2286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.920845970401825e-05,
+      "loss": 0.6235,
+      "step": 2287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9193533006151287e-05,
+      "loss": 1.0019,
+      "step": 2288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9178606758217273e-05,
+      "loss": 0.7688,
+      "step": 2289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.916368096854363e-05,
+      "loss": 0.8192,
+      "step": 2290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9148755645457573e-05,
+      "loss": 1.1423,
+      "step": 2291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9133830797286013e-05,
+      "loss": 0.8919,
+      "step": 2292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9118906432355626e-05,
+      "loss": 0.6649,
+      "step": 2293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9103982558992803e-05,
+      "loss": 0.7534,
+      "step": 2294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.908905918552366e-05,
+      "loss": 0.9028,
+      "step": 2295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9074136320274053e-05,
+      "loss": 0.6448,
+      "step": 2296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.905921397156953e-05,
+      "loss": 0.6398,
+      "step": 2297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.904429214773537e-05,
+      "loss": 1.0216,
+      "step": 2298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.902937085709654e-05,
+      "loss": 0.749,
+      "step": 2299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9014450107977736e-05,
+      "loss": 0.6522,
+      "step": 2300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8999529908703325e-05,
+      "loss": 0.6724,
+      "step": 2301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8984610267597385e-05,
+      "loss": 0.6413,
+      "step": 2302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8969691192983665e-05,
+      "loss": 0.6078,
+      "step": 2303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8954772693185625e-05,
+      "loss": 0.6422,
+      "step": 2304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.893985477652638e-05,
+      "loss": 1.0162,
+      "step": 2305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8924937451328723e-05,
+      "loss": 0.6965,
+      "step": 2306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891002072591514e-05,
+      "loss": 0.7187,
+      "step": 2307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889510460860774e-05,
+      "loss": 0.934,
+      "step": 2308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880189107728336e-05,
+      "loss": 0.7,
+      "step": 2309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.886527423159837e-05,
+      "loss": 0.7118,
+      "step": 2310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8850359988538952e-05,
+      "loss": 0.618,
+      "step": 2311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8835446386870818e-05,
+      "loss": 0.7054,
+      "step": 2312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8820533434914373e-05,
+      "loss": 0.6724,
+      "step": 2313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8805621140989636e-05,
+      "loss": 1.1074,
+      "step": 2314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8790709513416276e-05,
+      "loss": 0.5612,
+      "step": 2315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8775798560513575e-05,
+      "loss": 0.6627,
+      "step": 2316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8760888290600442e-05,
+      "loss": 1.0211,
+      "step": 2317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.874597871199542e-05,
+      "loss": 0.6249,
+      "step": 2318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.873106983301665e-05,
+      "loss": 0.6892,
+      "step": 2319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.871616166198189e-05,
+      "loss": 1.0117,
+      "step": 2320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.870125420720849e-05,
+      "loss": 0.854,
+      "step": 2321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8686347477013434e-05,
+      "loss": 0.6892,
+      "step": 2322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.867144147971326e-05,
+      "loss": 0.6963,
+      "step": 2323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8656536223624127e-05,
+      "loss": 0.5811,
+      "step": 2324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.864163171706176e-05,
+      "loss": 0.6497,
+      "step": 2325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.862672796834149e-05,
+      "loss": 0.7476,
+      "step": 2326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8611824985778197e-05,
+      "loss": 0.6178,
+      "step": 2327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8596922777686348e-05,
+      "loss": 0.9216,
+      "step": 2328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8582021352379987e-05,
+      "loss": 0.598,
+      "step": 2329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8567120718172702e-05,
+      "loss": 0.7514,
+      "step": 2330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8552220883377657e-05,
+      "loss": 0.9447,
+      "step": 2331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.853732185630756e-05,
+      "loss": 0.6478,
+      "step": 2332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8522423645274667e-05,
+      "loss": 0.6237,
+      "step": 2333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8507526258590783e-05,
+      "loss": 0.6095,
+      "step": 2334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.849262970456726e-05,
+      "loss": 0.627,
+      "step": 2335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8477733991514967e-05,
+      "loss": 0.771,
+      "step": 2336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8462839127744324e-05,
+      "loss": 0.7628,
+      "step": 2337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8447945121565257e-05,
+      "loss": 0.6802,
+      "step": 2338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8433051981287236e-05,
+      "loss": 0.6135,
+      "step": 2339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8418159715219236e-05,
+      "loss": 1.1133,
+      "step": 2340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.840326833166973e-05,
+      "loss": 0.6987,
+      "step": 2341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8388377838946732e-05,
+      "loss": 0.6119,
+      "step": 2342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8373488245357716e-05,
+      "loss": 0.865,
+      "step": 2343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8358599559209713e-05,
+      "loss": 1.0367,
+      "step": 2344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.834371178880918e-05,
+      "loss": 0.9162,
+      "step": 2345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.832882494246212e-05,
+      "loss": 0.6766,
+      "step": 2346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8313939028473976e-05,
+      "loss": 0.6713,
+      "step": 2347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8299054055149715e-05,
+      "loss": 0.7242,
+      "step": 2348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8284170030793735e-05,
+      "loss": 1.0032,
+      "step": 2349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8269286963709943e-05,
+      "loss": 0.9383,
+      "step": 2350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8254404862201688e-05,
+      "loss": 0.7558,
+      "step": 2351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8239523734571776e-05,
+      "loss": 0.7138,
+      "step": 2352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8224643589122493e-05,
+      "loss": 0.7183,
+      "step": 2353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8209764434155558e-05,
+      "loss": 0.7228,
+      "step": 2354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.819488627797215e-05,
+      "loss": 0.688,
+      "step": 2355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8180009128872878e-05,
+      "loss": 0.6251,
+      "step": 2356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.81651329951578e-05,
+      "loss": 0.5744,
+      "step": 2357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8150257885126398e-05,
+      "loss": 0.7167,
+      "step": 2358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.81353838070776e-05,
+      "loss": 0.6158,
+      "step": 2359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.812051076930973e-05,
+      "loss": 0.7362,
+      "step": 2360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8105638780120565e-05,
+      "loss": 0.6475,
+      "step": 2361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.809076784780727e-05,
+      "loss": 0.6808,
+      "step": 2362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8075897980666424e-05,
+      "loss": 0.7329,
+      "step": 2363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8061029186994034e-05,
+      "loss": 1.0223,
+      "step": 2364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8046161475085476e-05,
+      "loss": 1.2047,
+      "step": 2365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8031294853235554e-05,
+      "loss": 0.8012,
+      "step": 2366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.801642932973844e-05,
+      "loss": 1.0148,
+      "step": 2367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.80015649128877e-05,
+      "loss": 0.769,
+      "step": 2368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.798670161097629e-05,
+      "loss": 0.6822,
+      "step": 2369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7971839432296536e-05,
+      "loss": 0.7357,
+      "step": 2370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7956978385140136e-05,
+      "loss": 0.583,
+      "step": 2371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7942118477798168e-05,
+      "loss": 0.6874,
+      "step": 2372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7927259718561056e-05,
+      "loss": 0.5837,
+      "step": 2373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.791240211571861e-05,
+      "loss": 0.686,
+      "step": 2374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7897545677559966e-05,
+      "loss": 0.6919,
+      "step": 2375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7882690412373616e-05,
+      "loss": 0.8414,
+      "step": 2376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7867836328447427e-05,
+      "loss": 0.7923,
+      "step": 2377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.785298343406856e-05,
+      "loss": 0.6293,
+      "step": 2378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.783813173752356e-05,
+      "loss": 0.6951,
+      "step": 2379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7823281247098258e-05,
+      "loss": 0.7447,
+      "step": 2380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7808431971077855e-05,
+      "loss": 0.7304,
+      "step": 2381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7793583917746834e-05,
+      "loss": 0.707,
+      "step": 2382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7778737095389043e-05,
+      "loss": 0.8709,
+      "step": 2383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7763891512287586e-05,
+      "loss": 0.6098,
+      "step": 2384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7749047176724935e-05,
+      "loss": 0.7006,
+      "step": 2385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7734204096982814e-05,
+      "loss": 0.6984,
+      "step": 2386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7719362281342272e-05,
+      "loss": 0.9834,
+      "step": 2387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7704521738083667e-05,
+      "loss": 1.1501,
+      "step": 2388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.768968247548661e-05,
+      "loss": 0.7075,
+      "step": 2389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.767484450183003e-05,
+      "loss": 0.6293,
+      "step": 2390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7660007825392115e-05,
+      "loss": 0.5924,
+      "step": 2391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7645172454450352e-05,
+      "loss": 0.7449,
+      "step": 2392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.763033839728147e-05,
+      "loss": 0.7244,
+      "step": 2393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.76155056621615e-05,
+      "loss": 1.0147,
+      "step": 2394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.76006742573657e-05,
+      "loss": 0.6628,
+      "step": 2395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7585844191168615e-05,
+      "loss": 0.6383,
+      "step": 2396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7571015471844026e-05,
+      "loss": 1.1082,
+      "step": 2397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7556188107664966e-05,
+      "loss": 0.6767,
+      "step": 2398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7541362106903714e-05,
+      "loss": 1.171,
+      "step": 2399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7526537477831784e-05,
+      "loss": 0.6655,
+      "step": 2400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.751171422871994e-05,
+      "loss": 0.5961,
+      "step": 2401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7496892367838152e-05,
+      "loss": 0.9407,
+      "step": 2402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.748207190345564e-05,
+      "loss": 0.7585,
+      "step": 2403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7467252843840822e-05,
+      "loss": 0.8562,
+      "step": 2404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.745243519726135e-05,
+      "loss": 0.6807,
+      "step": 2405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7437618971984073e-05,
+      "loss": 0.6524,
+      "step": 2406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.742280417627507e-05,
+      "loss": 0.8215,
+      "step": 2407
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.740799081839959e-05,
+      "loss": 1.0766,
+      "step": 2408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7393178906622113e-05,
+      "loss": 0.7469,
+      "step": 2409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.737836844920629e-05,
+      "loss": 0.6057,
+      "step": 2410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7363559454414957e-05,
+      "loss": 0.6667,
+      "step": 2411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.734875193051016e-05,
+      "loss": 1.0203,
+      "step": 2412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7333945885753093e-05,
+      "loss": 0.6466,
+      "step": 2413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7319141328404157e-05,
+      "loss": 0.7886,
+      "step": 2414
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.730433826672289e-05,
+      "loss": 0.6612,
+      "step": 2415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.728953670896803e-05,
+      "loss": 0.6603,
+      "step": 2416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7274736663397435e-05,
+      "loss": 0.8797,
+      "step": 2417
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7259938138268165e-05,
+      "loss": 0.7674,
+      "step": 2418
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.724514114183639e-05,
+      "loss": 0.7991,
+      "step": 2419
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7230345682357467e-05,
+      "loss": 0.7136,
+      "step": 2420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7215551768085856e-05,
+      "loss": 0.7641,
+      "step": 2421
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7200759407275177e-05,
+      "loss": 0.6899,
+      "step": 2422
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7185968608178188e-05,
+      "loss": 0.7085,
+      "step": 2423
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7171179379046758e-05,
+      "loss": 0.6738,
+      "step": 2424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7156391728131896e-05,
+      "loss": 0.6586,
+      "step": 2425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7141605663683717e-05,
+      "loss": 0.6018,
+      "step": 2426
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7126821193951467e-05,
+      "loss": 1.0117,
+      "step": 2427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.711203832718349e-05,
+      "loss": 0.5889,
+      "step": 2428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.709725707162724e-05,
+      "loss": 0.5639,
+      "step": 2429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7082477435529263e-05,
+      "loss": 0.7393,
+      "step": 2430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7067699427135222e-05,
+      "loss": 0.7083,
+      "step": 2431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7052923054689855e-05,
+      "loss": 1.0699,
+      "step": 2432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7038148326436984e-05,
+      "loss": 0.7414,
+      "step": 2433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7023375250619534e-05,
+      "loss": 0.581,
+      "step": 2434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7008603835479486e-05,
+      "loss": 0.6407,
+      "step": 2435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.699383408925791e-05,
+      "loss": 0.635,
+      "step": 2436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.697906602019493e-05,
+      "loss": 0.7342,
+      "step": 2437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6964299636529763e-05,
+      "loss": 0.595,
+      "step": 2438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.694953494650064e-05,
+      "loss": 0.7004,
+      "step": 2439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6934771958344894e-05,
+      "loss": 0.7693,
+      "step": 2440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6920010680298877e-05,
+      "loss": 0.9298,
+      "step": 2441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6905251120598e-05,
+      "loss": 0.6043,
+      "step": 2442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6890493287476715e-05,
+      "loss": 0.7138,
+      "step": 2443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6875737189168508e-05,
+      "loss": 0.8515,
+      "step": 2444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6860982833905902e-05,
+      "loss": 0.6844,
+      "step": 2445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6846230229920422e-05,
+      "loss": 1.1233,
+      "step": 2446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.683147938544267e-05,
+      "loss": 0.6484,
+      "step": 2447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.68167303087022e-05,
+      "loss": 0.7292,
+      "step": 2448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.680198300792765e-05,
+      "loss": 1.1129,
+      "step": 2449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6787237491346596e-05,
+      "loss": 0.7671,
+      "step": 2450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6772493767185672e-05,
+      "loss": 1.0842,
+      "step": 2451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6757751843670487e-05,
+      "loss": 0.6747,
+      "step": 2452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6743011729025654e-05,
+      "loss": 0.5527,
+      "step": 2453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6728273431474775e-05,
+      "loss": 0.6582,
+      "step": 2454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.671353695924044e-05,
+      "loss": 0.573,
+      "step": 2455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.669880232054421e-05,
+      "loss": 1.0309,
+      "step": 2456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6684069523606634e-05,
+      "loss": 0.6825,
+      "step": 2457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6669338576647237e-05,
+      "loss": 0.6609,
+      "step": 2458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.66546094878845e-05,
+      "loss": 0.7819,
+      "step": 2459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6639882265535875e-05,
+      "loss": 0.7248,
+      "step": 2460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6625156917817767e-05,
+      "loss": 0.9692,
+      "step": 2461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6610433452945545e-05,
+      "loss": 0.6294,
+      "step": 2462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6595711879133515e-05,
+      "loss": 0.9855,
+      "step": 2463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6580992204594944e-05,
+      "loss": 0.6566,
+      "step": 2464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6566274437542014e-05,
+      "loss": 0.7919,
+      "step": 2465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6551558586185876e-05,
+      "loss": 0.7247,
+      "step": 2466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6536844658736586e-05,
+      "loss": 0.7845,
+      "step": 2467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6522132663403133e-05,
+      "loss": 0.6537,
+      "step": 2468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6507422608393433e-05,
+      "loss": 0.8947,
+      "step": 2469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6492714501914317e-05,
+      "loss": 1.0046,
+      "step": 2470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.647800835217153e-05,
+      "loss": 0.7179,
+      "step": 2471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.646330416736972e-05,
+      "loss": 0.6835,
+      "step": 2472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6448601955712455e-05,
+      "loss": 0.6648,
+      "step": 2473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.643390172540217e-05,
+      "loss": 0.982,
+      "step": 2474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6419203484640233e-05,
+      "loss": 0.8966,
+      "step": 2475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.640450724162687e-05,
+      "loss": 0.7178,
+      "step": 2476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.638981300456122e-05,
+      "loss": 0.9026,
+      "step": 2477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6375120781641282e-05,
+      "loss": 1.0217,
+      "step": 2478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.636043058106393e-05,
+      "loss": 1.0166,
+      "step": 2479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.634574241102493e-05,
+      "loss": 0.6392,
+      "step": 2480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6331056279718895e-05,
+      "loss": 0.5631,
+      "step": 2481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6316372195339315e-05,
+      "loss": 0.6803,
+      "step": 2482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.630169016607853e-05,
+      "loss": 0.6075,
+      "step": 2483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6287010200127735e-05,
+      "loss": 0.8304,
+      "step": 2484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.627233230567697e-05,
+      "loss": 0.6736,
+      "step": 2485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.625765649091513e-05,
+      "loss": 0.8437,
+      "step": 2486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6242982764029937e-05,
+      "loss": 0.746,
+      "step": 2487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.622831113320796e-05,
+      "loss": 0.631,
+      "step": 2488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6213641606634587e-05,
+      "loss": 0.8806,
+      "step": 2489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6198974192494045e-05,
+      "loss": 0.6289,
+      "step": 2490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.618430889896937e-05,
+      "loss": 0.9277,
+      "step": 2491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6169645734242418e-05,
+      "loss": 0.6258,
+      "step": 2492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6154984706493867e-05,
+      "loss": 0.7954,
+      "step": 2493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.614032582390319e-05,
+      "loss": 0.7347,
+      "step": 2494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6125669094648667e-05,
+      "loss": 0.6121,
+      "step": 2495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.611101452690738e-05,
+      "loss": 0.8969,
+      "step": 2496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.609636212885521e-05,
+      "loss": 0.7846,
+      "step": 2497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6081711908666807e-05,
+      "loss": 1.0622,
+      "step": 2498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6067063874515635e-05,
+      "loss": 0.6926,
+      "step": 2499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6052418034573908e-05,
+      "loss": 0.889,
+      "step": 2500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6037774397012643e-05,
+      "loss": 0.9627,
+      "step": 2501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6023132970001616e-05,
+      "loss": 0.978,
+      "step": 2502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6008493761709352e-05,
+      "loss": 0.622,
+      "step": 2503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5993856780303186e-05,
+      "loss": 0.6945,
+      "step": 2504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5979222033949146e-05,
+      "loss": 0.7848,
+      "step": 2505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.596458953081208e-05,
+      "loss": 0.69,
+      "step": 2506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.594995927905553e-05,
+      "loss": 0.5798,
+      "step": 2507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5935331286841824e-05,
+      "loss": 0.6688,
+      "step": 2508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5920705562331982e-05,
+      "loss": 0.7272,
+      "step": 2509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5906082113685815e-05,
+      "loss": 0.7225,
+      "step": 2510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5891460949061812e-05,
+      "loss": 0.8702,
+      "step": 2511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5876842076617235e-05,
+      "loss": 0.6894,
+      "step": 2512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.586222550450803e-05,
+      "loss": 0.6705,
+      "step": 2513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5847611240888866e-05,
+      "loss": 0.6938,
+      "step": 2514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5832999293913144e-05,
+      "loss": 0.6861,
+      "step": 2515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.581838967173295e-05,
+      "loss": 0.6765,
+      "step": 2516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5803782382499096e-05,
+      "loss": 0.8994,
+      "step": 2517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5789177434361065e-05,
+      "loss": 0.6973,
+      "step": 2518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5774574835467057e-05,
+      "loss": 0.6959,
+      "step": 2519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.575997459396395e-05,
+      "loss": 0.5714,
+      "step": 2520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.574537671799731e-05,
+      "loss": 0.7619,
+      "step": 2521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.573078121571138e-05,
+      "loss": 0.8595,
+      "step": 2522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5716188095249086e-05,
+      "loss": 0.6348,
+      "step": 2523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5701597364752015e-05,
+      "loss": 0.6136,
+      "step": 2524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.568700903236043e-05,
+      "loss": 0.7,
+      "step": 2525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.567242310621325e-05,
+      "loss": 0.6269,
+      "step": 2526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5657839594448056e-05,
+      "loss": 0.5866,
+      "step": 2527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5643258505201077e-05,
+      "loss": 0.5765,
+      "step": 2528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.562867984660719e-05,
+      "loss": 0.7065,
+      "step": 2529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5614103626799927e-05,
+      "loss": 0.6561,
+      "step": 2530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5599529853911442e-05,
+      "loss": 0.6777,
+      "step": 2531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.558495853607255e-05,
+      "loss": 0.7093,
+      "step": 2532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5570389681412655e-05,
+      "loss": 0.6906,
+      "step": 2533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5555823298059833e-05,
+      "loss": 0.6227,
+      "step": 2534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5541259394140745e-05,
+      "loss": 0.7202,
+      "step": 2535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.55266979777807e-05,
+      "loss": 0.9921,
+      "step": 2536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.551213905710359e-05,
+      "loss": 0.6608,
+      "step": 2537
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.549758264023192e-05,
+      "loss": 0.8273,
+      "step": 2538
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5483028735286822e-05,
+      "loss": 0.7043,
+      "step": 2539
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5468477350387992e-05,
+      "loss": 0.5976,
+      "step": 2540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5453928493653764e-05,
+      "loss": 0.7383,
+      "step": 2541
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5439382173201003e-05,
+      "loss": 0.724,
+      "step": 2542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5424838397145223e-05,
