diff --git "a/lora-checkpoints-6/trainer_state.json" "b/lora-checkpoints-6/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/lora-checkpoints-6/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": 1.5267175572519084e-06,
+      "loss": 1.1546,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.053435114503817e-06,
+      "loss": 1.5863,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.580152671755725e-06,
+      "loss": 1.7073,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.106870229007634e-06,
+      "loss": 1.1306,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.633587786259543e-06,
+      "loss": 1.133,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.16030534351145e-06,
+      "loss": 1.2965,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0687022900763359e-05,
+      "loss": 1.1996,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2213740458015267e-05,
+      "loss": 1.0919,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3740458015267178e-05,
+      "loss": 1.3265,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5267175572519086e-05,
+      "loss": 1.3408,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6793893129770993e-05,
+      "loss": 1.2151,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.83206106870229e-05,
+      "loss": 1.2364,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.984732824427481e-05,
+      "loss": 1.065,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1374045801526718e-05,
+      "loss": 1.5166,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2900763358778628e-05,
+      "loss": 1.0161,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4427480916030535e-05,
+      "loss": 1.0615,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5954198473282442e-05,
+      "loss": 1.4817,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7480916030534355e-05,
+      "loss": 1.073,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.900763358778626e-05,
+      "loss": 1.0492,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.053435114503817e-05,
+      "loss": 0.9933,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2061068702290076e-05,
+      "loss": 0.9525,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.358778625954199e-05,
+      "loss": 0.9916,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.511450381679389e-05,
+      "loss": 0.9651,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.66412213740458e-05,
+      "loss": 0.9715,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.816793893129771e-05,
+      "loss": 1.3068,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.969465648854962e-05,
+      "loss": 0.7443,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.122137404580153e-05,
+      "loss": 1.4003,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2748091603053435e-05,
+      "loss": 1.1682,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4274809160305345e-05,
+      "loss": 0.8639,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5801526717557256e-05,
+      "loss": 0.8664,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7328244274809166e-05,
+      "loss": 0.9105,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.885496183206107e-05,
+      "loss": 0.982,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.038167938931297e-05,
+      "loss": 1.5038,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1908396946564884e-05,
+      "loss": 0.9041,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3435114503816794e-05,
+      "loss": 0.996,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.496183206106871e-05,
+      "loss": 1.0536,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.648854961832062e-05,
+      "loss": 0.9698,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.801526717557252e-05,
+      "loss": 0.8943,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.954198473282443e-05,
+      "loss": 0.966,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.106870229007635e-05,
+      "loss": 0.9689,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.259541984732826e-05,
+      "loss": 1.0545,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.412213740458015e-05,
+      "loss": 0.9816,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.564885496183206e-05,
+      "loss": 0.919,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.717557251908397e-05,
+      "loss": 0.9964,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.870229007633588e-05,
+      "loss": 0.8414,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.022900763358778e-05,
+      "loss": 1.1612,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.175572519083969e-05,
+      "loss": 1.3616,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.32824427480916e-05,
+      "loss": 0.7844,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.480916030534351e-05,
+      "loss": 0.8545,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.633587786259542e-05,
+      "loss": 1.1792,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.786259541984733e-05,
+      "loss": 0.9386,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.938931297709924e-05,
+      "loss": 0.7205,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.091603053435115e-05,
+      "loss": 0.8396,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.244274809160306e-05,
+      "loss": 0.8684,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.396946564885496e-05,
+      "loss": 0.9061,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.549618320610687e-05,
+      "loss": 0.8258,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.702290076335878e-05,
+      "loss": 0.9018,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.854961832061069e-05,
+      "loss": 0.8268,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.007633587786259e-05,
+      "loss": 0.9554,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.160305343511451e-05,
+      "loss": 0.8164,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.312977099236642e-05,
+      "loss": 0.821,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.465648854961833e-05,
+      "loss": 1.0248,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.618320610687024e-05,
+      "loss": 0.6802,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.770992366412214e-05,
+      "loss": 0.8276,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.923664122137405e-05,
+      "loss": 0.7858,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010076335877862595,
+      "loss": 0.8977,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010229007633587786,
+      "loss": 0.8679,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010381679389312977,
+      "loss": 0.904,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010534351145038168,
+      "loss": 0.8214,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010687022900763359,
+      "loss": 0.9459,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001083969465648855,
+      "loss": 0.785,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010992366412213742,
+      "loss": 1.2907,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011145038167938933,
+      "loss": 0.9872,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011297709923664124,
+      "loss": 1.2553,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011450381679389313,
+      "loss": 0.9864,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011603053435114504,
+      "loss": 0.9119,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011755725190839695,
+      "loss": 1.2752,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011908396946564886,