+      "loss": 0.658,
+      "step": 2543
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.541029717360046e-05,
+      "loss": 0.6518,
+      "step": 2544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.539575851067937e-05,
+      "loss": 0.746,
+      "step": 2545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5381222416493152e-05,
+      "loss": 0.707,
+      "step": 2546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.53666888991516e-05,
+      "loss": 0.9304,
+      "step": 2547
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5352157966763034e-05,
+      "loss": 0.6646,
+      "step": 2548
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.533762962743436e-05,
+      "loss": 0.7921,
+      "step": 2549
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.532310388927103e-05,
+      "loss": 0.935,
+      "step": 2550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5308580760377038e-05,
+      "loss": 0.6032,
+      "step": 2551
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5294060248854936e-05,
+      "loss": 0.8213,
+      "step": 2552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5279542362805797e-05,
+      "loss": 0.6684,
+      "step": 2553
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5265027110329247e-05,
+      "loss": 0.7931,
+      "step": 2554
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5250514499523434e-05,
+      "loss": 0.5881,
+      "step": 2555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5236004538485037e-05,
+      "loss": 0.6013,
+      "step": 2556
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5221497235309244e-05,
+      "loss": 0.7264,
+      "step": 2557
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5206992598089778e-05,
+      "loss": 0.7472,
+      "step": 2558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.519249063491886e-05,
+      "loss": 0.6194,
+      "step": 2559
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5177991353887231e-05,
+      "loss": 0.673,
+      "step": 2560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5163494763084128e-05,
+      "loss": 0.6123,
+      "step": 2561
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5149000870597284e-05,
+      "loss": 0.6937,
+      "step": 2562
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5134509684512937e-05,
+      "loss": 0.5897,
+      "step": 2563
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5120021212915802e-05,
+      "loss": 0.7236,
+      "step": 2564
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5105535463889103e-05,
+      "loss": 0.6721,
+      "step": 2565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5091052445514511e-05,
+      "loss": 0.8984,
+      "step": 2566
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5076572165872205e-05,
+      "loss": 0.6765,
+      "step": 2567
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5062094633040813e-05,
+      "loss": 0.7981,
+      "step": 2568
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5047619855097451e-05,
+      "loss": 0.7774,
+      "step": 2569
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5033147840117674e-05,
+      "loss": 0.7487,
+      "step": 2570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5018678596175523e-05,
+      "loss": 0.8912,
+      "step": 2571
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5004212131343476e-05,
+      "loss": 0.7414,
+      "step": 2572
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4989748453692445e-05,
+      "loss": 0.6727,
+      "step": 2573
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4975287571291834e-05,
+      "loss": 0.672,
+      "step": 2574
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4960829492209435e-05,
+      "loss": 0.6363,
+      "step": 2575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.494637422451152e-05,
+      "loss": 0.6024,
+      "step": 2576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4931921776262749e-05,
+      "loss": 0.8497,
+      "step": 2577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4917472155526263e-05,
+      "loss": 0.6781,
+      "step": 2578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4903025370363562e-05,
+      "loss": 0.5703,
+      "step": 2579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4888581428834626e-05,
+      "loss": 1.0258,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4874140338997793e-05,
+      "loss": 0.8157,
+      "step": 2581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4859702108909863e-05,
+      "loss": 0.7012,
+      "step": 2582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4845266746625999e-05,
+      "loss": 0.7355,
+      "step": 2583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4830834260199775e-05,
+      "loss": 0.6439,
+      "step": 2584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4816404657683179e-05,
+      "loss": 0.6839,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4801977947126558e-05,
+      "loss": 1.2481,
+      "step": 2586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4787554136578682e-05,
+      "loss": 0.5764,
+      "step": 2587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.477313323408667e-05,
+      "loss": 0.6839,
+      "step": 2588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.475871524769604e-05,
+      "loss": 0.619,
+      "step": 2589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4744300185450673e-05,
+      "loss": 0.623,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4729888055392827e-05,
+      "loss": 0.7631,
+      "step": 2591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4715478865563109e-05,
+      "loss": 0.6148,
+      "step": 2592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4701072624000508e-05,
+      "loss": 0.708,
+      "step": 2593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4686669338742341e-05,
+      "loss": 0.7172,
+      "step": 2594
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4672269017824303e-05,
+      "loss": 0.7768,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4657871669280418e-05,
+      "loss": 0.8445,
+      "step": 2596
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.464347730114305e-05,
+      "loss": 0.5519,
+      "step": 2597
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4629085921442916e-05,
+      "loss": 0.8913,
+      "step": 2598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4614697538209045e-05,
+      "loss": 1.1055,
+      "step": 2599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4600312159468819e-05,
+      "loss": 0.7606,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4585929793247917e-05,
+      "loss": 0.9334,
+      "step": 2601
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4571550447570359e-05,
+      "loss": 0.642,
+      "step": 2602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4557174130458465e-05,
+      "loss": 0.7064,
+      "step": 2603
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4542800849932879e-05,
+      "loss": 0.6264,
+      "step": 2604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4528430614012532e-05,
+      "loss": 0.8339,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4514063430714683e-05,
+      "loss": 0.6867,
+      "step": 2606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4499699308054865e-05,
+      "loss": 0.6068,
+      "step": 2607
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4485338254046903e-05,
+      "loss": 0.6332,
+      "step": 2608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4470980276702929e-05,
+      "loss": 0.8395,
+      "step": 2609
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.445662538403334e-05,
+      "loss": 0.6566,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4442273584046827e-05,
+      "loss": 1.0549,
+      "step": 2611
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4427924884750342e-05,
+      "loss": 0.9468,
+      "step": 2612
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4413579294149118e-05,
+      "loss": 0.8095,
+      "step": 2613
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4399236820246644e-05,
+      "loss": 0.5279,
+      "step": 2614
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4384897471044683e-05,
+      "loss": 0.9848,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4370561254543236e-05,
+      "loss": 0.751,
+      "step": 2616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4356228178740579e-05,
+      "loss": 0.6341,
+      "step": 2617
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4341898251633218e-05,
+      "loss": 0.6886,
+      "step": 2618
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4327571481215903e-05,
+      "loss": 0.8747,
+      "step": 2619
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4313247875481639e-05,
+      "loss": 0.7638,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4298927442421641e-05,
+      "loss": 0.6877,
+      "step": 2621
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4284610190025383e-05,
+      "loss": 0.7637,
+      "step": 2622
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4270296126280536e-05,
+      "loss": 0.7059,
+      "step": 2623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4255985259173013e-05,
+      "loss": 0.6261,
+      "step": 2624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4241677596686927e-05,
+      "loss": 0.6334,
+      "step": 2625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4227373146804625e-05,
+      "loss": 0.6917,
+      "step": 2626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4213071917506634e-05,
+      "loss": 0.6486,
+      "step": 2627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4198773916771709e-05,
+      "loss": 0.694,
+      "step": 2628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4184479152576782e-05,
+      "loss": 0.9748,
+      "step": 2629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4170187632897006e-05,
+      "loss": 0.9867,
+      "step": 2630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4155899365705703e-05,
+      "loss": 0.814,
+      "step": 2631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4141614358974369e-05,
+      "loss": 0.7457,
+      "step": 2632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4127332620672727e-05,
+      "loss": 0.988,
+      "step": 2633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4113054158768616e-05,
+      "loss": 0.5639,
+      "step": 2634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4098778981228109e-05,
+      "loss": 0.9089,
+      "step": 2635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4084507096015389e-05,
+      "loss": 0.4732,
+      "step": 2636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4070238511092855e-05,
+      "loss": 0.7691,
+      "step": 2637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4055973234421015e-05,
+      "loss": 0.7165,
+      "step": 2638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4041711273958573e-05,
+      "loss": 0.6115,
+      "step": 2639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4027452637662352e-05,
+      "loss": 0.6954,
+      "step": 2640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4013197333487351e-05,
+      "loss": 0.6409,
+      "step": 2641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3998945369386676e-05,
+      "loss": 0.6772,
+      "step": 2642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3984696753311588e-05,
+      "loss": 0.6258,
+      "step": 2643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3970451493211488e-05,
+      "loss": 0.6815,
+      "step": 2644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.395620959703388e-05,
+      "loss": 0.6433,
+      "step": 2645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3941971072724425e-05,
+      "loss": 0.9366,
+      "step": 2646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3927735928226867e-05,
+      "loss": 0.6968,
+      "step": 2647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3913504171483095e-05,
+      "loss": 0.5565,
+      "step": 2648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.389927581043308e-05,
+      "loss": 0.6511,
+      "step": 2649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3885050853014928e-05,
+      "loss": 1.0545,
+      "step": 2650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3870829307164816e-05,
+      "loss": 1.142,
+      "step": 2651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.385661118081705e-05,
+      "loss": 0.9014,
+      "step": 2652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3842396481903997e-05,
+      "loss": 0.6944,
+      "step": 2653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3828185218356127e-05,
+      "loss": 0.8028,
+      "step": 2654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3813977398102001e-05,
+      "loss": 0.9579,
+      "step": 2655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3799773029068235e-05,
+      "loss": 0.8663,
+      "step": 2656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.378557211917955e-05,
+      "loss": 0.6885,
+      "step": 2657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3771374676358713e-05,
+      "loss": 0.7385,
+      "step": 2658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3757180708526572e-05,
+      "loss": 0.6892,
+      "step": 2659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3742990223602017e-05,
+      "loss": 1.0949,
+      "step": 2660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3728803229502022e-05,
+      "loss": 0.9439,
+      "step": 2661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.371461973414159e-05,
+      "loss": 0.5161,
+      "step": 2662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3700439745433787e-05,
+      "loss": 0.5995,
+      "step": 2663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3686263271289717e-05,
+      "loss": 0.9348,
+      "step": 2664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3672090319618504e-05,
+      "loss": 0.6949,
+      "step": 2665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.365792089832736e-05,
+      "loss": 0.8854,
+      "step": 2666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3643755015321455e-05,
+      "loss": 0.9945,
+      "step": 2667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.362959267850406e-05,
+      "loss": 0.701,
+      "step": 2668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3615433895776395e-05,
+      "loss": 0.5807,
+      "step": 2669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3601278675037764e-05,
+      "loss": 0.7638,
+      "step": 2670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3587127024185426e-05,
+      "loss": 0.5363,
+      "step": 2671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3572978951114701e-05,
+      "loss": 0.7427,
+      "step": 2672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3558834463718859e-05,
+      "loss": 0.9206,
+      "step": 2673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3544693569889225e-05,
+      "loss": 0.8714,
+      "step": 2674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3530556277515063e-05,
+      "loss": 0.6154,
+      "step": 2675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3516422594483687e-05,
+      "loss": 0.7835,
+      "step": 2676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3502292528680343e-05,
+      "loss": 0.6382,
+      "step": 2677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3488166087988284e-05,
+      "loss": 0.6814,
+      "step": 2678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.347404328028875e-05,
+      "loss": 0.6126,
+      "step": 2679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3459924113460931e-05,
+      "loss": 0.8689,
+      "step": 2680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3445808595382011e-05,
+      "loss": 0.6596,
+      "step": 2681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3431696733927111e-05,
+      "loss": 0.7309,
+      "step": 2682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.341758853696934e-05,
+      "loss": 0.6082,
+      "step": 2683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3403484012379736e-05,
+      "loss": 0.7123,
+      "step": 2684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3389383168027311e-05,
+      "loss": 1.0135,
+      "step": 2685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3375286011779003e-05,
+      "loss": 0.6451,
+      "step": 2686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3361192551499715e-05,
+      "loss": 0.673,
+      "step": 2687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3347102795052268e-05,
+      "loss": 0.9422,
+      "step": 2688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.333301675029742e-05,
+      "loss": 0.6554,
+      "step": 2689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3318934425093876e-05,
+      "loss": 0.871,
+      "step": 2690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3304855827298242e-05,
+      "loss": 1.0291,
+      "step": 2691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.329078096476506e-05,
+      "loss": 0.6994,
+      "step": 2692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3276709845346778e-05,
+      "loss": 0.6969,
+      "step": 2693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3262642476893768e-05,
+      "loss": 0.9332,
+      "step": 2694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3248578867254296e-05,
+      "loss": 0.9483,
+      "step": 2695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3234519024274542e-05,
+      "loss": 0.7253,
+      "step": 2696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3220462955798574e-05,
+      "loss": 0.7288,
+      "step": 2697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3206410669668363e-05,
+      "loss": 0.6672,
+      "step": 2698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3192362173723774e-05,
+      "loss": 0.6108,
+      "step": 2699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3178317475802529e-05,
+      "loss": 0.666,
+      "step": 2700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3164276583740279e-05,
+      "loss": 1.022,
+      "step": 2701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3150239505370497e-05,
+      "loss": 0.7675,
+      "step": 2702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.313620624852458e-05,
+      "loss": 0.7007,
+      "step": 2703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3122176821031752e-05,
+      "loss": 0.8297,
+      "step": 2704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3108151230719132e-05,
+      "loss": 1.0253,
+      "step": 2705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3094129485411666e-05,
+      "loss": 0.6532,
+      "step": 2706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3080111592932196e-05,
+      "loss": 0.6132,
+      "step": 2707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3066097561101367e-05,
+      "loss": 0.6058,
+      "step": 2708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3052087397737718e-05,
+      "loss": 0.6408,
+      "step": 2709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3038081110657581e-05,
+      "loss": 0.9138,
+      "step": 2710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3024078707675175e-05,
+      "loss": 0.796,
+      "step": 2711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3010080196602508e-05,
+      "loss": 0.719,
+      "step": 2712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2996085585249442e-05,
+      "loss": 0.6514,
+      "step": 2713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2982094881423667e-05,
+      "loss": 0.9091,
+      "step": 2714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2968108092930667e-05,
+      "loss": 0.7203,
+      "step": 2715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.295412522757377e-05,
+      "loss": 0.6462,
+      "step": 2716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2940146293154099e-05,
+      "loss": 0.5906,
+      "step": 2717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2926171297470589e-05,
+      "loss": 1.0368,
+      "step": 2718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2912200248319975e-05,
+      "loss": 0.8119,
+      "step": 2719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.28982331534968e-05,
+      "loss": 0.784,
+      "step": 2720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2884270020793379e-05,
+      "loss": 0.633,
+      "step": 2721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2870310857999844e-05,
+      "loss": 0.7002,
+      "step": 2722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2856355672904091e-05,
+      "loss": 0.6816,
+      "step": 2723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.28424044732918e-05,
+      "loss": 0.572,
+      "step": 2724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2828457266946447e-05,
+      "loss": 0.7479,
+      "step": 2725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2814514061649246e-05,
+      "loss": 0.6215,
+      "step": 2726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.280057486517921e-05,
+      "loss": 0.8379,
+      "step": 2727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2786639685313098e-05,
+      "loss": 0.6538,
+      "step": 2728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2772708529825442e-05,
+      "loss": 0.6368,
+      "step": 2729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2758781406488508e-05,
+      "loss": 0.692,
+      "step": 2730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2744858323072333e-05,
+      "loss": 0.717,
+      "step": 2731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2730939287344684e-05,
+      "loss": 0.657,
+      "step": 2732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2717024307071086e-05,
+      "loss": 0.7022,
+      "step": 2733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2703113390014793e-05,
+      "loss": 0.6239,
+      "step": 2734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2689206543936776e-05,
+      "loss": 0.6665,
+      "step": 2735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2675303776595773e-05,
+      "loss": 1.0588,
+      "step": 2736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2661405095748202e-05,
+      "loss": 1.034,
+      "step": 2737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2647510509148248e-05,
+      "loss": 1.0751,
+      "step": 2738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2633620024547759e-05,
+      "loss": 0.699,
+      "step": 2739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2619733649696348e-05,
+      "loss": 0.5773,
+      "step": 2740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.260585139234129e-05,
+      "loss": 0.5948,
+      "step": 2741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2591973260227604e-05,
+      "loss": 0.7447,
+      "step": 2742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.257809926109796e-05,
+      "loss": 1.0176,
+      "step": 2743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2564229402692768e-05,
+      "loss": 0.6897,
+      "step": 2744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2550363692750099e-05,
+      "loss": 0.5365,
+      "step": 2745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2536502139005721e-05,
+      "loss": 0.7419,
+      "step": 2746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2522644749193083e-05,
+      "loss": 0.9678,
+      "step": 2747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.25087915310433e-05,
+      "loss": 0.5619,
+      "step": 2748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2494942492285182e-05,
+      "loss": 0.7681,
+      "step": 2749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.248109764064518e-05,
+      "loss": 0.7722,
+      "step": 2750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2467256983847435e-05,
+      "loss": 0.5753,
+      "step": 2751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2453420529613723e-05,
+      "loss": 0.6474,
+      "step": 2752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2439588285663497e-05,
+      "loss": 1.0895,
+      "step": 2753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2425760259713843e-05,
+      "loss": 0.6717,
+      "step": 2754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2411936459479516e-05,
+      "loss": 0.5889,
+      "step": 2755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2398116892672886e-05,
+      "loss": 0.7545,
+      "step": 2756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2384301567003982e-05,
+      "loss": 0.7453,
+      "step": 2757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2370490490180462e-05,
+      "loss": 0.5945,
+      "step": 2758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.23566836699076e-05,
+      "loss": 0.7378,
+      "step": 2759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2342881113888323e-05,
+      "loss": 0.6535,
+      "step": 2760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.232908282982315e-05,
+      "loss": 0.5962,
+      "step": 2761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2315288825410235e-05,
+      "loss": 0.7397,
+      "step": 2762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2301499108345332e-05,
+      "loss": 0.7242,
+      "step": 2763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.228771368632182e-05,
+      "loss": 0.6786,
+      "step": 2764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2273932567030662e-05,
+      "loss": 0.6696,
+      "step": 2765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2260155758160441e-05,
+      "loss": 0.6598,
+      "step": 2766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2246383267397306e-05,
+      "loss": 0.6508,
+      "step": 2767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2232615102425038e-05,
+      "loss": 0.8942,
+      "step": 2768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2218851270924967e-05,
+      "loss": 0.6649,
+      "step": 2769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2205091780576018e-05,
+      "loss": 0.6304,
+      "step": 2770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2191336639054705e-05,
+      "loss": 0.5538,
+      "step": 2771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2177585854035097e-05,
+      "loss": 0.9732,
+      "step": 2772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2163839433188852e-05,
+      "loss": 0.6663,
+      "step": 2773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2150097384185174e-05,
+      "loss": 0.7044,
+      "step": 2774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2136359714690848e-05,
+      "loss": 1.0905,
+      "step": 2775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2122626432370193e-05,
+      "loss": 0.6202,
+      "step": 2776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2108897544885103e-05,
+      "loss": 0.6473,
+      "step": 2777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2095173059895004e-05,
+      "loss": 0.973,
+      "step": 2778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2081452985056874e-05,
+      "loss": 0.7388,
+      "step": 2779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2067737328025224e-05,
+      "loss": 0.6059,
+      "step": 2780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2054026096452112e-05,
+      "loss": 0.8711,
+      "step": 2781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2040319297987112e-05,
+      "loss": 0.6584,
+      "step": 2782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2026616940277327e-05,
+      "loss": 0.7298,
+      "step": 2783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2012919030967398e-05,
+      "loss": 0.7213,
+      "step": 2784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1999225577699467e-05,
+      "loss": 0.7543,
+      "step": 2785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1985536588113203e-05,
+      "loss": 0.6779,
+      "step": 2786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.197185206984577e-05,
+      "loss": 0.5814,
+      "step": 2787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.195817203053185e-05,
+      "loss": 0.9261,
+      "step": 2788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.194449647780362e-05,
+      "loss": 0.7363,
+      "step": 2789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1930825419290761e-05,
+      "loss": 0.5952,
+      "step": 2790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1917158862620432e-05,
+      "loss": 0.6527,
+      "step": 2791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1903496815417301e-05,
+      "loss": 1.002,
+      "step": 2792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1889839285303508e-05,
+      "loss": 0.7415,
+      "step": 2793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1876186279898654e-05,
+      "loss": 0.7854,
+      "step": 2794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1862537806819863e-05,
+      "loss": 1.0194,
+      "step": 2795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1848893873681678e-05,
+      "loss": 0.6645,
+      "step": 2796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1835254488096157e-05,
+      "loss": 0.8147,
+      "step": 2797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1821619657672771e-05,
+      "loss": 0.7211,
+      "step": 2798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1807989390018509e-05,
+      "loss": 1.0397,
+      "step": 2799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1794363692737751e-05,
+      "loss": 0.6317,
+      "step": 2800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1780742573432385e-05,
+      "loss": 0.6839,
+      "step": 2801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.176712603970169e-05,
+      "loss": 0.9962,
+      "step": 2802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1753514099142442e-05,
+      "loss": 0.6576,
+      "step": 2803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1739906759348812e-05,
+      "loss": 0.6006,
+      "step": 2804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1726304027912416e-05,
+      "loss": 0.5659,
+      "step": 2805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1712705912422311e-05,
+      "loss": 0.7229,
+      "step": 2806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1699112420464964e-05,
+      "loss": 0.7534,
+      "step": 2807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1685523559624277e-05,
+      "loss": 0.6718,
+      "step": 2808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1671939337481542e-05,
+      "loss": 0.6849,
+      "step": 2809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1658359761615503e-05,
+      "loss": 0.6566,
+      "step": 2810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1644784839602278e-05,
+      "loss": 0.711,
+      "step": 2811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1631214579015402e-05,
+      "loss": 0.6497,
+      "step": 2812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.16176489874258e-05,
+      "loss": 0.7638,
+      "step": 2813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1604088072401813e-05,
+      "loss": 0.6043,
+      "step": 2814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1590531841509163e-05,
+      "loss": 0.7365,
+      "step": 2815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1576980302310927e-05,
+      "loss": 0.8397,
+      "step": 2816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1563433462367622e-05,
+      "loss": 0.6199,
+      "step": 2817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1549891329237093e-05,
+      "loss": 0.7338,
+      "step": 2818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1536353910474607e-05,
+      "loss": 0.5811,
+      "step": 2819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1522821213632743e-05,
+      "loss": 0.989,
+      "step": 2820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1509293246261495e-05,
+      "loss": 0.8159,
+      "step": 2821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.14957700159082e-05,
+      "loss": 0.6256,
+      "step": 2822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1482251530117541e-05,
+      "loss": 0.6921,
+      "step": 2823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.146873779643157e-05,
+      "loss": 0.9746,
+      "step": 2824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1455228822389687e-05,
+      "loss": 0.5532,
+      "step": 2825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1441724615528633e-05,
+      "loss": 0.6649,
+      "step": 2826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1428225183382486e-05,
+      "loss": 0.7277,
+      "step": 2827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1414730533482663e-05,
+      "loss": 0.8445,
+      "step": 2828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1401240673357904e-05,
+      "loss": 0.7061,
+      "step": 2829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1387755610534301e-05,
+      "loss": 1.0354,
+      "step": 2830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1374275352535247e-05,
+      "loss": 0.6601,
+      "step": 2831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1360799906881463e-05,
+      "loss": 0.6739,
+      "step": 2832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1347329281090973e-05,
+      "loss": 0.6116,
+      "step": 2833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1333863482679139e-05,
+      "loss": 1.0479,
+      "step": 2834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1320402519158609e-05,
+      "loss": 0.7794,
+      "step": 2835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1306946398039336e-05,
+      "loss": 0.6996,
+      "step": 2836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1293495126828562e-05,
+      "loss": 0.6778,
+      "step": 2837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1280048713030856e-05,
+      "loss": 0.7187,
+      "step": 2838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1266607164148042e-05,
+      "loss": 0.9543,
+      "step": 2839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1253170487679249e-05,
+      "loss": 1.0632,
+      "step": 2840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1239738691120878e-05,
+      "loss": 0.6201,
+      "step": 2841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1226311781966608e-05,
+      "loss": 0.6972,
+      "step": 2842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1212889767707406e-05,
+      "loss": 0.6219,
+      "step": 2843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1199472655831492e-05,
+      "loss": 0.7639,
+      "step": 2844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1186060453824354e-05,
+      "loss": 0.5656,
+      "step": 2845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.117265316916874e-05,
+      "loss": 0.6793,
+      "step": 2846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.115925080934467e-05,
+      "loss": 0.6062,
+      "step": 2847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1145853381829386e-05,
+      "loss": 0.7606,
+      "step": 2848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1132460894097407e-05,
+      "loss": 0.7764,
+      "step": 2849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1119073353620485e-05,
+      "loss": 0.8047,
+      "step": 2850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.110569076786761e-05,
+      "loss": 1.0784,
+      "step": 2851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1092313144305006e-05,
+      "loss": 0.5947,
+      "step": 2852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1078940490396128e-05,
+      "loss": 0.6644,
+      "step": 2853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1065572813601681e-05,
+      "loss": 0.6959,
+      "step": 2854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1052210121379558e-05,
+      "loss": 0.5957,
+      "step": 2855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1038852421184898e-05,
+      "loss": 0.732,
+      "step": 2856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.102549972047003e-05,
+      "loss": 0.6868,
+      "step": 2857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1012152026684526e-05,
+      "loss": 0.577,
+      "step": 2858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0998809347275141e-05,
+      "loss": 0.6827,
+      "step": 2859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0985471689685836e-05,
+      "loss": 0.611,
+      "step": 2860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0972139061357767e-05,
+      "loss": 0.9437,
+      "step": 2861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0958811469729306e-05,
+      "loss": 0.9268,
+      "step": 2862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0945488922235988e-05,
+      "loss": 0.5791,
+      "step": 2863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.093217142631055e-05,
+      "loss": 0.7407,
+      "step": 2864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.09188589893829e-05,
+      "loss": 0.9861,
+      "step": 2865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0905551618880124e-05,
+      "loss": 0.6177,
+      "step": 2866
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0892249322226503e-05,
+      "loss": 0.7814,
+      "step": 2867
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0878952106843461e-05,
+      "loss": 1.0153,
+      "step": 2868
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0865659980149599e-05,
+      "loss": 0.7703,
+      "step": 2869
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0852372949560666e-05,
+      "loss": 0.7037,
+      "step": 2870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0839091022489599e-05,
+      "loss": 0.731,
+      "step": 2871
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0825814206346456e-05,
+      "loss": 0.5994,
+      "step": 2872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0812542508538458e-05,
+      "loss": 0.7637,
+      "step": 2873
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0799275936469962e-05,
+      "loss": 0.676,
+      "step": 2874
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0786014497542467e-05,
+      "loss": 0.818,
+      "step": 2875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0772758199154639e-05,
+      "loss": 0.5692,
+      "step": 2876
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.075950704870221e-05,
+      "loss": 0.7497,
+      "step": 2877
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.074626105357811e-05,
+      "loss": 0.7075,
+      "step": 2878
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0733020221172337e-05,
+      "loss": 0.9548,
+      "step": 2879
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0719784558872066e-05,
+      "loss": 0.6093,
+      "step": 2880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.070655407406152e-05,
+      "loss": 0.5457,
+      "step": 2881
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.069332877412209e-05,
+      "loss": 0.6096,
+      "step": 2882
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0680108666432245e-05,
+      "loss": 0.7488,
+      "step": 2883
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0666893758367578e-05,
+      "loss": 0.8587,
+      "step": 2884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0653684057300753e-05,
+      "loss": 0.6466,
+      "step": 2885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0640479570601538e-05,
+      "loss": 0.6283,
+      "step": 2886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0627280305636817e-05,
+      "loss": 0.6022,
+      "step": 2887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0614086269770533e-05,
+      "loss": 0.6619,
+      "step": 2888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0600897470363725e-05,
+      "loss": 0.6307,
+      "step": 2889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.058771391477449e-05,
+      "loss": 0.6066,
+      "step": 2890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0574535610358034e-05,
+      "loss": 0.5305,
+      "step": 2891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0561362564466611e-05,
+      "loss": 0.9376,
+      "step": 2892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0548194784449542e-05,
+      "loss": 0.6372,
+      "step": 2893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0535032277653203e-05,
+      "loss": 0.7318,
+      "step": 2894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0521875051421056e-05,
+      "loss": 0.792,
+      "step": 2895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0508723113093592e-05,
+      "loss": 0.6987,
+      "step": 2896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0495576470008361e-05,
+      "loss": 0.6781,
+      "step": 2897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0482435129499952e-05,
+      "loss": 0.8384,
+      "step": 2898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0469299098899994e-05,
+      "loss": 0.9686,
+      "step": 2899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0456168385537176e-05,
+      "loss": 0.7762,
+      "step": 2900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0443042996737198e-05,
+      "loss": 0.6268,
+      "step": 2901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0429922939822794e-05,
+      "loss": 0.6735,
+      "step": 2902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0416808222113716e-05,
+      "loss": 0.5259,
+      "step": 2903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0403698850926767e-05,
+      "loss": 0.6655,
+      "step": 2904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0390594833575734e-05,
+      "loss": 0.6795,
+      "step": 2905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0377496177371431e-05,
+      "loss": 0.6491,
+      "step": 2906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.036440288962167e-05,
+      "loss": 0.8759,
+      "step": 2907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0351314977631298e-05,
+      "loss": 0.7001,
+      "step": 2908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0338232448702134e-05,
+      "loss": 0.6449,
+      "step": 2909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0325155310132988e-05,
+      "loss": 0.5511,
+      "step": 2910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0312083569219696e-05,
+      "loss": 0.6847,
+      "step": 2911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0299017233255044e-05,
+      "loss": 0.8169,
+      "step": 2912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.028595630952885e-05,
+      "loss": 0.7433,
+      "step": 2913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0272900805327852e-05,
+      "loss": 0.5531,
+      "step": 2914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0259850727935823e-05,
+      "loss": 0.5739,
+      "step": 2915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.024680608463346e-05,
+      "loss": 0.7999,
+      "step": 2916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0233766882698478e-05,
+      "loss": 0.8004,
+      "step": 2917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0220733129405496e-05,
+      "loss": 0.5943,
+      "step": 2918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.020770483202615e-05,
+      "loss": 0.5407,
+      "step": 2919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0194681997828999e-05,
+      "loss": 0.7524,
+      "step": 2920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0181664634079565e-05,
+      "loss": 0.8184,
+      "step": 2921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0168652748040317e-05,
+      "loss": 0.7929,
+      "step": 2922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0155646346970654e-05,
+      "loss": 0.5856,
+      "step": 2923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0142645438126948e-05,
+      "loss": 0.6337,
+      "step": 2924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0129650028762476e-05,
+      "loss": 0.797,
+      "step": 2925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.011666012612746e-05,
+      "loss": 1.0093,
+      "step": 2926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0103675737469033e-05,
+      "loss": 0.873,
+      "step": 2927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0090696870031293e-05,
+      "loss": 0.6993,
+      "step": 2928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0077723531055214e-05,
+      "loss": 0.9254,
+      "step": 2929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0064755727778705e-05,
+      "loss": 0.7077,
+      "step": 2930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0051793467436577e-05,
+      "loss": 0.6339,
+      "step": 2931
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0038836757260568e-05,
+      "loss": 0.7526,
+      "step": 2932
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0025885604479302e-05,
+      "loss": 1.0509,
+      "step": 2933
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0012940016318306e-05,
+      "loss": 0.6955,
+      "step": 2934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 1.0301,
+      "step": 2935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.987065562743705e-06,
+      "loss": 0.7346,
+      "step": 2936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.974136711765624e-06,
+      "loss": 0.7166,
+      "step": 2937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.961213454278844e-06,
+      "loss": 0.8262,
+      "step": 2938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.948295797493328e-06,
+      "loss": 0.9132,
+      "step": 2939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.935383748615908e-06,
+      "loss": 0.7282,
+      "step": 2940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.922477314850314e-06,
+      "loss": 0.7373,
+      "step": 2941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.909576503397115e-06,
+      "loss": 0.7763,
+      "step": 2942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.896681321453756e-06,
+      "loss": 0.5851,
+      "step": 2943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.883791776214537e-06,
+      "loss": 0.5672,
+      "step": 2944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.87090787487061e-06,
+      "loss": 0.6099,
+      "step": 2945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.858029624610004e-06,
+      "loss": 0.8039,
+      "step": 2946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.845157032617547e-06,
+      "loss": 0.7523,
+      "step": 2947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.832290106074958e-06,
+      "loss": 0.7356,
+      "step": 2948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.819428852160759e-06,
+      "loss": 0.7551,
+      "step": 2949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.80657327805035e-06,
+      "loss": 0.8359,
+      "step": 2950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.793723390915904e-06,
+      "loss": 0.8453,
+      "step": 2951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.780879197926472e-06,
+      "loss": 0.5877,
+      "step": 2952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.768040706247897e-06,
+      "loss": 0.6841,
+      "step": 2953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.755207923042875e-06,
+      "loss": 0.9945,
+      "step": 2954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.742380855470875e-06,
+      "loss": 0.7102,
+      "step": 2955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.729559510688194e-06,
+      "loss": 0.9543,
+      "step": 2956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.716743895847955e-06,
+      "loss": 0.6783,
+      "step": 2957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.703934018100066e-06,
+      "loss": 0.7283,
+      "step": 2958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.69112988459123e-06,
+      "loss": 0.973,
+      "step": 2959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.67833150246495e-06,
+      "loss": 1.0404,
+      "step": 2960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.665538878861538e-06,
+      "loss": 0.6945,
+      "step": 2961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.652752020918069e-06,
+      "loss": 0.5312,
+      "step": 2962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.639970935768407e-06,
+      "loss": 0.6348,
+      "step": 2963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.627195630543198e-06,
+      "loss": 0.7787,
+      "step": 2964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.614426112369876e-06,
+      "loss": 0.6114,
+      "step": 2965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.601662388372625e-06,
+      "loss": 0.5817,
+      "step": 2966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.588904465672409e-06,
+      "loss": 0.7035,
+      "step": 2967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.57615235138695e-06,
+      "loss": 0.7531,
+      "step": 2968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.563406052630721e-06,
+      "loss": 0.7342,
+      "step": 2969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.550665576514984e-06,
+      "loss": 0.5203,
+      "step": 2970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.537930930147716e-06,
+      "loss": 0.9159,
+      "step": 2971
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.525202120633659e-06,
+      "loss": 0.6888,
+      "step": 2972
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.512479155074284e-06,
+      "loss": 0.7013,
+      "step": 2973
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.499762040567828e-06,
+      "loss": 0.6934,
+      "step": 2974
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.487050784209241e-06,
+      "loss": 1.0293,
+      "step": 2975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.474345393090211e-06,
+      "loss": 0.6756,
+      "step": 2976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.461645874299152e-06,
+      "loss": 0.6995,
+      "step": 2977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.44895223492121e-06,
+      "loss": 0.723,
+      "step": 2978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.436264482038256e-06,
+      "loss": 0.7283,
+      "step": 2979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.423582622728836e-06,
+      "loss": 0.6745,
+      "step": 2980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.41090666406826e-06,
+      "loss": 1.0363,
+      "step": 2981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.39823661312852e-06,
+      "loss": 0.792,
+      "step": 2982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.385572476978316e-06,
+      "loss": 0.6184,
+      "step": 2983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.372914262683038e-06,
+      "loss": 0.6641,
+      "step": 2984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.360261977304796e-06,
+      "loss": 0.611,
+      "step": 2985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.347615627902377e-06,
+      "loss": 0.7597,
+      "step": 2986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.334975221531255e-06,
+      "loss": 0.6121,
+      "step": 2987
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.32234076524358e-06,
+      "loss": 0.7684,
+      "step": 2988
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.309712266088214e-06,
+      "loss": 0.5917,
+      "step": 2989
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.297089731110662e-06,
+      "loss": 0.6057,
+      "step": 2990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.28447316735312e-06,
+      "loss": 0.8443,
+      "step": 2991
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.271862581854443e-06,
+      "loss": 0.6742,
+      "step": 2992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.259257981650149e-06,
+      "loss": 0.6831,
+      "step": 2993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.24665937377244e-06,
+      "loss": 0.7495,
+      "step": 2994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.234066765250145e-06,
+      "loss": 0.6334,
+      "step": 2995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.221480163108762e-06,
+      "loss": 0.6345,
+      "step": 2996
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.208899574370424e-06,
+      "loss": 0.7481,
+      "step": 2997
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.196325006053938e-06,
+      "loss": 0.6849,
+      "step": 2998
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.183756465174724e-06,
+      "loss": 0.6214,
+      "step": 2999
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.171193958744848e-06,
+      "loss": 0.9502,
+      "step": 3000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.158637493773015e-06,
+      "loss": 0.8132,
+      "step": 3001
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.146087077264541e-06,
+      "loss": 0.6951,
+      "step": 3002
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.133542716221407e-06,
+      "loss": 0.6889,
+      "step": 3003
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.121004417642158e-06,
+      "loss": 0.6607,
+      "step": 3004
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.108472188522013e-06,
+      "loss": 0.6747,
+      "step": 3005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.095946035852765e-06,
+      "loss": 0.5981,
+      "step": 3006
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.083425966622854e-06,
+      "loss": 0.6016,
+      "step": 3007
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.070911987817275e-06,
+      "loss": 0.6489,
+      "step": 3008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.058404106417676e-06,
+      "loss": 0.7993,
+      "step": 3009
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.045902329402265e-06,
+      "loss": 0.6157,
+      "step": 3010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.033406663745887e-06,
+      "loss": 0.584,
+      "step": 3011
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.02091711641992e-06,
+      "loss": 0.5948,
+      "step": 3012
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.008433694392383e-06,
+      "loss": 0.9796,
+      "step": 3013
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.995956404627845e-06,
+      "loss": 0.7949,
+      "step": 3014
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.983485254087465e-06,
+      "loss": 1.0707,
+      "step": 3015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.971020249728975e-06,
+      "loss": 0.7881,
+      "step": 3016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.958561398506671e-06,
+      "loss": 0.7426,
+      "step": 3017
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.946108707371437e-06,
+      "loss": 0.8773,
+      "step": 3018
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.9336621832707e-06,
+      "loss": 0.6931,
+      "step": 3019
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.921221833148455e-06,
+      "loss": 0.5352,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.908787663945237e-06,
+      "loss": 0.6153,
+      "step": 3021
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.896359682598168e-06,
+      "loss": 0.6107,
+      "step": 3022
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.883937896040884e-06,
+      "loss": 0.5924,
+      "step": 3023
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.871522311203579e-06,
+      "loss": 0.7123,
+      "step": 3024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.859112935012983e-06,
+      "loss": 0.6635,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.84670977439236e-06,
+      "loss": 0.7104,
+      "step": 3026
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.834312836261521e-06,
+      "loss": 0.9216,
+      "step": 3027
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.821922127536786e-06,
+      "loss": 1.039,
+      "step": 3028
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.809537655131015e-06,
+      "loss": 0.7916,
+      "step": 3029
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.797159425953568e-06,
+      "loss": 0.7515,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.784787446910355e-06,
+      "loss": 1.1241,
+      "step": 3031
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.772421724903768e-06,
+      "loss": 0.6052,
+      "step": 3032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.760062266832724e-06,
+      "loss": 0.6941,
+      "step": 3033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.747709079592632e-06,
+      "loss": 0.6901,
+      "step": 3034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.735362170075427e-06,
+      "loss": 0.616,
+      "step": 3035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.723021545169528e-06,
+      "loss": 0.7262,
+      "step": 3036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.710687211759821e-06,
+      "loss": 0.9349,
+      "step": 3037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.698359176727731e-06,
+      "loss": 0.6705,
+      "step": 3038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.68603744695113e-06,
+      "loss": 1.0015,
+      "step": 3039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.673722029304407e-06,
+      "loss": 0.623,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.661412930658387e-06,
+      "loss": 0.7426,
+      "step": 3041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.64911015788041e-06,
+      "loss": 0.5339,
+      "step": 3042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.636813717834253e-06,
+      "loss": 0.7463,
+      "step": 3043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.6245236173802e-06,
+      "loss": 0.6672,
+      "step": 3044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.612239863374947e-06,
+      "loss": 0.7208,
+      "step": 3045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.599962462671697e-06,
+      "loss": 0.6347,
+      "step": 3046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.587691422120073e-06,
+      "loss": 0.7994,
+      "step": 3047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.575426748566184e-06,
+      "loss": 1.0501,
+      "step": 3048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.563168448852551e-06,
+      "loss": 0.6915,
+      "step": 3049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.55091652981815e-06,
+      "loss": 0.5816,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.538670998298422e-06,
+      "loss": 0.747,
+      "step": 3051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.526431861125216e-06,
+      "loss": 0.626,
+      "step": 3052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.514199125126818e-06,
+      "loss": 0.6499,
+      "step": 3053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.501972797127947e-06,
+      "loss": 0.6195,
+      "step": 3054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.489752883949758e-06,
+      "loss": 0.8651,
+      "step": 3055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.477539392409808e-06,
+      "loss": 0.7407,
+      "step": 3056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.465332329322084e-06,
+      "loss": 0.6607,
+      "step": 3057
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.453131701496971e-06,