+      "loss": 0.8041,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012061068702290077,
+      "loss": 0.8004,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001221374045801527,
+      "loss": 0.9727,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001236641221374046,
+      "loss": 0.8864,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001251908396946565,
+      "loss": 0.9085,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001267175572519084,
+      "loss": 0.8761,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001282442748091603,
+      "loss": 1.2123,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012977099236641222,
+      "loss": 0.833,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013129770992366413,
+      "loss": 0.9421,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013282442748091604,
+      "loss": 0.9944,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013435114503816795,
+      "loss": 0.8912,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013587786259541986,
+      "loss": 0.8884,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013740458015267177,
+      "loss": 0.8342,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013893129770992368,
+      "loss": 0.9156,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014045801526717556,
+      "loss": 0.7217,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014198473282442747,
+      "loss": 0.8196,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014351145038167938,
+      "loss": 0.9143,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001450381679389313,
+      "loss": 0.7995,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001465648854961832,
+      "loss": 0.9695,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001480916030534351,
+      "loss": 0.7807,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014961832061068702,
+      "loss": 1.0232,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015114503816793893,
+      "loss": 0.8656,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015267175572519084,
+      "loss": 0.8392,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015419847328244275,
+      "loss": 0.8116,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015572519083969466,
+      "loss": 0.8195,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015725190839694657,
+      "loss": 0.8049,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015877862595419848,
+      "loss": 0.8516,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001603053435114504,
+      "loss": 0.7971,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001618320610687023,
+      "loss": 0.7467,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016335877862595422,
+      "loss": 0.9368,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016488549618320613,
+      "loss": 1.0069,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016641221374045804,
+      "loss": 0.8476,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016793893129770992,
+      "loss": 0.9071,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016946564885496183,
+      "loss": 0.6956,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017099236641221374,
+      "loss": 0.9655,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017251908396946565,
+      "loss": 1.1701,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017404580152671756,
+      "loss": 0.7871,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017557251908396947,
+      "loss": 0.7961,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017709923664122138,
+      "loss": 1.0126,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001786259541984733,
+      "loss": 0.8955,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018015267175572518,
+      "loss": 0.8689,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001816793893129771,
+      "loss": 1.24,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018320610687022902,
+      "loss": 0.8132,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018473282442748093,
+      "loss": 0.9538,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018625954198473284,
+      "loss": 1.246,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018778625954198475,
+      "loss": 0.9238,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018931297709923666,
+      "loss": 0.8019,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019083969465648857,
+      "loss": 0.9632,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019236641221374049,
+      "loss": 0.8522,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019389312977099237,
+      "loss": 1.1283,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019541984732824428,
+      "loss": 0.8429,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001969465648854962,
+      "loss": 1.0552,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001984732824427481,
+      "loss": 0.8149,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9576,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999997210468995,
+      "loss": 1.1295,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999998884187754,
+      "loss": 0.9733,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999974894230294,
+      "loss": 0.9696,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999955367535052,
+      "loss": 0.9453,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999930261802696,
+      "loss": 1.2909,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999899577047238,
+      "loss": 0.9438,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999863313285797,
+      "loss": 0.8253,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999821470538607,
+      "loss": 1.273,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999977404882901,
+      "loss": 0.7705,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999972104818346,
+      "loss": 0.8617,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999966246863153,
+      "loss": 1.1986,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999598310205903,
+      "loss": 0.761,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999952857294237,
+      "loss": 0.9156,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999945325687984,
+      "loss": 1.1796,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999937236206033,
+      "loss": 0.913,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999285888528976,
+      "loss": 0.8992,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999193836334019,
+      "loss": 0.8251,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999096205526815,
+      "loss": 0.8035,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998992996161834,
+      "loss": 0.9628,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998884208296657,
+      "loss": 0.7468,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998769841991977,
+      "loss": 0.7993,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998649897311603,
+      "loss": 0.9362,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998524374322446,
+      "loss": 0.8359,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998393273094544,
+      "loss": 0.7985,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998256593701033,
+      "loss": 0.8121,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999811433621817,
+      "loss": 1.014,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997966500725318,
+      "loss": 0.7722,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999781308730496,
+      "loss": 0.8934,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997654096042688,