+      "loss": 1.0357,
+      "step": 3058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.440937515741293e-06,
+      "loss": 0.968,
+      "step": 3059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.428749778858249e-06,
+      "loss": 0.819,
+      "step": 3060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.416568497647456e-06,
+      "loss": 0.7022,
+      "step": 3061
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.40439367890493e-06,
+      "loss": 0.7187,
+      "step": 3062
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.392225329423066e-06,
+      "loss": 0.6909,
+      "step": 3063
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.380063455990677e-06,
+      "loss": 0.7702,
+      "step": 3064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.36790806539294e-06,
+      "loss": 1.0056,
+      "step": 3065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.355759164411426e-06,
+      "loss": 0.7203,
+      "step": 3066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.34361675982407e-06,
+      "loss": 0.7586,
+      "step": 3067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.331480858405212e-06,
+      "loss": 0.8378,
+      "step": 3068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.31935146692554e-06,
+      "loss": 0.7325,
+      "step": 3069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.307228592152116e-06,
+      "loss": 0.7405,
+      "step": 3070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.295112240848366e-06,
+      "loss": 0.961,
+      "step": 3071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.283002419774073e-06,
+      "loss": 0.5955,
+      "step": 3072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.2708991356854e-06,
+      "loss": 0.6662,
+      "step": 3073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.258802395334817e-06,
+      "loss": 0.9811,
+      "step": 3074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.24671220547119e-06,
+      "loss": 0.9236,
+      "step": 3075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.2346285728397e-06,
+      "loss": 0.5772,
+      "step": 3076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.222551504181899e-06,
+      "loss": 0.8213,
+      "step": 3077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.210481006235633e-06,
+      "loss": 0.8436,
+      "step": 3078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.198417085735126e-06,
+      "loss": 0.645,
+      "step": 3079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.186359749410902e-06,
+      "loss": 0.7258,
+      "step": 3080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.174309003989842e-06,
+      "loss": 0.7582,
+      "step": 3081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.162264856195106e-06,
+      "loss": 0.9561,
+      "step": 3082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.15022731274622e-06,
+      "loss": 0.8072,
+      "step": 3083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.138196380358997e-06,
+      "loss": 0.6759,
+      "step": 3084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.126172065745566e-06,
+      "loss": 0.7082,
+      "step": 3085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.114154375614369e-06,
+      "loss": 0.961,
+      "step": 3086
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.102143316670143e-06,
+      "loss": 0.5586,
+      "step": 3087
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.090138895613946e-06,
+      "loss": 0.7053,
+      "step": 3088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.078141119143113e-06,
+      "loss": 0.6535,
+      "step": 3089
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.066149993951274e-06,
+      "loss": 0.6197,
+      "step": 3090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.05416552672835e-06,
+      "loss": 0.6493,
+      "step": 3091
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.042187724160559e-06,
+      "loss": 0.9645,
+      "step": 3092
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.030216592930388e-06,
+      "loss": 0.8334,
+      "step": 3093
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.018252139716607e-06,
+      "loss": 0.7352,
+      "step": 3094
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.006294371194253e-06,
+      "loss": 1.0418,
+      "step": 3095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.994343294034635e-06,
+      "loss": 0.681,
+      "step": 3096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.982398914905348e-06,
+      "loss": 0.9879,
+      "step": 3097
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.970461240470224e-06,
+      "loss": 0.6909,
+      "step": 3098
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.95853027738937e-06,
+      "loss": 0.6065,
+      "step": 3099
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.946606032319138e-06,
+      "loss": 0.7906,
+      "step": 3100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.93468851191215e-06,
+      "loss": 0.7268,
+      "step": 3101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.922777722817256e-06,
+      "loss": 0.7594,
+      "step": 3102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.910873671679562e-06,
+      "loss": 0.638,
+      "step": 3103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.898976365140402e-06,
+      "loss": 0.6973,
+      "step": 3104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.887085809837376e-06,
+      "loss": 0.9858,
+      "step": 3105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.875202012404295e-06,
+      "loss": 0.5697,
+      "step": 3106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.86332497947118e-06,
+      "loss": 0.9863,
+      "step": 3107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.851454717664328e-06,
+      "loss": 0.7795,
+      "step": 3108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.839591233606214e-06,
+      "loss": 0.9737,
+      "step": 3109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.827734533915573e-06,
+      "loss": 0.5756,
+      "step": 3110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.815884625207297e-06,
+      "loss": 0.5672,
+      "step": 3111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.80404151409255e-06,
+      "loss": 0.8836,
+      "step": 3112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.792205207178665e-06,
+      "loss": 0.6245,
+      "step": 3113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.780375711069208e-06,
+      "loss": 1.0019,
+      "step": 3114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.7685530323639e-06,
+      "loss": 0.862,
+      "step": 3115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.756737177658709e-06,
+      "loss": 0.6848,
+      "step": 3116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.744928153545768e-06,
+      "loss": 0.646,
+      "step": 3117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.733125966613402e-06,
+      "loss": 0.9633,
+      "step": 3118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.721330623446126e-06,
+      "loss": 0.6587,
+      "step": 3119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.70954213062463e-06,
+      "loss": 0.6418,
+      "step": 3120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.697760494725797e-06,
+      "loss": 0.5973,
+      "step": 3121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.685985722322668e-06,
+      "loss": 0.6042,
+      "step": 3122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.674217819984466e-06,
+      "loss": 0.6255,
+      "step": 3123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.662456794276567e-06,
+      "loss": 0.9707,
+      "step": 3124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.650702651760533e-06,
+      "loss": 0.9264,
+      "step": 3125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.638955398994068e-06,
+      "loss": 0.9228,
+      "step": 3126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.627215042531035e-06,
+      "loss": 0.7797,
+      "step": 3127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.61548158892145e-06,
+      "loss": 0.7028,
+      "step": 3128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.603755044711485e-06,
+      "loss": 0.8639,
+      "step": 3129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.592035416443451e-06,
+      "loss": 0.6235,
+      "step": 3130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5803227106558005e-06,
+      "loss": 0.9349,
+      "step": 3131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.568616933883122e-06,
+      "loss": 0.6251,
+      "step": 3132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.556918092656136e-06,
+      "loss": 0.6267,
+      "step": 3133
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5452261935017135e-06,
+      "loss": 0.925,
+      "step": 3134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5335412429428276e-06,
+      "loss": 0.8591,
+      "step": 3135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.521863247498588e-06,
+      "loss": 0.8349,
+      "step": 3136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.510192213684211e-06,
+      "loss": 0.6289,
+      "step": 3137
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.498528148011055e-06,
+      "loss": 0.8365,
+      "step": 3138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.486871056986567e-06,
+      "loss": 0.6186,
+      "step": 3139
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.475220947114308e-06,
+      "loss": 0.7299,
+      "step": 3140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.4635778248939484e-06,
+      "loss": 0.8944,
+      "step": 3141
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.451941696821259e-06,
+      "loss": 0.7322,
+      "step": 3142
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.440312569388106e-06,
+      "loss": 0.8592,
+      "step": 3143
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.428690449082448e-06,
+      "loss": 0.7467,
+      "step": 3144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.417075342388347e-06,
+      "loss": 1.0641,
+      "step": 3145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.405467255785941e-06,
+      "loss": 0.914,
+      "step": 3146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.393866195751451e-06,
+      "loss": 0.8879,
+      "step": 3147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3822721687571715e-06,
+      "loss": 0.6651,
+      "step": 3148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.370685181271499e-06,
+      "loss": 0.8698,
+      "step": 3149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.359105239758879e-06,
+      "loss": 0.6291,
+      "step": 3150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.347532350679831e-06,
+      "loss": 0.7694,
+      "step": 3151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.335966520490942e-06,
+      "loss": 0.7267,
+      "step": 3152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3244077556448555e-06,
+      "loss": 0.5634,
+      "step": 3153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.312856062590288e-06,
+      "loss": 1.11,
+      "step": 3154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.301311447771997e-06,
+      "loss": 0.5807,
+      "step": 3155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.289773917630792e-06,
+      "loss": 0.7522,
+      "step": 3156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.278243478603528e-06,
+      "loss": 0.5563,
+      "step": 3157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.26672013712312e-06,
+      "loss": 0.6226,
+      "step": 3158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.255203899618508e-06,
+      "loss": 0.7211,
+      "step": 3159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.243694772514669e-06,
+      "loss": 0.627,
+      "step": 3160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.232192762232613e-06,
+      "loss": 0.5714,
+      "step": 3161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.220697875189393e-06,
+      "loss": 0.8383,
+      "step": 3162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.209210117798071e-06,
+      "loss": 0.6118,
+      "step": 3163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.197729496467743e-06,
+      "loss": 0.7393,
+      "step": 3164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.186256017603512e-06,
+      "loss": 0.7838,
+      "step": 3165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1747896876065026e-06,
+      "loss": 0.7796,
+      "step": 3166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.163330512873867e-06,
+      "loss": 0.6166,
+      "step": 3167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.151878499798726e-06,
+      "loss": 0.6865,
+      "step": 3168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.140433654770245e-06,
+      "loss": 0.7256,
+      "step": 3169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.128995984173566e-06,
+      "loss": 0.5544,
+      "step": 3170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.117565494389851e-06,
+      "loss": 0.9617,
+      "step": 3171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.106142191796217e-06,
+      "loss": 0.6422,
+      "step": 3172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.094726082765817e-06,
+      "loss": 0.7705,
+      "step": 3173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.083317173667754e-06,
+      "loss": 0.7334,
+      "step": 3174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.071915470867148e-06,
+      "loss": 0.6145,
+      "step": 3175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.060520980725061e-06,
+      "loss": 1.0532,
+      "step": 3176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0491337095985475e-06,
+      "loss": 0.7623,
+      "step": 3177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.037753663840656e-06,
+      "loss": 0.7794,
+      "step": 3178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.026380849800371e-06,
+      "loss": 0.6489,
+      "step": 3179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.015015273822661e-06,
+      "loss": 0.7341,
+      "step": 3180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.00365694224844e-06,
+      "loss": 0.759,
+      "step": 3181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.99230586141461e-06,
+      "loss": 0.7451,
+      "step": 3182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.9809620376539975e-06,
+      "loss": 0.7288,
+      "step": 3183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.969625477295396e-06,
+      "loss": 0.5883,
+      "step": 3184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.958296186663535e-06,
+      "loss": 0.8363,
+      "step": 3185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.946974172079111e-06,
+      "loss": 0.5623,
+      "step": 3186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.9356594398587375e-06,
+      "loss": 0.7996,
+      "step": 3187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.924351996314971e-06,
+      "loss": 0.7966,
+      "step": 3188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.913051847756311e-06,
+      "loss": 0.6021,
+      "step": 3189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.901759000487167e-06,
+      "loss": 0.9222,
+      "step": 3190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.890473460807907e-06,
+      "loss": 0.8566,
+      "step": 3191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.879195235014795e-06,
+      "loss": 0.7198,
+      "step": 3192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.867924329400024e-06,
+      "loss": 0.7552,
+      "step": 3193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.856660750251693e-06,
+      "loss": 0.8746,
+      "step": 3194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.845404503853836e-06,
+      "loss": 0.7291,
+      "step": 3195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.834155596486381e-06,
+      "loss": 1.1551,
+      "step": 3196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.822914034425159e-06,
+      "loss": 0.5933,
+      "step": 3197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8116798239418995e-06,
+      "loss": 0.5337,
+      "step": 3198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.800452971304254e-06,
+      "loss": 0.6264,
+      "step": 3199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.789233482775752e-06,
+      "loss": 0.7356,
+      "step": 3200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.7780213646157965e-06,
+      "loss": 0.5988,
+      "step": 3201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.766816623079717e-06,
+      "loss": 0.7454,
+      "step": 3202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.755619264418696e-06,
+      "loss": 0.5898,
+      "step": 3203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.744429294879828e-06,
+      "loss": 0.6759,
+      "step": 3204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.733246720706041e-06,
+      "loss": 0.693,
+      "step": 3205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.722071548136184e-06,
+      "loss": 0.5542,
+      "step": 3206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.710903783404936e-06,
+      "loss": 0.6007,
+      "step": 3207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.699743432742889e-06,
+      "loss": 0.762,
+      "step": 3208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6885905023764395e-06,
+      "loss": 0.7064,
+      "step": 3209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6774449985279e-06,
+      "loss": 0.7231,
+      "step": 3210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.666306927415409e-06,
+      "loss": 0.7667,
+      "step": 3211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.655176295252965e-06,
+      "loss": 0.8904,
+      "step": 3212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.644053108250417e-06,
+      "loss": 0.7411,
+      "step": 3213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.63293737261345e-06,
+      "loss": 0.6609,
+      "step": 3214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.621829094543621e-06,
+      "loss": 0.9738,
+      "step": 3215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6107282802382985e-06,
+      "loss": 0.8458,
+      "step": 3216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.599634935890693e-06,
+      "loss": 0.7683,
+      "step": 3217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.588549067689846e-06,
+      "loss": 0.8051,
+      "step": 3218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.577470681820646e-06,
+      "loss": 0.7013,
+      "step": 3219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.5663997844637814e-06,
+      "loss": 0.7958,
+      "step": 3220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.555336381795783e-06,
+      "loss": 0.6886,
+      "step": 3221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.544280479988985e-06,
+      "loss": 0.789,
+      "step": 3222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.533232085211539e-06,
+      "loss": 0.8806,
+      "step": 3223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.522191203627426e-06,
+      "loss": 0.6562,
+      "step": 3224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.511157841396416e-06,
+      "loss": 0.7423,
+      "step": 3225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.500132004674091e-06,
+      "loss": 0.7938,
+      "step": 3226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.489113699611826e-06,
+      "loss": 0.9111,
+      "step": 3227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.478102932356816e-06,
+      "loss": 1.061,
+      "step": 3228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.467099709052027e-06,
+      "loss": 0.567,
+      "step": 3229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.4561040358362284e-06,
+      "loss": 0.6253,
+      "step": 3230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.445115918843967e-06,
+      "loss": 0.8614,
+      "step": 3231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.4341353642055934e-06,
+      "loss": 0.6522,
+      "step": 3232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.423162378047223e-06,
+      "loss": 0.6397,
+      "step": 3233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.412196966490751e-06,
+      "loss": 0.8109,
+      "step": 3234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.401239135653849e-06,
+      "loss": 0.7526,
+      "step": 3235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.390288891649951e-06,
+      "loss": 0.5311,
+      "step": 3236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.379346240588289e-06,
+      "loss": 0.9136,
+      "step": 3237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.368411188573807e-06,
+      "loss": 0.7329,
+      "step": 3238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.357483741707258e-06,
+      "loss": 0.8686,
+      "step": 3239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.346563906085119e-06,
+      "loss": 0.5285,
+      "step": 3240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.335651687799656e-06,
+      "loss": 0.9209,
+      "step": 3241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.324747092938832e-06,
+      "loss": 0.5749,
+      "step": 3242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.313850127586414e-06,
+      "loss": 0.6037,
+      "step": 3243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.302960797821871e-06,
+      "loss": 0.8014,
+      "step": 3244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.292079109720446e-06,
+      "loss": 0.7709,
+      "step": 3245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.281205069353082e-06,
+      "loss": 0.9608,
+      "step": 3246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.270338682786472e-06,
+      "loss": 0.6458,
+      "step": 3247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.259479956083056e-06,
+      "loss": 0.7396,
+      "step": 3248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.248628895300976e-06,
+      "loss": 1.0055,
+      "step": 3249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.237785506494107e-06,
+      "loss": 0.6218,
+      "step": 3250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.226949795712039e-06,
+      "loss": 0.573,
+      "step": 3251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.21612176900009e-06,
+      "loss": 0.8438,
+      "step": 3252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.20530143239928e-06,
+      "loss": 0.6376,
+      "step": 3253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.194488791946343e-06,
+      "loss": 0.6573,
+      "step": 3254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1836838536737075e-06,
+      "loss": 0.5094,
+      "step": 3255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1728866236095335e-06,
+      "loss": 0.6369,
+      "step": 3256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.162097107777651e-06,
+      "loss": 0.7084,
+      "step": 3257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.151315312197601e-06,
+      "loss": 0.7554,
+      "step": 3258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.140541242884614e-06,
+      "loss": 0.63,
+      "step": 3259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.129774905849604e-06,
+      "loss": 0.9561,
+      "step": 3260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.11901630709919e-06,
+      "loss": 0.7689,
+      "step": 3261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.108265452635653e-06,
+      "loss": 0.6389,
+      "step": 3262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.097522348456961e-06,
+      "loss": 0.7216,
+      "step": 3263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.086787000556753e-06,
+      "loss": 0.689,
+      "step": 3264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.0760594149243625e-06,
+      "loss": 0.6989,
+      "step": 3265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.065339597544764e-06,
+      "loss": 0.7585,
+      "step": 3266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.054627554398611e-06,
+      "loss": 1.138,
+      "step": 3267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.043923291462215e-06,
+      "loss": 0.6983,
+      "step": 3268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.033226814707562e-06,
+      "loss": 1.0001,
+      "step": 3269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.022538130102285e-06,
+      "loss": 0.6587,
+      "step": 3270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.011857243609644e-06,
+      "loss": 0.6132,
+      "step": 3271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.001184161188596e-06,
+      "loss": 0.6531,
+      "step": 3272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.990518888793702e-06,
+      "loss": 0.7982,
+      "step": 3273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.979861432375205e-06,
+      "loss": 0.8267,
+      "step": 3274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.96921179787894e-06,
+      "loss": 0.6834,
+      "step": 3275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.958569991246426e-06,
+      "loss": 1.0099,
+      "step": 3276
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.947936018414781e-06,
+      "loss": 0.6038,
+      "step": 3277
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.937309885316771e-06,
+      "loss": 0.6851,
+      "step": 3278
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.926691597880771e-06,
+      "loss": 0.8691,
+      "step": 3279
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.916081162030802e-06,
+      "loss": 0.8358,
+      "step": 3280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.905478583686488e-06,
+      "loss": 0.5873,
+      "step": 3281
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.89488386876307e-06,
+      "loss": 0.6392,
+      "step": 3282
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.884297023171408e-06,
+      "loss": 0.6806,
+      "step": 3283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.873718052817963e-06,
+      "loss": 0.716,
+      "step": 3284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.863146963604818e-06,
+      "loss": 0.6695,
+      "step": 3285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.852583761429644e-06,
+      "loss": 0.7805,
+      "step": 3286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8420284521857175e-06,
+      "loss": 0.6162,
+      "step": 3287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.831481041761904e-06,
+      "loss": 0.7137,
+      "step": 3288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8209415360426815e-06,
+      "loss": 0.5886,
+      "step": 3289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.810409940908102e-06,
+      "loss": 0.665,
+      "step": 3290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.799886262233803e-06,
+      "loss": 0.6102,
+      "step": 3291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.789370505891014e-06,
+      "loss": 0.7787,
+      "step": 3292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.778862677746533e-06,
+      "loss": 0.6443,
+      "step": 3293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.768362783662756e-06,
+      "loss": 0.5592,
+      "step": 3294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.757870829497631e-06,
+      "loss": 1.0721,
+      "step": 3295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.747386821104685e-06,
+      "loss": 0.7351,
+      "step": 3296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.736910764333008e-06,
+      "loss": 0.6655,
+      "step": 3297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.726442665027269e-06,
+      "loss": 0.6756,
+      "step": 3298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.715982529027668e-06,
+      "loss": 0.6006,
+      "step": 3299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.705530362169993e-06,
+      "loss": 0.6773,
+      "step": 3300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.695086170285562e-06,
+      "loss": 0.7029,
+      "step": 3301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.684649959201274e-06,
+      "loss": 0.6686,
+      "step": 3302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.674221734739538e-06,
+      "loss": 0.8099,
+      "step": 3303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6638015027183225e-06,
+      "loss": 0.7845,
+      "step": 3304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.653389268951152e-06,
+      "loss": 0.9362,
+      "step": 3305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.642985039247073e-06,
+      "loss": 0.9927,
+      "step": 3306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.632588819410667e-06,
+      "loss": 0.6791,
+      "step": 3307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.622200615242044e-06,
+      "loss": 0.6488,
+      "step": 3308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.611820432536861e-06,
+      "loss": 0.6899,
+      "step": 3309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.601448277086279e-06,
+      "loss": 0.6537,
+      "step": 3310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.591084154676989e-06,
+      "loss": 0.95,
+      "step": 3311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.58072807109119e-06,
+      "loss": 1.1233,
+      "step": 3312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.570380032106622e-06,
+      "loss": 0.63,
+      "step": 3313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.560040043496511e-06,
+      "loss": 0.7625,
+      "step": 3314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.549708111029601e-06,
+      "loss": 0.6047,
+      "step": 3315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.5393842404701425e-06,
+      "loss": 0.7936,
+      "step": 3316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.5290684375778805e-06,
+      "loss": 0.5165,
+      "step": 3317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.518760708108075e-06,
+      "loss": 0.7266,
+      "step": 3318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.50846105781147e-06,
+      "loss": 0.6484,
+      "step": 3319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4981694924343e-06,
+      "loss": 0.7994,
+      "step": 3320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4878860177182916e-06,
+      "loss": 0.6861,
+      "step": 3321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.477610639400668e-06,
+      "loss": 0.737,
+      "step": 3322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.467343363214124e-06,
+      "loss": 0.888,
+      "step": 3323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.457084194886831e-06,
+      "loss": 0.6431,
+      "step": 3324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.44683314014244e-06,
+      "loss": 0.6697,
+      "step": 3325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.436590204700092e-06,
+      "loss": 0.6619,
+      "step": 3326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.426355394274381e-06,
+      "loss": 0.7242,
+      "step": 3327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.416128714575355e-06,
+      "loss": 0.6129,
+      "step": 3328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.405910171308557e-06,
+      "loss": 0.6797,
+      "step": 3329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3956997701749696e-06,
+      "loss": 0.7476,
+      "step": 3330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.38549751687105e-06,
+      "loss": 0.7242,
+      "step": 3331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.37530341708868e-06,
+      "loss": 0.7962,
+      "step": 3332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.365117476515225e-06,
+      "loss": 0.7198,
+      "step": 3333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.354939700833475e-06,
+      "loss": 0.5945,
+      "step": 3334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.344770095721692e-06,
+      "loss": 0.7427,
+      "step": 3335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.334608666853536e-06,
+      "loss": 0.6996,
+      "step": 3336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.324455419898151e-06,
+      "loss": 0.6518,
+      "step": 3337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.31431036052009e-06,
+      "loss": 0.9013,
+      "step": 3338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3041734943793435e-06,
+      "loss": 0.65,
+      "step": 3339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.294044827131335e-06,
+      "loss": 0.6626,
+      "step": 3340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2839243644269e-06,
+      "loss": 0.6485,
+      "step": 3341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.273812111912325e-06,
+      "loss": 0.79,
+      "step": 3342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.263708075229286e-06,
+      "loss": 0.6866,
+      "step": 3343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.253612260014893e-06,
+      "loss": 1.0718,
+      "step": 3344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2435246719016565e-06,
+      "loss": 0.666,
+      "step": 3345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.233445316517514e-06,
+      "loss": 0.6515,
+      "step": 3346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.223374199485798e-06,
+      "loss": 0.9424,
+      "step": 3347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.213311326425246e-06,
+      "loss": 0.7662,
+      "step": 3348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.203256702949988e-06,
+      "loss": 0.9362,
+      "step": 3349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.193210334669578e-06,
+      "loss": 0.8563,
+      "step": 3350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.18317222718894e-06,
+      "loss": 0.7094,
+      "step": 3351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.173142386108392e-06,
+      "loss": 0.7751,
+      "step": 3352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.163120817023648e-06,
+      "loss": 0.5928,
+      "step": 3353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.153107525525798e-06,
+      "loss": 0.6751,
+      "step": 3354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.143102517201333e-06,
+      "loss": 0.7414,
+      "step": 3355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.133105797632096e-06,
+      "loss": 0.8889,
+      "step": 3356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.123117372395327e-06,
+      "loss": 1.0613,
+      "step": 3357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.11313724706362e-06,
+      "loss": 0.9561,
+      "step": 3358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.103165427204964e-06,
+      "loss": 0.5674,
+      "step": 3359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.093201918382686e-06,
+      "loss": 0.7153,
+      "step": 3360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.083246726155499e-06,
+      "loss": 1.1701,
+      "step": 3361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.073299856077461e-06,
+      "loss": 0.716,
+      "step": 3362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.063361313697985e-06,
+      "loss": 0.6355,
+      "step": 3363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.053431104561867e-06,
+      "loss": 0.5751,
+      "step": 3364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.043509234209207e-06,
+      "loss": 0.6703,
+      "step": 3365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.033595708175496e-06,
+      "loss": 0.7498,
+      "step": 3366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0236905319915405e-06,
+      "loss": 0.9779,
+      "step": 3367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.013793711183515e-06,
+      "loss": 0.6186,
+      "step": 3368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0039052512728955e-06,
+      "loss": 0.5963,
+      "step": 3369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.994025157776534e-06,
+      "loss": 0.7343,
+      "step": 3370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.984153436206583e-06,
+      "loss": 0.7377,
+      "step": 3371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.97429009207055e-06,
+      "loss": 0.5858,
+      "step": 3372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.964435130871248e-06,
+      "loss": 0.6515,
+      "step": 3373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.954588558106812e-06,
+      "loss": 0.6631,
+      "step": 3374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.944750379270724e-06,
+      "loss": 0.9339,
+      "step": 3375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.934920599851756e-06,
+      "loss": 0.6735,
+      "step": 3376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.925099225334003e-06,
+      "loss": 1.0401,
+      "step": 3377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.915286261196865e-06,
+      "loss": 0.6633,
+      "step": 3378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.905481712915068e-06,
+      "loss": 0.6243,
+      "step": 3379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.895685585958623e-06,
+      "loss": 0.5896,
+      "step": 3380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.885897885792852e-06,
+      "loss": 0.7936,
+      "step": 3381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.876118617878367e-06,
+      "loss": 0.8822,
+      "step": 3382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.866347787671093e-06,
+      "loss": 0.7774,
+      "step": 3383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.856585400622231e-06,
+      "loss": 0.6705,
+      "step": 3384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.846831462178278e-06,
+      "loss": 0.6541,
+      "step": 3385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.837085977781018e-06,
+      "loss": 0.634,
+      "step": 3386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8273489528675075e-06,
+      "loss": 0.6971,
+      "step": 3387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.81762039287011e-06,
+      "loss": 0.7889,
+      "step": 3388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.80790030321644e-06,
+      "loss": 0.5801,
+      "step": 3389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.798188689329395e-06,
+      "loss": 0.7856,
+      "step": 3390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.78848555662714e-06,
+      "loss": 0.8035,
+      "step": 3391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.778790910523128e-06,
+      "loss": 0.6453,
+      "step": 3392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7691047564260505e-06,
+      "loss": 0.9279,
+      "step": 3393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.759427099739877e-06,
+      "loss": 1.0039,
+      "step": 3394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.749757945863825e-06,
+      "loss": 0.7392,
+      "step": 3395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.740097300192385e-06,
+      "loss": 1.0129,
+      "step": 3396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7304451681152964e-06,
+      "loss": 1.0881,
+      "step": 3397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.720801555017522e-06,
+      "loss": 0.6116,
+      "step": 3398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.711166466279311e-06,
+      "loss": 0.8534,
+      "step": 3399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.701539907276129e-06,
+      "loss": 0.7062,
+      "step": 3400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.691921883378707e-06,
+      "loss": 0.8156,
+      "step": 3401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.682312399952982e-06,
+      "loss": 0.9968,
+      "step": 3402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.67271146236016e-06,
+      "loss": 0.8197,
+      "step": 3403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.66311907595665e-06,
+      "loss": 0.6336,
+      "step": 3404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.653535246094123e-06,
+      "loss": 0.9061,
+      "step": 3405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.643959978119432e-06,
+      "loss": 0.6317,
+      "step": 3406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6343932773747e-06,
+      "loss": 1.2322,
+      "step": 3407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.62483514919724e-06,
+      "loss": 1.1264,
+      "step": 3408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.61528559891959e-06,
+      "loss": 0.6102,
+      "step": 3409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.605744631869504e-06,
+      "loss": 0.5765,
+      "step": 3410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.596212253369943e-06,
+      "loss": 0.6144,
+      "step": 3411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.586688468739087e-06,
+      "loss": 0.5979,
+      "step": 3412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.577173283290315e-06,
+      "loss": 0.5835,
+      "step": 3413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.567666702332203e-06,
+      "loss": 0.6872,
+      "step": 3414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.558168731168528e-06,
+      "loss": 0.7544,
+      "step": 3415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.548679375098277e-06,
+      "loss": 0.7175,
+      "step": 3416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.539198639415618e-06,
+      "loss": 0.6157,
+      "step": 3417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.529726529409908e-06,
+      "loss": 0.7208,
+      "step": 3418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.520263050365692e-06,
+      "loss": 0.7893,
+      "step": 3419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.510808207562718e-06,
+      "loss": 0.9989,
+      "step": 3420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.501362006275889e-06,
+      "loss": 0.6391,
+      "step": 3421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.491924451775307e-06,
+      "loss": 0.6703,
+      "step": 3422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.482495549326235e-06,
+      "loss": 0.5488,
+      "step": 3423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.473075304189116e-06,
+      "loss": 0.6336,
+      "step": 3424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.463663721619573e-06,
+      "loss": 0.6439,
+      "step": 3425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.45426080686838e-06,
+      "loss": 0.7167,
+      "step": 3426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.44486656518148e-06,
+      "loss": 0.9227,
+      "step": 3427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.435481001799975e-06,
+      "loss": 0.7687,
+      "step": 3428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.42610412196014e-06,
+      "loss": 0.6397,
+      "step": 3429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.416735930893388e-06,
+      "loss": 0.7853,
+      "step": 3430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4073764338262915e-06,
+      "loss": 0.8367,
+      "step": 3431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.398025635980574e-06,
+      "loss": 0.7611,
+      "step": 3432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3886835425731e-06,
+      "loss": 1.0435,
+      "step": 3433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.379350158815882e-06,
+      "loss": 0.5393,
+      "step": 3434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.37002548991607e-06,
+      "loss": 0.6118,
+      "step": 3435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.360709541075962e-06,
+      "loss": 1.016,
+      "step": 3436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.351402317492981e-06,
+      "loss": 0.6708,
+      "step": 3437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.342103824359682e-06,
+      "loss": 0.6129,
+      "step": 3438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3328140668637485e-06,
+      "loss": 0.7443,
+      "step": 3439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3235330501880045e-06,
+      "loss": 0.6082,
+      "step": 3440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.314260779510382e-06,
+      "loss": 0.7022,
+      "step": 3441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.304997260003935e-06,
+      "loss": 0.9688,
+      "step": 3442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.295742496836843e-06,
+      "loss": 0.7103,
+      "step": 3443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.286496495172387e-06,
+      "loss": 0.5981,
+      "step": 3444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.277259260168982e-06,
+      "loss": 0.6637,
+      "step": 3445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.268030796980131e-06,
+      "loss": 0.6208,
+      "step": 3446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.258811110754453e-06,
+      "loss": 0.7299,
+      "step": 3447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.24960020663566e-06,
+      "loss": 0.6126,
+      "step": 3448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.240398089762585e-06,
+      "loss": 0.6814,
+      "step": 3449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.231204765269141e-06,
+      "loss": 0.6905,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.222020238284341e-06,
+      "loss": 0.5227,
+      "step": 3451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.212844513932284e-06,
+      "loss": 0.7354,
+      "step": 3452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.203677597332172e-06,
+      "loss": 0.5992,
+      "step": 3453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.194519493598281e-06,
+      "loss": 0.5966,
+      "step": 3454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1853702078399736e-06,
+      "loss": 0.5515,
+      "step": 3455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.176229745161695e-06,
+      "loss": 0.6054,
+      "step": 3456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.167098110662957e-06,
+      "loss": 0.5531,
+      "step": 3457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.157975309438373e-06,
+      "loss": 0.7102,
+      "step": 3458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1488613465775905e-06,
+      "loss": 0.9969,
+      "step": 3459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.139756227165359e-06,
+      "loss": 0.7448,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.130659956281473e-06,
+      "loss": 1.0087,
+      "step": 3461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.121572539000813e-06,
+      "loss": 0.6852,
+      "step": 3462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.112493980393284e-06,
+      "loss": 0.8152,
+      "step": 3463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.103424285523887e-06,
+      "loss": 0.7818,
+      "step": 3464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.094363459452651e-06,
+      "loss": 0.8399,
+      "step": 3465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0853115072346815e-06,
+      "loss": 0.8996,
+      "step": 3466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.076268433920103e-06,
+      "loss": 1.0874,
+      "step": 3467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.067234244554101e-06,
+      "loss": 0.7357,
+      "step": 3468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.05820894417692e-06,
+      "loss": 0.8782,
+      "step": 3469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.049192537823823e-06,
+      "loss": 0.7325,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.040185030525119e-06,
+      "loss": 0.5537,
+      "step": 3471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.031186427306146e-06,
+      "loss": 0.6328,
+      "step": 3472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.022196733187296e-06,
+      "loss": 0.6054,
+      "step": 3473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.013215953183964e-06,
+      "loss": 0.6374,
+      "step": 3474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.004244092306586e-06,
+      "loss": 0.5826,
+      "step": 3475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.995281155560613e-06,
+      "loss": 0.7585,
+      "step": 3476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.986327147946534e-06,
+      "loss": 0.8602,
+      "step": 3477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.977382074459837e-06,
+      "loss": 0.6565,
+      "step": 3478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.968445940091039e-06,
+      "loss": 1.0233,
+      "step": 3479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9595187498256616e-06,
+      "loss": 0.945,
+      "step": 3480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.950600508644233e-06,
+      "loss": 0.7537,
+      "step": 3481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.94169122152231e-06,
+      "loss": 0.5607,
+      "step": 3482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.932790893430431e-06,
+      "loss": 1.0212,
+      "step": 3483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.923899529334143e-06,
+      "loss": 0.7529,
+      "step": 3484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.915017134193992e-06,
+      "loss": 0.7796,
+      "step": 3485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9061437129655285e-06,
+      "loss": 0.7297,
+      "step": 3486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8972792705992855e-06,
+      "loss": 0.6319,
+      "step": 3487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.888423812040789e-06,
+      "loss": 0.6065,
+      "step": 3488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.879577342230556e-06,
+      "loss": 0.731,
+      "step": 3489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8707398661040805e-06,
+      "loss": 0.5652,
+      "step": 3490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.861911388591864e-06,
+      "loss": 0.4938,
+      "step": 3491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.85309191461934e-06,
+      "loss": 0.5971,
+      "step": 3492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.844281449106972e-06,
+      "loss": 0.9782,
+      "step": 3493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8354799969701575e-06,
+      "loss": 0.4733,
+      "step": 3494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8266875631193e-06,
+      "loss": 0.5938,
+      "step": 3495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.817904152459731e-06,
+      "loss": 0.9688,
+      "step": 3496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8091297698917838e-06,
+      "loss": 0.6857,
+      "step": 3497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.800364420310734e-06,
+      "loss": 0.6638,
+      "step": 3498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.791608108606837e-06,
+      "loss": 0.8869,
+      "step": 3499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.782860839665272e-06,
+      "loss": 0.9838,
+      "step": 3500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.774122618366214e-06,
+      "loss": 0.6747,
+      "step": 3501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.765393449584762e-06,
+      "loss": 0.6552,
+      "step": 3502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7566733381909746e-06,
+      "loss": 0.8827,
+      "step": 3503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7479622890498558e-06,
+      "loss": 0.7511,
+      "step": 3504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7392603070213508e-06,
+      "loss": 0.7376,
+      "step": 3505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.730567396960356e-06,
+      "loss": 0.7036,
+      "step": 3506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7218835637166997e-06,
+      "loss": 0.5535,
+      "step": 3507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.713208812135141e-06,
+      "loss": 0.6809,
+      "step": 3508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.704543147055375e-06,
+      "loss": 0.5873,
+      "step": 3509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6958865733120397e-06,
+      "loss": 0.594,
+      "step": 3510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6872390957346872e-06,
+      "loss": 0.7474,
+      "step": 3511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.678600719147798e-06,
+      "loss": 0.8562,
+      "step": 3512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.669971448370775e-06,
+      "loss": 0.5184,
+      "step": 3513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.661351288217938e-06,
+      "loss": 0.7039,
+      "step": 3514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6527402434985405e-06,
+      "loss": 0.6215,
+      "step": 3515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644138319016728e-06,
+      "loss": 0.6637,
+      "step": 3516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6355455195715683e-06,
+      "loss": 0.9945,
+      "step": 3517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.626961849957036e-06,
+      "loss": 0.6932,
+      "step": 3518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6183873149620197e-06,
+      "loss": 0.6831,
+      "step": 3519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.609821919370302e-06,
+      "loss": 0.8456,
+      "step": 3520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6012656679605716e-06,
+      "loss": 0.6183,
+      "step": 3521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5927185655064078e-06,
+      "loss": 0.5633,
+      "step": 3522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5841806167763026e-06,
+      "loss": 0.5085,
+      "step": 3523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.575651826533628e-06,
+      "loss": 0.8756,
+      "step": 3524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5671321995366383e-06,
+      "loss": 0.5696,
+      "step": 3525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5586217405384993e-06,
+      "loss": 0.842,
+      "step": 3526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.550120454287238e-06,
+      "loss": 0.6809,
+      "step": 3527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.541628345525792e-06,
+      "loss": 0.6535,
+      "step": 3528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.533145418991939e-06,
+      "loss": 0.65,
+      "step": 3529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.524671679418372e-06,
+      "loss": 0.5958,
+      "step": 3530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5162071315326317e-06,
+      "loss": 0.7628,
+      "step": 3531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5077517800571604e-06,
+      "loss": 0.6605,
+      "step": 3532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4993056297092263e-06,
+      "loss": 0.7341,
+      "step": 3533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.49086868520101e-06,
+      "loss": 0.7491,
+      "step": 3534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4824409512395184e-06,
+      "loss": 0.9362,
+      "step": 3535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.474022432526658e-06,
+      "loss": 0.6707,
+      "step": 3536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4656131337591537e-06,
+      "loss": 0.8615,
+      "step": 3537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4572130596286058e-06,
+      "loss": 0.5885,
+      "step": 3538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.448822214821477e-06,
+      "loss": 1.0542,
+      "step": 3539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4404406040190684e-06,
+      "loss": 0.6144,
+      "step": 3540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.432068231897534e-06,
+      "loss": 1.0395,
+      "step": 3541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4237051031278633e-06,
+      "loss": 0.8149,
+      "step": 3542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.415351222375911e-06,
+      "loss": 0.7217,
+      "step": 3543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.407006594302351e-06,
+      "loss": 0.6213,
+      "step": 3544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.398671223562706e-06,
+      "loss": 0.7084,
+      "step": 3545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3903451148073252e-06,
+      "loss": 0.7604,
+      "step": 3546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3820282726814037e-06,
+      "loss": 0.6371,
+      "step": 3547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3737207018249584e-06,
+      "loss": 0.9076,
+      "step": 3548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.365422406872831e-06,
+      "loss": 0.795,
+      "step": 3549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3571333924546945e-06,
+      "loss": 0.6742,
+      "step": 3550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3488536631950352e-06,
+      "loss": 0.57,
+      "step": 3551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3405832237131743e-06,
+      "loss": 0.6244,
+      "step": 3552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.332322078623238e-06,
+      "loss": 0.6126,
+      "step": 3553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3240702325341713e-06,
+      "loss": 0.5819,
+      "step": 3554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.315827690049722e-06,
+      "loss": 0.6224,
+      "step": 3555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.307594455768468e-06,
+      "loss": 0.6907,
+      "step": 3556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2993705342837765e-06,
+      "loss": 0.7028,
+      "step": 3557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2911559301838226e-06,
+      "loss": 0.9887,
+      "step": 3558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2829506480515884e-06,
+      "loss": 0.6121,
+      "step": 3559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.274754692464843e-06,
+      "loss": 0.7258,
+      "step": 3560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2665680679961765e-06,
+      "loss": 0.7486,
+      "step": 3561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.258390779212941e-06,
+      "loss": 0.7232,
+      "step": 3562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2502228306773077e-06,
+      "loss": 0.6048,
+      "step": 3563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2420642269462153e-06,
+      "loss": 0.965,
+      "step": 3564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.233914972571419e-06,
+      "loss": 0.7053,
+      "step": 3565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.225775072099413e-06,
+      "loss": 0.548,
+      "step": 3566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2176445300715174e-06,
+      "loss": 0.7293,
+      "step": 3567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2095233510238e-06,
+      "loss": 0.7795,
+      "step": 3568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2014115394871316e-06,
+      "loss": 0.632,
+      "step": 3569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.193309099987123e-06,
+      "loss": 0.9863,
+      "step": 3570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.185216037044192e-06,
+      "loss": 1.1583,
+      "step": 3571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1771323551735e-06,
+      "loss": 0.7386,
+      "step": 3572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.169058058884984e-06,
+      "loss": 0.7729,
+      "step": 3573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1609931526833473e-06,
+      "loss": 0.8154,
+      "step": 3574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1529376410680434e-06,
+      "loss": 0.525,
+      "step": 3575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1448915285333025e-06,
+      "loss": 0.6984,
+      "step": 3576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1368548195680958e-06,
+      "loss": 0.6857,
+      "step": 3577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1288275186561525e-06,
+      "loss": 0.6225,
+      "step": 3578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1208096302759496e-06,
+      "loss": 0.7675,
+      "step": 3579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1128011589007266e-06,
+      "loss": 0.6119,
+      "step": 3580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.104802108998455e-06,
+      "loss": 1.0357,
+      "step": 3581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0968124850318506e-06,
+      "loss": 0.6096,
+      "step": 3582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.088832291458379e-06,
+      "loss": 0.6275,
+      "step": 3583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.080861532730233e-06,
+      "loss": 1.0428,
+      "step": 3584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0729002132943563e-06,
+      "loss": 0.6609,
+      "step": 3585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.064948337592415e-06,
+      "loss": 0.8169,
+      "step": 3586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0570059100608108e-06,
+      "loss": 0.6019,
+      "step": 3587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.049072935130668e-06,
+      "loss": 0.9519,
+      "step": 3588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.041149417227851e-06,
+      "loss": 0.7797,
+      "step": 3589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.033235360772935e-06,
+      "loss": 0.5875,
+      "step": 3590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.025330770181223e-06,
+      "loss": 0.7375,
+      "step": 3591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0174356498627278e-06,
+      "loss": 0.6244,
+      "step": 3592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0095500042222016e-06,
+      "loss": 0.679,
+      "step": 3593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0016738376590804e-06,
+      "loss": 0.6667,
+      "step": 3594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.993807154567525e-06,
+      "loss": 0.6552,
+      "step": 3595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9859499593364205e-06,
+      "loss": 0.768,
+      "step": 3596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9781022563493355e-06,
+      "loss": 0.7706,
+      "step": 3597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9702640499845548e-06,
+      "loss": 0.6171,
+      "step": 3598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9624353446150557e-06,
+      "loss": 0.6315,
+      "step": 3599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.954616144608533e-06,
+      "loss": 0.6182,
+      "step": 3600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.94680645432736e-06,
+      "loss": 0.9114,
+      "step": 3601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9390062781286133e-06,
+      "loss": 0.6703,
+      "step": 3602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9312156203640542e-06,
+      "loss": 0.9762,
+      "step": 3603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.923434485380148e-06,
+      "loss": 0.8852,
+      "step": 3604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.915662877518033e-06,
+      "loss": 0.986,
+      "step": 3605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.907900801113539e-06,
+      "loss": 0.635,
+      "step": 3606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9001482604971754e-06,
+      "loss": 0.6664,
+      "step": 3607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8924052599941267e-06,
+      "loss": 0.689,
+      "step": 3608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8846718039242726e-06,
+      "loss": 0.6405,
+      "step": 3609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8769478966021515e-06,
+      "loss": 0.8123,
+      "step": 3610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8692335423369776e-06,
+      "loss": 0.5863,
+      "step": 3611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8615287454326335e-06,
+      "loss": 1.1429,
+      "step": 3612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8538335101876825e-06,
+      "loss": 0.5923,
+      "step": 3613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8461478408953393e-06,
+      "loss": 0.5676,
+      "step": 3614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.838471741843485e-06,
+      "loss": 0.779,
+      "step": 3615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8308052173146607e-06,
+      "loss": 0.655,
+      "step": 3616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8231482715860756e-06,
+      "loss": 0.643,
+      "step": 3617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.815500908929587e-06,
+      "loss": 0.6435,
+      "step": 3618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8078631336116923e-06,
+      "loss": 0.5232,
+      "step": 3619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8002349498935676e-06,
+      "loss": 0.8552,
+      "step": 3620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7926163620310154e-06,
+      "loss": 0.6103,
+      "step": 3621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7850073742745063e-06,
+      "loss": 0.7293,
+      "step": 3622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.777407990869123e-06,
+      "loss": 0.7812,
+      "step": 3623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.769818216054625e-06,
+      "loss": 1.0039,
+      "step": 3624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7622380540653827e-06,
+      "loss": 0.8863,
+      "step": 3625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7546675091304285e-06,
+      "loss": 0.9285,
+      "step": 3626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7471065854734003e-06,
+      "loss": 0.7351,
+      "step": 3627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7395552873125985e-06,
+      "loss": 0.6683,
+      "step": 3628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7320136188609314e-06,
+      "loss": 1.0095,
+      "step": 3629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7244815843259464e-06,
+      "loss": 0.7096,
+      "step": 3630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.716959187909809e-06,
+      "loss": 0.579,
+      "step": 3631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7094464338093087e-06,
+      "loss": 0.6787,
+      "step": 3632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7019433262158657e-06,
+      "loss": 0.66,
+      "step": 3633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6944498693155053e-06,
+      "loss": 0.6331,
+      "step": 3634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6869660672888743e-06,
+      "loss": 0.9274,
+      "step": 3635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.679491924311226e-06,
+      "loss": 0.7433,
+      "step": 3636
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.672027444552443e-06,
+      "loss": 0.6517,
+      "step": 3637
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6645726321769982e-06,
+      "loss": 0.9336,
+      "step": 3638
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6571274913439803e-06,
+      "loss": 0.7092,
+      "step": 3639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6496920262070757e-06,
+      "loss": 0.7242,
+      "step": 3640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.642266240914573e-06,
+      "loss": 0.8224,
+      "step": 3641
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.634850139609375e-06,
+      "loss": 0.7544,
+      "step": 3642
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6274437264289663e-06,
+      "loss": 0.9213,
+      "step": 3643
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6200470055054263e-06,
+      "loss": 0.7723,
+      "step": 3644
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.612659980965433e-06,
+      "loss": 0.6782,
+      "step": 3645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.605282656930257e-06,
+      "loss": 0.7358,
+      "step": 3646
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5979150375157503e-06,
+      "loss": 0.6127,
+      "step": 3647
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.590557126832356e-06,
+      "loss": 0.713,
+      "step": 3648
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5832089289850903e-06,
+      "loss": 0.7152,
+      "step": 3649
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5758704480735696e-06,
+      "loss": 0.636,
+      "step": 3650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.56854168819197e-06,
+      "loss": 0.7016,
+      "step": 3651
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.561222653429056e-06,
+      "loss": 0.5739,
+      "step": 3652
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5539133478681622e-06,
+      "loss": 0.8533,
+      "step": 3653
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.546613775587188e-06,
+      "loss": 0.7065,
+      "step": 3654
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5393239406586266e-06,
+      "loss": 0.5733,
+      "step": 3655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5320438471495013e-06,
+      "loss": 0.59,
+      "step": 3656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.524773499121438e-06,
+      "loss": 0.8671,
+      "step": 3657
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.517512900630599e-06,
+      "loss": 0.5646,
+      "step": 3658
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.510262055727728e-06,
+      "loss": 0.595,
+      "step": 3659
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5030209684581007e-06,
+      "loss": 0.6469,
+      "step": 3660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4957896428615767e-06,
+      "loss": 0.6195,
+      "step": 3661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4885680829725533e-06,
+      "loss": 0.5974,
+      "step": 3662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4813562928199896e-06,
+      "loss": 0.6238,
+      "step": 3663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4741542764273808e-06,
+      "loss": 1.0022,
+      "step": 3664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4669620378127723e-06,
+      "loss": 0.6903,
+      "step": 3665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.459779580988768e-06,
+      "loss": 0.6013,
+      "step": 3666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.452606909962505e-06,
+      "loss": 0.9366,
+      "step": 3667
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4454440287356553e-06,
+      "loss": 0.7633,
+      "step": 3668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.438290941304433e-06,
+      "loss": 0.5948,
+      "step": 3669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4311476516596e-06,
+      "loss": 0.7194,
+      "step": 3670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4240141637864343e-06,
+      "loss": 0.5995,
+      "step": 3671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4168904816647555e-06,
+      "loss": 0.657,
+      "step": 3672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.409776609268906e-06,
+      "loss": 0.7118,
+      "step": 3673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4026725505677662e-06,
+      "loss": 0.657,
+      "step": 3674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.395578309524731e-06,
+      "loss": 0.7911,
+      "step": 3675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3884938900977207e-06,
+      "loss": 0.6562,
+      "step": 3676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.381419296239178e-06,
+      "loss": 0.8423,
+      "step": 3677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.374354531896057e-06,
+      "loss": 0.9423,
+      "step": 3678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.367299601009845e-06,
+      "loss": 0.7361,
+      "step": 3679
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.360254507516524e-06,
+      "loss": 0.7806,
+      "step": 3680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3532192553465972e-06,
+      "loss": 0.8095,
+      "step": 3681
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.346193848425069e-06,
+      "loss": 0.7115,
+      "step": 3682
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3391782906714686e-06,
+      "loss": 0.6981,
+      "step": 3683
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3321725859998147e-06,
+      "loss": 0.704,
+      "step": 3684
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.325176738318633e-06,
+      "loss": 0.7359,
+      "step": 3685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3181907515309466e-06,
+      "loss": 0.7106,
+      "step": 3686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3112146295342864e-06,
+      "loss": 1.004,
+      "step": 3687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3042483762206767e-06,
+      "loss": 0.9859,
+      "step": 3688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2972919954766202e-06,
+      "loss": 0.754,
+      "step": 3689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.290345491183139e-06,
+      "loss": 0.6004,
+      "step": 3690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2834088672157195e-06,
+      "loss": 0.6368,
+      "step": 3691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2764821274443617e-06,
+      "loss": 0.8698,
+      "step": 3692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2695652757335206e-06,
+      "loss": 0.6393,
+      "step": 3693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.262658315942161e-06,
+      "loss": 0.714,
+      "step": 3694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2557612519237117e-06,
+      "loss": 0.5387,
+      "step": 3695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2488740875261005e-06,
+      "loss": 0.6683,
+      "step": 3696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2419968265917012e-06,
+      "loss": 0.7073,
+      "step": 3697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2351294729573935e-06,
+      "loss": 0.5813,
+      "step": 3698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2282720304545123e-06,
+      "loss": 0.5587,
+      "step": 3699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2214245029088665e-06,
+      "loss": 0.7111,
+      "step": 3700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2145868941407354e-06,
+      "loss": 0.623,
+      "step": 3701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2077592079648593e-06,
+      "loss": 1.1318,
+      "step": 3702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2009414481904524e-06,
+      "loss": 0.5263,
+      "step": 3703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1941336186211835e-06,
+      "loss": 0.6944,
+      "step": 3704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.187335723055184e-06,
+      "loss": 0.974,
+      "step": 3705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1805477652850347e-06,
+      "loss": 0.5422,
+      "step": 3706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1737697490977892e-06,
+      "loss": 0.8572,
+      "step": 3707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1670016782749424e-06,
+      "loss": 0.8305,
+      "step": 3708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.160243556592441e-06,
+      "loss": 0.7302,
+      "step": 3709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.153495387820681e-06,
+      "loss": 0.8133,
+      "step": 3710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1467571757245076e-06,
+      "loss": 0.9667,
+      "step": 3711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1400289240632155e-06,
+      "loss": 0.5804,
+      "step": 3712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1333106365905353e-06,
+      "loss": 0.5757,
+      "step": 3713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1266023170546446e-06,
+      "loss": 0.6132,
+      "step": 3714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.119903969198147e-06,
+      "loss": 0.6447,
+      "step": 3715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.113215596758105e-06,
+      "loss": 0.6817,
+      "step": 3716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1065372034659948e-06,
+      "loss": 0.7426,
+      "step": 3717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.099868793047737e-06,
+      "loss": 0.6469,
+      "step": 3718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0932103692236726e-06,
+      "loss": 0.6189,
+      "step": 3719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0865619357085885e-06,
+      "loss": 0.6025,
+      "step": 3720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0799234962116798e-06,
+      "loss": 0.7966,
+      "step": 3721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.073295054436575e-06,
+      "loss": 0.8133,
+      "step": 3722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.066676614081322e-06,
+      "loss": 0.9139,
+      "step": 3723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.060068178838386e-06,
+      "loss": 0.7128,
+      "step": 3724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0534697523946656e-06,
+      "loss": 0.8256,
+      "step": 3725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0468813384314478e-06,
+      "loss": 0.6098,
+      "step": 3726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0403029406244634e-06,
+      "loss": 0.7121,
+      "step": 3727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0337345626438343e-06,
+      "loss": 0.6513,
+      "step": 3728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0271762081541023e-06,
+      "loss": 0.6629,
+      "step": 3729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0206278808142076e-06,
+      "loss": 0.848,
+      "step": 3730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.014089584277512e-06,
+      "loss": 0.9045,
+      "step": 3731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.007561322191769e-06,
+      "loss": 0.6794,
+      "step": 3732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0010430981991335e-06,
+      "loss": 0.6557,
+      "step": 3733
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.994534915936164e-06,
+      "loss": 0.5724,
+      "step": 3734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9880367790338127e-06,
+      "loss": 0.6769,
+      "step": 3735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9815486911174387e-06,
+      "loss": 0.7841,
+      "step": 3736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.975070655806781e-06,
+      "loss": 0.7789,
+      "step": 3737
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9686026767159763e-06,
+      "loss": 0.6957,
+      "step": 3738
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9621447574535458e-06,
+      "loss": 0.6393,
+      "step": 3739
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9556969016224124e-06,
+      "loss": 0.654,
+      "step": 3740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9492591128198703e-06,
+      "loss": 0.7104,
+      "step": 3741
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.942831394637601e-06,
+      "loss": 0.6518,
+      "step": 3742
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.936413750661663e-06,
+      "loss": 1.1178,
+      "step": 3743
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9300061844725103e-06,
+      "loss": 0.6526,
+      "step": 3744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.923608699644959e-06,
+      "loss": 0.5288,
+      "step": 3745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9172212997482044e-06,
+      "loss": 0.6923,
+      "step": 3746
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9108439883458184e-06,
+      "loss": 0.6107,
+      "step": 3747
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9044767689957378e-06,
+      "loss": 0.895,
+      "step": 3748
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8981196452502803e-06,
+      "loss": 0.6,
+      "step": 3749
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8917726206561228e-06,
+      "loss": 0.71,
+      "step": 3750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8854356987543077e-06,
+      "loss": 0.6301,
+      "step": 3751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8791088830802406e-06,
+      "loss": 0.7205,
+      "step": 3752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8727921771637024e-06,
+      "loss": 0.6315,
+      "step": 3753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8664855845288076e-06,
+      "loss": 0.6608,
+      "step": 3754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8601891086940549e-06,
+      "loss": 0.7632,
+      "step": 3755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8539027531722785e-06,
+      "loss": 0.5879,
+      "step": 3756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8476265214706867e-06,
+      "loss": 0.896,
+      "step": 3757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.841360417090816e-06,
+      "loss": 0.6747,
+      "step": 3758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8351044435285682e-06,
+      "loss": 0.7613,
+      "step": 3759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.828858604274193e-06,
+      "loss": 0.6063,
+      "step": 3760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8226229028122832e-06,
+      "loss": 0.6378,
+      "step": 3761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8163973426217717e-06,
+      "loss": 1.0155,
+      "step": 3762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8101819271759358e-06,
+      "loss": 0.7408,
+      "step": 3763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8039766599424014e-06,
+      "loss": 0.7986,
+      "step": 3764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7977815443831193e-06,
+      "loss": 0.7016,
+      "step": 3765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7915965839543869e-06,
+      "loss": 0.7003,
+      "step": 3766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7854217821068264e-06,
+      "loss": 0.9265,
+      "step": 3767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7792571422854044e-06,
+      "loss": 1.0865,
+      "step": 3768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7731026679294095e-06,
+      "loss": 0.9482,
+      "step": 3769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7669583624724597e-06,
+      "loss": 0.7511,
+      "step": 3770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7608242293425038e-06,
+      "loss": 0.5892,
+      "step": 3771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7547002719618067e-06,
+      "loss": 0.5928,
+      "step": 3772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7485864937469688e-06,
+      "loss": 0.7842,
+      "step": 3773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7424828981089037e-06,
+      "loss": 0.9015,
+      "step": 3774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.736389488452843e-06,
+      "loss": 0.6667,
+      "step": 3775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7303062681783344e-06,
+      "loss": 0.6392,
+      "step": 3776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.724233240679254e-06,
+      "loss": 1.2897,
+      "step": 3777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7181704093437756e-06,
+      "loss": 0.7529,
+      "step": 3778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7121177775543895e-06,
+      "loss": 0.7064,
+      "step": 3779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.706075348687901e-06,
+      "loss": 0.6364,
+      "step": 3780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7000431261154093e-06,
+      "loss": 0.5587,
+      "step": 3781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.694021113202342e-06,
+      "loss": 0.8338,
+      "step": 3782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6880093133084053e-06,
+      "loss": 0.5523,
+      "step": 3783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6820077297876269e-06,
+      "loss": 0.7214,
+      "step": 3784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6760163659883222e-06,
+      "loss": 0.6712,
+      "step": 3785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6700352252531194e-06,
+      "loss": 0.6497,
+      "step": 3786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640643109189225e-06,
+      "loss": 1.0323,
+      "step": 3787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6581036263169492e-06,
+      "loss": 0.807,
+      "step": 3788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6521531747726992e-06,
+      "loss": 0.5602,
+      "step": 3789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6462129596059728e-06,
+      "loss": 0.5867,
+      "step": 3790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6402829841308432e-06,
+      "loss": 1.0793,
+      "step": 3791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6343632516556818e-06,
+      "loss": 0.8179,
+      "step": 3792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6284537654831512e-06,
+      "loss": 0.6629,
+      "step": 3793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6225545289101851e-06,
+      "loss": 0.6926,
+      "step": 3794
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6166655452280066e-06,
+      "loss": 1.0916,
+      "step": 3795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6107868177221098e-06,
+      "loss": 0.9955,
+      "step": 3796
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6049183496722842e-06,
+      "loss": 0.7302,
+      "step": 3797
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5990601443525755e-06,
+      "loss": 0.8382,
+      "step": 3798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5932122050313182e-06,
+      "loss": 0.8082,
+      "step": 3799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5873745349711067e-06,
+      "loss": 0.7089,
+      "step": 3800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5815471374288227e-06,
+      "loss": 0.9433,
+      "step": 3801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5757300156556032e-06,
+      "loss": 0.6898,
+      "step": 3802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5699231728968567e-06,
+      "loss": 0.7192,
+      "step": 3803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.564126612392256e-06,
+      "loss": 0.5435,
+      "step": 3804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5583403373757344e-06,
+      "loss": 1.1356,
+      "step": 3805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5525643510754985e-06,
+      "loss": 0.6655,
+      "step": 3806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5467986567140015e-06,
+      "loss": 0.7629,
+      "step": 3807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5410432575079636e-06,
+      "loss": 0.6942,
+      "step": 3808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5352981566683522e-06,
+      "loss": 0.6155,
+      "step": 3809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.529563357400401e-06,
+      "loss": 0.5302,
+      "step": 3810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5238388629035884e-06,
+      "loss": 0.5873,
+      "step": 3811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5181246763716417e-06,
+      "loss": 0.5467,
+      "step": 3812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5124208009925424e-06,
+      "loss": 0.9137,
+      "step": 3813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5067272399485222e-06,
+      "loss": 0.6465,
+      "step": 3814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.501043996416054e-06,
+      "loss": 0.5334,
+      "step": 3815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4953710735658456e-06,
+      "loss": 0.5683,
+      "step": 3816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.489708474562863e-06,
+      "loss": 0.7358,
+      "step": 3817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.484056202566304e-06,
+      "loss": 0.7834,
+      "step": 3818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4784142607296104e-06,
+      "loss": 0.6356,
+      "step": 3819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4727826522004485e-06,
+      "loss": 0.5716,
+      "step": 3820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4671613801207318e-06,
+      "loss": 0.8283,
+      "step": 3821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4615504476266028e-06,
+      "loss": 1.0045,
+      "step": 3822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4559498578484399e-06,
+      "loss": 0.9966,
+      "step": 3823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4503596139108366e-06,
+      "loss": 0.8247,
+      "step": 3824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4447797189326318e-06,
+      "loss": 0.7623,
+      "step": 3825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.439210176026884e-06,
+      "loss": 0.6416,
+      "step": 3826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4336509883008742e-06,
+      "loss": 0.5402,
+      "step": 3827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.428102158856106e-06,
+      "loss": 0.6981,
+      "step": 3828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4225636907883034e-06,
+      "loss": 0.5004,
+      "step": 3829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4170355871874208e-06,
+      "loss": 0.7014,
+      "step": 3830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.411517851137616e-06,
+      "loss": 0.7276,
+      "step": 3831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4060104857172685e-06,
+      "loss": 1.0135,
+      "step": 3832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4005134939989673e-06,
+      "loss": 0.9122,
+      "step": 3833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3950268790495281e-06,
+      "loss": 0.6689,
+      "step": 3834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.38955064392996e-06,
+      "loss": 0.9275,
+      "step": 3835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3840847916954924e-06,
+      "loss": 0.5291,
+      "step": 3836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3786293253955508e-06,
+      "loss": 0.6187,
+      "step": 3837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3731842480737845e-06,
+      "loss": 0.6805,
+      "step": 3838
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3677495627680281e-06,
+      "loss": 0.7554,
+      "step": 3839
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3623252725103298e-06,
+      "loss": 0.6153,
+      "step": 3840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.356911380326935e-06,
+      "loss": 0.7432,
+      "step": 3841
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3515078892382816e-06,
+      "loss": 0.9366,
+      "step": 3842
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3461148022590176e-06,
+      "loss": 0.6985,
+      "step": 3843
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3407321223979807e-06,
+      "loss": 1.1322,
+      "step": 3844
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3353598526581956e-06,
+      "loss": 0.6841,
+      "step": 3845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3299979960368848e-06,
+      "loss": 0.61,
+      "step": 3846
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3246465555254683e-06,
+      "loss": 0.5668,
+      "step": 3847
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.319305534109543e-06,
+      "loss": 0.6911,
+      "step": 3848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.313974934768898e-06,
+      "loss": 0.9842,
+      "step": 3849
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3086547604775102e-06,
+      "loss": 0.6321,
+      "step": 3850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3033450142035298e-06,
+      "loss": 0.6405,
+      "step": 3851
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2980456989093115e-06,
+      "loss": 0.6975,
+      "step": 3852
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2927568175513616e-06,
+      "loss": 0.6574,
+      "step": 3853
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2874783730803908e-06,
+      "loss": 0.7354,
+      "step": 3854
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2822103684412657e-06,
+      "loss": 0.6666,
+      "step": 3855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.276952806573053e-06,
+      "loss": 0.7112,
+      "step": 3856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.271705690408962e-06,
+      "loss": 0.5385,
+      "step": 3857
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2664690228764044e-06,
+      "loss": 0.8235,
+      "step": 3858
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2612428068969428e-06,
+      "loss": 0.7186,
+      "step": 3859
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2560270453863233e-06,
+      "loss": 0.7104,
+      "step": 3860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.250821741254442e-06,
+      "loss": 0.9065,
+      "step": 3861
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2456268974053699e-06,
+      "loss": 0.6721,
+      "step": 3862
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2404425167373524e-06,
+      "loss": 0.7134,
+      "step": 3863
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2352686021427806e-06,
+      "loss": 0.5983,
+      "step": 3864
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2301051565082168e-06,
+      "loss": 0.6656,
+      "step": 3865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.224952182714374e-06,
+      "loss": 0.8065,
+      "step": 3866
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2198096836361351e-06,
+      "loss": 0.706,
+      "step": 3867
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2146776621425295e-06,
+      "loss": 0.6216,
+      "step": 3868
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2095561210967422e-06,
+      "loss": 0.8223,
+      "step": 3869
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2044450633561123e-06,
+      "loss": 0.6662,
+      "step": 3870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.199344491772134e-06,
+      "loss": 0.7954,
+      "step": 3871
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.194254409190445e-06,
+      "loss": 0.8654,
+      "step": 3872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1891748184508378e-06,
+      "loss": 0.6419,
+      "step": 3873
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1841057223872432e-06,
+      "loss": 0.5762,
+      "step": 3874
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.179047123827739e-06,
+      "loss": 0.7714,
+      "step": 3875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1739990255945566e-06,
+      "loss": 0.7447,
+      "step": 3876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1689614305040564e-06,
+      "loss": 0.587,
+      "step": 3877
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.163934341366746e-06,
+      "loss": 0.5809,
+      "step": 3878
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.158917760987266e-06,
+      "loss": 0.5439,
+      "step": 3879
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1539116921644e-06,
+      "loss": 0.6024,
+      "step": 3880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1489161376910674e-06,
+      "loss": 0.6107,
+      "step": 3881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1439311003543163e-06,
+      "loss": 0.5761,
+      "step": 3882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1389565829353267e-06,
+      "loss": 0.7313,
+      "step": 3883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1339925882094205e-06,
+      "loss": 0.6261,
+      "step": 3884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1290391189460381e-06,
+      "loss": 0.9352,
+      "step": 3885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1240961779087446e-06,
+      "loss": 0.9936,
+      "step": 3886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1191637678552447e-06,
+      "loss": 0.5659,
+      "step": 3887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1142418915373576e-06,
+      "loss": 1.0768,
+      "step": 3888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.109330551701029e-06,
+      "loss": 0.9692,
+      "step": 3889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1044297510863245e-06,
+      "loss": 0.6708,
+      "step": 3890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.099539492427435e-06,
+      "loss": 0.626,
+      "step": 3891
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0946597784526646e-06,
+      "loss": 0.9596,
+      "step": 3892
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0897906118844338e-06,
+      "loss": 0.6134,
+      "step": 3893
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.08493199543928e-06,
+      "loss": 0.7748,
+      "step": 3894
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0800839318278621e-06,
+      "loss": 0.6459,
+      "step": 3895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0752464237549387e-06,
+      "loss": 0.7554,
+      "step": 3896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0704194739193863e-06,
+      "loss": 0.6824,
+      "step": 3897
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.065603085014193e-06,
+      "loss": 0.7917,
+      "step": 3898
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.060797259726447e-06,
+      "loss": 0.7409,
+      "step": 3899
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.056002000737355e-06,
+      "loss": 0.6267,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0512173107222145e-06,
+      "loss": 0.6376,
+      "step": 3901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0464431923504415e-06,
+      "loss": 0.6534,
+      "step": 3902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0416796482855363e-06,
+      "loss": 0.6071,
+      "step": 3903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0369266811851174e-06,
+      "loss": 0.6957,
+      "step": 3904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0321842937008953e-06,
+      "loss": 0.5735,
+      "step": 3905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0274524884786729e-06,
+      "loss": 0.6869,
+      "step": 3906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0227312681583523e-06,
+      "loss": 0.6885,
+      "step": 3907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0180206353739376e-06,
+      "loss": 0.9761,
+      "step": 3908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0133205927535195e-06,
+      "loss": 0.8669,
+      "step": 3909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0086311429192741e-06,
+      "loss": 0.7852,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0039522884874818e-06,
+      "loss": 0.9424,
+      "step": 3911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.99284032068497e-07,
+      "loss": 0.5488,
+      "step": 3912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.946263762667808e-07,
+      "loss": 0.5744,
+      "step": 3913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.89979323680854e-07,
+      "loss": 0.7868,
+      "step": 3914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.85342876903348e-07,
+      "loss": 1.0192,
+      "step": 3915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.807170385209553e-07,
+      "loss": 0.7124,
+      "step": 3916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.761018111144716e-07,
+      "loss": 0.9352,
+      "step": 3917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.714971972587483e-07,
+      "loss": 0.7686,
+      "step": 3918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.669031995227374e-07,
+      "loss": 0.7432,
+      "step": 3919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.623198204694528e-07,
+      "loss": 0.6462,
+      "step": 3920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.577470626559938e-07,
+      "loss": 0.6436,
+      "step": 3921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.531849286335282e-07,
+      "loss": 0.6117,
+      "step": 3922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.48633420947298e-07,
+      "loss": 0.7972,
+      "step": 3923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.44092542136621e-07,
+      "loss": 1.0037,
+      "step": 3924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.395622947348814e-07,
+      "loss": 0.8076,
+      "step": 3925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.35042681269529e-07,
+      "loss": 0.7303,
+      "step": 3926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.305337042620844e-07,
+      "loss": 0.7892,
+      "step": 3927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.260353662281374e-07,
+      "loss": 0.6681,
+      "step": 3928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.215476696773363e-07,
+      "loss": 0.6294,
+      "step": 3929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.170706171133958e-07,
+      "loss": 0.7045,
+      "step": 3930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.126042110340894e-07,
+      "loss": 0.6226,
+      "step": 3931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.081484539312524e-07,
+      "loss": 0.9182,
+      "step": 3932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.037033482907853e-07,
+      "loss": 0.7109,
+      "step": 3933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.992688965926333e-07,
+      "loss": 0.6134,
+      "step": 3934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.948451013108062e-07,
+      "loss": 0.6922,
+      "step": 3935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.904319649133675e-07,
+      "loss": 0.5779,
+      "step": 3936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.860294898624344e-07,
+      "loss": 0.5244,
+      "step": 3937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.816376786141734e-07,
+      "loss": 0.6677,
+      "step": 3938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.772565336188066e-07,
+      "loss": 0.7274,
+      "step": 3939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.728860573205944e-07,
+      "loss": 0.7795,
+      "step": 3940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.685262521578597e-07,
+      "loss": 0.6555,
+      "step": 3941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.641771205629657e-07,
+      "loss": 0.7145,
+      "step": 3942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.598386649623136e-07,
+      "loss": 0.6392,
+      "step": 3943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.555108877763585e-07,
+      "loss": 0.6755,
+      "step": 3944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.511937914195911e-07,
+      "loss": 0.8004,
+      "step": 3945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.468873783005538e-07,
+      "loss": 0.9172,
+      "step": 3946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.42591650821809e-07,
+      "loss": 0.7969,
+      "step": 3947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.383066113799821e-07,
+      "loss": 0.5786,
+      "step": 3948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.340322623657138e-07,
+      "loss": 0.9882,
+      "step": 3949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.297686061636989e-07,
+      "loss": 0.5821,
+      "step": 3950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.25515645152648e-07,
+      "loss": 0.7413,
+      "step": 3951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.212733817053231e-07,
+      "loss": 0.9265,
+      "step": 3952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.170418181884998e-07,
+      "loss": 1.0789,
+      "step": 3953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.128209569630052e-07,
+      "loss": 0.6083,
+      "step": 3954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.086108003836757e-07,
+      "loss": 0.6978,
+      "step": 3955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.044113507993834e-07,
+      "loss": 0.601,
+      "step": 3956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.00222610553032e-07,
+      "loss": 0.5948,
+      "step": 3957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.960445819815432e-07,
+      "loss": 0.667,
+      "step": 3958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.918772674158659e-07,
+      "loss": 0.7868,
+      "step": 3959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.877206691809669e-07,
+      "loss": 0.9085,
+      "step": 3960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.835747895958445e-07,
+      "loss": 0.9382,
+      "step": 3961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.794396309735064e-07,
+      "loss": 0.7084,
+      "step": 3962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.753151956209848e-07,
+      "loss": 0.5574,
+      "step": 3963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.712014858393235e-07,
+      "loss": 0.7896,
+      "step": 3964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.670985039235956e-07,
+      "loss": 0.585,
+      "step": 3965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.630062521628723e-07,
+      "loss": 0.5971,
+      "step": 3966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.589247328402494e-07,
+      "loss": 0.6854,
+      "step": 3967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.548539482328321e-07,
+      "loss": 0.6354,
+      "step": 3968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.507939006117327e-07,
+      "loss": 0.5841,
+      "step": 3969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.467445922420836e-07,
+      "loss": 0.8075,
+      "step": 3970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.427060253830154e-07,
+      "loss": 0.5752,
+      "step": 3971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.386782022876682e-07,
+      "loss": 0.6718,
+      "step": 3972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.346611252031887e-07,
+      "loss": 0.6631,
+      "step": 3973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.306547963707334e-07,
+      "loss": 0.6977,
+      "step": 3974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.266592180254561e-07,
+      "loss": 0.9639,
+      "step": 3975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.226743923965141e-07,
+      "loss": 0.7614,
+      "step": 3976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.187003217070665e-07,
+      "loss": 0.7636,
+      "step": 3977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.147370081742689e-07,
+      "loss": 0.9765,
+      "step": 3978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.107844540092879e-07,
+      "loss": 0.6204,
+      "step": 3979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.068426614172686e-07,
+      "loss": 0.6265,
+      "step": 3980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.02911632597365e-07,
+      "loss": 0.6233,
+      "step": 3981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.989913697427186e-07,
+      "loss": 0.7194,
+      "step": 3982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.950818750404797e-07,
+      "loss": 0.61,
+      "step": 3983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.911831506717659e-07,
+      "loss": 0.6199,
+      "step": 3984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.872951988117083e-07,
+      "loss": 0.6188,
+      "step": 3985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.834180216294139e-07,
+      "loss": 0.5814,
+      "step": 3986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.795516212879904e-07,
+      "loss": 0.6073,
+      "step": 3987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.75695999944519e-07,
+      "loss": 0.7839,
+      "step": 3988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.71851159750081e-07,
+      "loss": 1.142,
+      "step": 3989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.680171028497362e-07,
+      "loss": 0.6589,
+      "step": 3990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.641938313825247e-07,
+      "loss": 0.9616,
+      "step": 3991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.603813474814757e-07,
+      "loss": 0.6638,
+      "step": 3992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.565796532735946e-07,
+      "loss": 0.8143,
+      "step": 3993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.527887508798758e-07,
+      "loss": 0.8812,
+      "step": 3994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.490086424152831e-07,
+      "loss": 0.6011,
+      "step": 3995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.452393299887649e-07,
+      "loss": 0.7597,
+      "step": 3996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.414808157032415e-07,
+      "loss": 0.669,
+      "step": 3997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.377331016556132e-07,
+      "loss": 0.5529,
+      "step": 3998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.33996189936752e-07,
+      "loss": 1.0063,
+      "step": 3999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.302700826315055e-07,
+      "loss": 0.7494,
+      "step": 4000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.265547818186913e-07,
+      "loss": 0.931,
+      "step": 4001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.228502895710953e-07,
+      "loss": 0.786,
+      "step": 4002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.191566079554822e-07,
+      "loss": 0.5724,
+      "step": 4003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.154737390325771e-07,
+      "loss": 0.7047,
+      "step": 4004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.118016848570763e-07,
+      "loss": 1.1019,
+      "step": 4005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.081404474776386e-07,
+      "loss": 0.7667,
+      "step": 4006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.044900289368971e-07,
+      "loss": 0.6731,
+      "step": 4007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.008504312714381e-07,
+      "loss": 0.7024,
+      "step": 4008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.972216565118172e-07,
+      "loss": 0.8241,
+      "step": 4009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.936037066825484e-07,
+      "loss": 1.0598,
+      "step": 4010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.899965838021126e-07,
+      "loss": 0.8041,
+      "step": 4011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.864002898829446e-07,
+      "loss": 0.6159,
+      "step": 4012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.828148269314326e-07,
+      "loss": 0.7369,
+      "step": 4013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.792401969479388e-07,
+      "loss": 0.5998,
+      "step": 4014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.756764019267635e-07,
+      "loss": 0.6122,
+      "step": 4015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.721234438561763e-07,
+      "loss": 0.7098,
+      "step": 4016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.685813247183891e-07,
+      "loss": 0.8715,
+      "step": 4017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.650500464895747e-07,
+      "loss": 0.777,
+      "step": 4018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.61529611139855e-07,
+      "loss": 0.8942,
+      "step": 4019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.580200206333053e-07,
+      "loss": 0.6398,
+      "step": 4020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.545212769279396e-07,
+      "loss": 0.6856,
+      "step": 4021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.510333819757385e-07,
+      "loss": 0.588,
+      "step": 4022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.475563377226146e-07,
+      "loss": 0.9207,
+      "step": 4023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.440901461084358e-07,
+      "loss": 0.7035,
+      "step": 4024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.406348090670088e-07,
+      "loss": 0.5577,
+      "step": 4025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.371903285260871e-07,
+      "loss": 0.7287,
+      "step": 4026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.337567064073712e-07,
+      "loss": 0.7214,
+      "step": 4027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.303339446264977e-07,
+      "loss": 0.7452,
+      "step": 4028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.269220450930479e-07,
+      "loss": 0.8914,
+      "step": 4029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.235210097105392e-07,
+      "loss": 0.7974,
+      "step": 4030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.201308403764338e-07,
+      "loss": 0.6814,
+      "step": 4031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.167515389821254e-07,
+      "loss": 0.6301,
+      "step": 4032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.133831074129503e-07,
+      "loss": 0.8045,
+      "step": 4033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.100255475481719e-07,
+      "loss": 0.6816,
+      "step": 4034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.066788612610007e-07,
+      "loss": 0.7032,
+      "step": 4035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.033430504185699e-07,
+      "loss": 1.018,
+      "step": 4036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.000181168819507e-07,
+      "loss": 0.6514,
+      "step": 4037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.967040625061414e-07,
+      "loss": 0.6661,
+      "step": 4038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.934008891400721e-07,
+      "loss": 0.6456,
+      "step": 4039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.901085986266085e-07,
+      "loss": 0.6466,
+      "step": 4040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.868271928025392e-07,
+      "loss": 0.7289,
+      "step": 4041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.835566734985797e-07,
+      "loss": 0.5965,
+      "step": 4042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.802970425393705e-07,
+      "loss": 0.5972,
+      "step": 4043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.770483017434857e-07,
+      "loss": 0.6059,
+      "step": 4044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.738104529234111e-07,
+      "loss": 0.7091,
+      "step": 4045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7058349788556834e-07,
+      "loss": 0.5626,
+      "step": 4046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6736743843029507e-07,
+      "loss": 0.7665,
+      "step": 4047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.641622763518472e-07,
+      "loss": 1.0324,
+      "step": 4048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.609680134384054e-07,
+      "loss": 0.9384,
+      "step": 4049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.577846514720685e-07,
+      "loss": 0.6232,
+      "step": 4050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.546121922288582e-07,
+      "loss": 0.774,
+      "step": 4051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5145063747870535e-07,
+      "loss": 0.9436,
+      "step": 4052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4829998898546335e-07,
+      "loss": 0.8038,
+      "step": 4053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.451602485068929e-07,
+      "loss": 0.6845,
+      "step": 4054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.420314177946816e-07,
+      "loss": 0.6686,
+      "step": 4055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.389134985944199e-07,
+      "loss": 0.5787,
+      "step": 4056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3580649264561626e-07,
+      "loss": 0.9965,
+      "step": 4057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.327104016816841e-07,
+      "loss": 0.7295,
+      "step": 4058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.296252274299595e-07,
+      "loss": 0.7896,
+      "step": 4059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2655097161167224e-07,
+      "loss": 0.9811,
+      "step": 4060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2348763594197484e-07,
+      "loss": 0.6655,
+      "step": 4061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.204352221299179e-07,
+      "loss": 0.6601,
+      "step": 4062
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1739373187846156e-07,
+      "loss": 0.9011,
+      "step": 4063
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1436316688447277e-07,
+      "loss": 0.6559,
+      "step": 4064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1134352883872574e-07,
+      "loss": 0.6121,
+      "step": 4065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.083348194258907e-07,
+      "loss": 0.8859,
+      "step": 4066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0533704032454714e-07,
+      "loss": 0.6093,
+      "step": 4067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0235019320717495e-07,
+      "loss": 1.0222,
+      "step": 4068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.993742797401545e-07,
+      "loss": 0.6912,
+      "step": 4069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.964093015837667e-07,
+      "loss": 0.7628,
+      "step": 4070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.934552603921926e-07,
+      "loss": 0.6857,
+      "step": 4071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9051215781350295e-07,
+      "loss": 0.6737,
+      "step": 4072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8757999548968416e-07,
+      "loss": 0.6414,
+      "step": 4073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8465877505659886e-07,
+      "loss": 0.7446,
+      "step": 4074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.817484981440167e-07,
+      "loss": 0.7068,
+      "step": 4075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.788491663755989e-07,
+      "loss": 0.6511,
+      "step": 4076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.759607813689048e-07,
+      "loss": 1.0974,
+      "step": 4077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7308334473537654e-07,
+      "loss": 0.5868,
+      "step": 4078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7021685808035667e-07,
+      "loss": 0.6846,
+      "step": 4079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6736132300307257e-07,
+      "loss": 0.6418,
+      "step": 4080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6451674109665214e-07,
+      "loss": 0.9228,
+      "step": 4081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.616831139480992e-07,
+      "loss": 0.5805,
+      "step": 4082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5886044313831137e-07,
+      "loss": 1.0169,
+      "step": 4083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5604873024208007e-07,
+      "loss": 0.8167,
+      "step": 4084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5324797682807056e-07,
+      "loss": 0.6536,
+      "step": 4085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5045818445884617e-07,
+      "loss": 0.6211,
+      "step": 4086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4767935469084413e-07,
+      "loss": 0.5871,
+      "step": 4087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.449114890743954e-07,
+      "loss": 0.6708,
+      "step": 4088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4215458915370705e-07,
+      "loss": 0.6139,
+      "step": 4089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3940865646687084e-07,
+      "loss": 0.7366,
+      "step": 4090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.366736925458591e-07,
+      "loss": 0.6714,
+      "step": 4091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3394969891652915e-07,
+      "loss": 1.0406,
+      "step": 4092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.312366770986075e-07,
+      "loss": 0.6876,
+      "step": 4093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.285346286057101e-07,
+      "loss": 0.9386,
+      "step": 4094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.258435549453265e-07,
+      "loss": 0.7569,
+      "step": 4095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.231634576188203e-07,
+      "loss": 0.6644,
+      "step": 4096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.204943381214376e-07,
+      "loss": 0.7216,
+      "step": 4097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1783619794229616e-07,
+      "loss": 0.6409,
+      "step": 4098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1518903856438966e-07,
+      "loss": 0.7342,
+      "step": 4099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1255286146457897e-07,
+      "loss": 0.9093,
+      "step": 4100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.099276681136121e-07,
+      "loss": 0.7516,
+      "step": 4101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0731345997609516e-07,
+      "loss": 0.7022,
+      "step": 4102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0471023851051496e-07,
+      "loss": 0.5462,
+      "step": 4103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.021180051692185e-07,
+      "loss": 0.9061,
+      "step": 4104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9953676139843567e-07,
+      "loss": 0.6584,
+      "step": 4105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.969665086382567e-07,
+      "loss": 0.893,
+      "step": 4106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.944072483226368e-07,
+      "loss": 0.8239,
+      "step": 4107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9185898187940933e-07,
+      "loss": 0.6246,
+      "step": 4108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.893217107302637e-07,
+      "loss": 0.5866,
+      "step": 4109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8679543629076324e-07,
+      "loss": 0.5771,
+      "step": 4110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.842801599703271e-07,
+      "loss": 0.9268,
+      "step": 4111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8177588317224835e-07,
+      "loss": 0.6521,
+      "step": 4112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7928260729367604e-07,
+      "loss": 0.6568,
+      "step": 4113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.768003337256242e-07,
+      "loss": 0.6492,
+      "step": 4114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.743290638529694e-07,
+      "loss": 0.9555,
+      "step": 4115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7186879905444886e-07,
+      "loss": 0.7898,
+      "step": 4116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6941954070265783e-07,
+      "loss": 0.5708,
+      "step": 4117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.669812901640545e-07,
+      "loss": 0.6784,
+      "step": 4118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6455404879895286e-07,
+      "loss": 0.6845,
+      "step": 4119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.621378179615253e-07,
+      "loss": 0.6991,
+      "step": 4120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.597325989998045e-07,
+      "loss": 0.6478,
+      "step": 4121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.57338393255675e-07,
+      "loss": 0.6021,
+      "step": 4122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5495520206487935e-07,
+      "loss": 0.9221,
+      "step": 4123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5258302675701396e-07,
+      "loss": 1.0818,
+      "step": 4124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.502218686555313e-07,
+      "loss": 0.6608,
+      "step": 4125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.478717290777377e-07,
+      "loss": 0.6448,
+      "step": 4126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4553260933478653e-07,
+      "loss": 0.8769,
+      "step": 4127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.432045107316916e-07,
+      "loss": 0.6018,
+      "step": 4128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4088743456730957e-07,
+      "loss": 0.6961,
+      "step": 4129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.38581382134353e-07,
+      "loss": 0.9184,
+      "step": 4130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3628635471938388e-07,
+      "loss": 0.8358,
+      "step": 4131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3400235360281354e-07,
+      "loss": 0.7591,
+      "step": 4132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.31729380058896e-07,
+      "loss": 1.0261,
+      "step": 4133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2946743535574133e-07,
+      "loss": 0.6662,
+      "step": 4134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.272165207553001e-07,
+      "loss": 0.8089,
+      "step": 4135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.249766375133722e-07,
+      "loss": 0.989,
+      "step": 4136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2274778687960242e-07,
+      "loss": 0.9163,
+      "step": 4137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2052997009748055e-07,
+      "loss": 0.7598,
+      "step": 4138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.183231884043413e-07,
+      "loss": 0.6524,
+      "step": 4139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1612744303135979e-07,
+      "loss": 0.6205,
+      "step": 4140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1394273520355834e-07,
+      "loss": 0.6824,
+      "step": 4141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1176906613979532e-07,
+      "loss": 0.8742,
+      "step": 4142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0960643705277839e-07,
+      "loss": 1.0175,
+      "step": 4143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0745484914904911e-07,
+      "loss": 0.6094,
+      "step": 4144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0531430362899173e-07,
+      "loss": 0.6922,
+      "step": 4145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0318480168683098e-07,
+      "loss": 0.5819,
+      "step": 4146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0106634451062977e-07,
+      "loss": 0.6198,
+      "step": 4147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9895893328228723e-07,
+      "loss": 0.6869,
+      "step": 4148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9686256917754055e-07,
+      "loss": 0.7477,
+      "step": 4149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.947772533659653e-07,
+      "loss": 0.6547,
+      "step": 4150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.927029870109731e-07,
+      "loss": 0.5975,
+      "step": 4151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9063977126980938e-07,
+      "loss": 0.6605,
+      "step": 4152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8858760729355107e-07,
+      "loss": 0.7455,
+      "step": 4153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8654649622712017e-07,
+      "loss": 0.6735,
+      "step": 4154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.845164392092591e-07,
+      "loss": 0.6242,
+      "step": 4155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8249743737255298e-07,
+      "loss": 0.741,
+      "step": 4156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8048949184341413e-07,
+      "loss": 0.77,
+      "step": 4157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7849260374208644e-07,
+      "loss": 0.701,
+      "step": 4158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7650677418264982e-07,
+      "loss": 0.633,
+      "step": 4159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.745320042730092e-07,
+      "loss": 0.9637,
+      "step": 4160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.725682951148988e-07,
+      "loss": 0.7645,
+      "step": 4161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7061564780388672e-07,
+      "loss": 0.782,
+      "step": 4162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6867406342936599e-07,
+      "loss": 0.8737,
+      "step": 4163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6674354307455675e-07,
+      "loss": 0.6664,
+      "step": 4164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6482408781651305e-07,
+      "loss": 0.7068,
+      "step": 4165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.629156987261049e-07,
+      "loss": 0.7896,
+      "step": 4166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6101837686804067e-07,
+      "loss": 0.6239,
+      "step": 4167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591321233008447e-07,
+      "loss": 0.618,
+      "step": 4168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5725693907686856e-07,
+      "loss": 0.595,
+      "step": 4169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.553928252422887e-07,
+      "loss": 0.6064,
+      "step": 4170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5353978283711103e-07,
+      "loss": 0.6741,
+      "step": 4171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5169781289515296e-07,
+      "loss": 0.5833,
+      "step": 4172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4986691644406582e-07,
+      "loss": 1.0581,
+      "step": 4173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4804709450531474e-07,
+      "loss": 0.6888,
+      "step": 4174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4623834809419203e-07,
+      "loss": 0.7535,
+      "step": 4175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4444067821980602e-07,
+      "loss": 0.6131,
+      "step": 4176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4265408588509e-07,
+      "loss": 0.676,
+      "step": 4177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4087857208679556e-07,
+      "loss": 0.6129,
+      "step": 4178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3911413781549032e-07,
+      "loss": 0.5827,
+      "step": 4179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3736078405556242e-07,
+      "loss": 1.0065,
+      "step": 4180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3561851178522267e-07,
+      "loss": 0.976,
+      "step": 4181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3388732197649356e-07,
+      "loss": 0.6922,
+      "step": 4182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3216721559521583e-07,
+      "loss": 0.6002,
+      "step": 4183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3045819360105072e-07,
+      "loss": 0.9856,
+      "step": 4184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2876025694746662e-07,
+      "loss": 0.6207,
+      "step": 4185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2707340658175692e-07,
+      "loss": 0.6129,
+      "step": 4186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2539764344502658e-07,
+      "loss": 0.5964,
+      "step": 4187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2373296847219218e-07,
+      "loss": 0.7676,
+      "step": 4188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2207938259198638e-07,
+      "loss": 0.8809,
+      "step": 4189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2043688672695564e-07,
+      "loss": 0.5901,
+      "step": 4190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1880548179345808e-07,
+      "loss": 0.6372,
+      "step": 4191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1718516870166563e-07,
+      "loss": 0.7162,
+      "step": 4192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1557594835555963e-07,
+      "loss": 0.6269,
+      "step": 4193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1397782165293303e-07,
+      "loss": 0.8518,
+      "step": 4194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1239078948539483e-07,
+      "loss": 0.6122,
+      "step": 4195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1081485273835679e-07,
+      "loss": 0.6892,
+      "step": 4196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0925001229104448e-07,
+      "loss": 0.6372,
+      "step": 4197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0769626901649288e-07,
+      "loss": 0.6779,
+      "step": 4198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0615362378154193e-07,
+      "loss": 0.6768,
+      "step": 4199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0462207744684761e-07,
+      "loss": 0.7883,
+      "step": 4200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0310163086686642e-07,
+      "loss": 0.7071,
+      "step": 4201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.015922848898665e-07,
+      "loss": 1.0889,
+      "step": 4202
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.000940403579187e-07,
+      "loss": 0.8862,
+      "step": 4203
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.860689810690549e-08,
+      "loss": 0.6439,
+      "step": 4204
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.713085896650987e-08,
+      "loss": 0.6298,
+      "step": 4205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.566592376022643e-08,
+      "loss": 0.6094,
+      "step": 4206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.42120933053503e-08,
+      "loss": 0.9047,
+      "step": 4207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.276936841298379e-08,
+      "loss": 0.7546,
+      "step": 4208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.133774988802746e-08,
+      "loss": 0.927,
+      "step": 4209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.991723852919576e-08,
+      "loss": 0.9979,
+      "step": 4210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.850783512899918e-08,
+      "loss": 0.6831,
+      "step": 4211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.710954047375097e-08,
+      "loss": 0.5807,
+      "step": 4212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.572235534356932e-08,
+      "loss": 0.6803,
+      "step": 4213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.434628051237292e-08,
+      "loss": 0.6194,
+      "step": 4214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.298131674788323e-08,
+      "loss": 0.6068,
+      "step": 4215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.16274648116222e-08,
+      "loss": 0.6226,
+      "step": 4216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.028472545891008e-08,
+      "loss": 0.8076,
+      "step": 4217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.895309943887208e-08,
+      "loss": 0.722,
+      "step": 4218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.763258749442948e-08,
+      "loss": 0.5895,
+      "step": 4219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.632319036230629e-08,
+      "loss": 0.6619,
+      "step": 4220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.502490877302038e-08,
+      "loss": 0.8153,
+      "step": 4221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.373774345089013e-08,
+      "loss": 0.6433,
+      "step": 4222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.246169511403666e-08,
+      "loss": 0.9898,
+      "step": 4223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.119676447437273e-08,
+      "loss": 0.6483,
+      "step": 4224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.99429522376116e-08,
+      "loss": 0.5781,
+      "step": 4225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.870025910326483e-08,
+      "loss": 0.6665,
+      "step": 4226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.746868576463339e-08,
+      "loss": 0.9212,
+      "step": 4227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.624823290882542e-08,
+      "loss": 0.8502,
+      "step": 4228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.50389012167385e-08,
+      "loss": 0.6387,
+      "step": 4229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.384069136306181e-08,
+      "loss": 0.684,
+      "step": 4230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.265360401628728e-08,
+      "loss": 0.5779,
+      "step": 4231
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.14776398387007e-08,
+      "loss": 0.6035,
+      "step": 4232
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.031279948637503e-08,
+      "loss": 0.6007,
+      "step": 4233
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9159083609183764e-08,
+      "loss": 0.6602,
+      "step": 4234
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.8016492850792026e-08,
+      "loss": 0.6471,
+      "step": 4235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.688502784865879e-08,
+      "loss": 0.6239,
+      "step": 4236
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.576468923403466e-08,
+      "loss": 0.9875,
+      "step": 4237
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.465547763196411e-08,
+      "loss": 0.6562,
+      "step": 4238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.3557393661283205e-08,
+      "loss": 0.6608,
+      "step": 4239
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.24704379346197e-08,
+      "loss": 0.6398,
+      "step": 4240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.139461105839516e-08,
+      "loss": 0.69,
+      "step": 4241
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.032991363281392e-08,
+      "loss": 0.5852,
+      "step": 4242
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.927634625188304e-08,
+      "loss": 1.0121,
+      "step": 4243
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.823390950339235e-08,
+      "loss": 0.7085,
+      "step": 4244
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.720260396892329e-08,
+      "loss": 0.6032,
+      "step": 4245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.618243022384894e-08,
+      "loss": 0.5883,
+      "step": 4246
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.517338883732958e-08,
+      "loss": 0.5965,
+      "step": 4247
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.417548037231489e-08,
+      "loss": 0.8022,
+      "step": 4248
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.31887053855462e-08,
+      "loss": 0.5829,
+      "step": 4249
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.22130644275498e-08,
+      "loss": 0.6144,
+      "step": 4250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.124855804264361e-08,
+      "loss": 0.5541,
+      "step": 4251
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0295186768928295e-08,
+      "loss": 0.6198,
+      "step": 4252
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9352951138300624e-08,
+      "loss": 0.7128,
+      "step": 4253
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.842185167643342e-08,
+      "loss": 0.7275,
+      "step": 4254
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.750188890279782e-08,
+      "loss": 0.8573,
+      "step": 4255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.65930633306455e-08,
+      "loss": 0.5831,
+      "step": 4256
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5695375467015295e-08,
+      "loss": 0.7318,
+      "step": 4257
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.480882581273326e-08,
+      "loss": 0.6072,
+      "step": 4258
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.393341486241264e-08,
+      "loss": 0.6291,
+      "step": 4259
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.306914310444498e-08,
+      "loss": 0.6515,
+      "step": 4260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2216011021020124e-08,
+      "loss": 0.6951,
+      "step": 4261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.137401908809956e-08,
+      "loss": 0.6193,
+      "step": 4262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0543167775440864e-08,
+      "loss": 0.8636,
+      "step": 4263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9723457546577683e-08,
+      "loss": 0.6293,
+      "step": 4264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.891488885883309e-08,
+      "loss": 0.6382,
+      "step": 4265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8117462163310683e-08,
+      "loss": 0.6961,
+      "step": 4266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.733117790490347e-08,
+      "loss": 0.8646,
+      "step": 4267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.655603652228056e-08,
+      "loss": 0.8296,
+      "step": 4268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5792038447898237e-08,
+      "loss": 0.7147,
+      "step": 4269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5039184107999992e-08,
+      "loss": 0.5986,
+      "step": 4270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4297473922603176e-08,
+      "loss": 0.6456,
+      "step": 4271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.356690830551456e-08,
+      "loss": 0.7244,
+      "step": 4272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.284748766432143e-08,
+      "loss": 0.5842,
+      "step": 4273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.213921240039163e-08,
+      "loss": 1.0687,
+      "step": 4274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1442082908877947e-08,
+      "loss": 0.788,
+      "step": 4275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0756099578713717e-08,
+      "loss": 0.7302,
+      "step": 4276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0081262792610578e-08,
+      "loss": 0.5841,
+      "step": 4277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.941757292706736e-08,
+      "loss": 0.9479,
+      "step": 4278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8765030352358992e-08,
+      "loss": 0.5493,
+      "step": 4279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.812363543254092e-08,
+      "loss": 0.7251,
+      "step": 4280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.749338852545579e-08,
+      "loss": 0.5865,
+      "step": 4281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6874289982720117e-08,
+      "loss": 0.6587,
+      "step": 4282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.626634014973094e-08,
+      "loss": 0.6422,
+      "step": 4283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5669539365670283e-08,
+      "loss": 0.7565,
+      "step": 4284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5083887963496247e-08,
+      "loss": 0.5746,
+      "step": 4285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4509386269945246e-08,
+      "loss": 0.6161,
+      "step": 4286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3946034605538671e-08,
+      "loss": 1.0367,
+      "step": 4287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3393833284571778e-08,
+      "loss": 0.5537,
+      "step": 4288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2852782615122572e-08,
+      "loss": 0.614,
+      "step": 4289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2322882899045152e-08,
+      "loss": 0.7597,
+      "step": 4290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1804134431974146e-08,
+      "loss": 0.5928,
+      "step": 4291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1296537503322492e-08,
+      "loss": 0.625,
+      "step": 4292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0800092396281436e-08,
+      "loss": 0.7807,
+      "step": 4293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.031479938782054e-08,
+      "loss": 1.1089,
+      "step": 4294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.840658748689891e-09,
+      "loss": 0.7523,
+      "step": 4295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.377670743413447e-09,
+      "loss": 0.6357,
+      "step": 4296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.925835630295699e-09,
+      "loss": 0.5811,
+      "step": 4297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.485153661419443e-09,
+      "loss": 0.7021,
+      "step": 4298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.05562508264135e-09,
+      "loss": 0.5573,
+      "step": 4299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.637250133600837e-09,
+      "loss": 1.3074,
+      "step": 4300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.2300290477089754e-09,
+      "loss": 0.6943,
+      "step": 4301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.833962052159582e-09,
+      "loss": 0.7822,
+      "step": 4302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.4490493679181254e-09,
+      "loss": 0.9893,
+      "step": 4303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0752912097328254e-09,
+      "loss": 0.8778,
+      "step": 4304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.71268778612355e-09,
+      "loss": 0.6665,
+      "step": 4305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.361239299390697e-09,
+      "loss": 0.6046,
+      "step": 4306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0209459456085354e-09,
+      "loss": 0.6387,
+      "step": 4307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.691807914627422e-09,
+      "loss": 0.7579,
+      "step": 4308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.373825390078245e-09,
+      "loss": 0.5757,
+      "step": 4309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.066998549363544e-09,
+      "loss": 0.7812,
+      "step": 4310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.771327563664162e-09,
+      "loss": 0.5553,
+      "step": 4311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.486812597937039e-09,
+      "loss": 0.568,
+      "step": 4312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2134538109152014e-09,
+      "loss": 0.7547,
+      "step": 4313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9512513551077647e-09,
+      "loss": 0.603,
+      "step": 4314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7002053767977152e-09,
+      "loss": 0.7053,
+      "step": 4315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4603160160441288e-09,
+      "loss": 0.7119,
+      "step": 4316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2315834066866105e-09,
+      "loss": 1.0643,
+      "step": 4317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0140076763341954e-09,
+      "loss": 0.8102,
+      "step": 4318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8075889463720076e-09,
+      "loss": 1.0664,
+      "step": 4319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6123273319657017e-09,
+      "loss": 0.6853,
+      "step": 4320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4282229420503612e-09,
+      "loss": 0.6619,
+      "step": 4321
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2552758793393793e-09,
+      "loss": 0.7038,
+      "step": 4322
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.093486240322239e-09,
+      "loss": 0.6181,
+      "step": 4323
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.428541152622927e-10,
+      "loss": 0.7138,
+      "step": 4324
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.033795881989826e-10,
+      "loss": 0.6351,
+      "step": 4325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.750627369411789e-10,
+      "loss": 0.7845,
+      "step": 4326
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.579036330827236e-10,
+      "loss": 0.7671,
+      "step": 4327
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.519023419868873e-10,
+      "loss": 0.6342,
+      "step": 4328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.57058922788589e-10,
+      "loss": 0.7431,
+      "step": 4329
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.73373428405499e-10,
+      "loss": 1.0083,
+      "step": 4330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0084590552693627e-10,
+      "loss": 0.729,
+      "step": 4331
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.394763946116484e-10,
+      "loss": 0.6677,
+      "step": 4332
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.926492990335434e-11,
+      "loss": 0.9514,
+      "step": 4333
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.021153941076051e-11,
+      "loss": 0.8446,
+      "step": 4334
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.231624492532447e-11,
+      "loss": 0.6859,
+      "step": 4335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.579062010152569e-12,
+      "loss": 0.844,
+      "step": 4336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.0944,
+      "step": 4337
+    },
+    {
+      "epoch": 1.0,
+      "step": 4337,
+      "total_flos": 179355569651712.0,
+      "train_loss": 0.783293194936557,
+      "train_runtime": 43655.3879,
+      "train_samples_per_second": 1.589,
+      "train_steps_per_second": 0.099
+    }
+  ],
+  "max_steps": 4337,
+  "num_train_epochs": 1,
+  "total_flos": 179355569651712.0,
+  "trial_name": null,
+  "trial_params": null
+}