+      "loss": 0.7475,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997489527027199,
+      "loss": 0.8147,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997319380350305,
+      "loss": 0.8434,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997143656106937,
+      "loss": 0.9454,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996962354395135,
+      "loss": 0.864,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996775475316042,
+      "loss": 0.8543,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996583018973922,
+      "loss": 0.8873,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996384985476146,
+      "loss": 1.1531,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996181374933202,
+      "loss": 0.9526,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995972187458681,
+      "loss": 1.2017,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999575742316929,
+      "loss": 0.958,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995537082184855,
+      "loss": 1.0781,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995311164628293,
+      "loss": 1.0902,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995079670625656,
+      "loss": 1.1738,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999484260030609,
+      "loss": 1.1341,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999459995380186,
+      "loss": 0.7757,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994351731248338,
+      "loss": 0.9079,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994097932784014,
+      "loss": 0.7422,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993838558550478,
+      "loss": 0.974,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999357360869244,
+      "loss": 0.8435,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993303083357716,
+      "loss": 0.7936,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999302698269723,
+      "loss": 0.9566,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992745306865028,
+      "loss": 0.844,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992458056018253,
+      "loss": 1.0124,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992165230317167,
+      "loss": 1.1091,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991866829925135,
+      "loss": 0.7593,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999156285500864,
+      "loss": 0.8104,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991253305737274,
+      "loss": 0.8159,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999093818228373,
+      "loss": 0.6799,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999061748482382,
+      "loss": 0.868,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990291213536467,
+      "loss": 0.9316,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989959368603695,
+      "loss": 0.9541,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989621950210643,
+      "loss": 0.8874,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998927895854556,
+      "loss": 1.1573,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988930393799806,
+      "loss": 1.0309,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998857625616784,
+      "loss": 0.8342,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988216545847244,
+      "loss": 0.7609,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199878512630387,
+      "loss": 0.793,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987480407946003,
+      "loss": 0.7988,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998710398077605,
+      "loss": 0.7915,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998672198173886,
+      "loss": 0.85,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998633441104755,
+      "loss": 1.3173,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985941268918346,
+      "loss": 0.7593,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985542555570586,
+      "loss": 0.7129,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985138271226715,
+      "loss": 0.8979,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984728416112279,
+      "loss": 0.8509,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998431299045595,
+      "loss": 0.7998,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998389199448949,
+      "loss": 0.6884,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983465428447778,
+      "loss": 0.9169,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983033292568796,
+      "loss": 1.2225,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982595587093632,
+      "loss": 0.8136,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982152312266493,
+      "loss": 0.858,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981703468334677,
+      "loss": 0.9517,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981249055548602,
+      "loss": 0.7487,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980789074161784,
+      "loss": 0.803,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980323524430854,
+      "loss": 0.7833,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979852406615538,
+      "loss": 0.8543,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997937572097868,
+      "loss": 0.7173,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978893467786227,
+      "loss": 1.0233,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978405647307228,
+      "loss": 0.6834,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977912259813843,
+      "loss": 0.8346,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977413305581336,
+      "loss": 1.1292,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976908784888076,
+      "loss": 1.2794,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997639869801554,
+      "loss": 0.7956,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975883045248306,
+      "loss": 1.2634,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997536182687406,
+      "loss": 1.1791,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974835043183595,
+      "loss": 1.0454,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974302694470802,
+      "loss": 1.2369,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997376478103269,
+      "loss": 0.8643,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997322130316936,
+      "loss": 0.7962,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997267226118402,
+      "loss": 1.2207,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972117655382983,
+      "loss": 0.7815,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971557486075673,
+      "loss": 0.8343,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970991753574607,
+      "loss": 0.8784,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997042045819541,
+      "loss": 1.2791,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996984360025681,
+      "loss": 0.8633,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996926118008065,
+      "loss": 0.8007,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968673197991857,
+      "loss": 0.8133,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996807965431847,
+      "loss": 0.9939,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967480549391633,
+      "loss": 0.9059,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966875883545588,
+      "loss": 0.9679,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966265657117685,
+      "loss": 0.8924,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965649870448367,
+      "loss": 0.9238,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965028523881195,
+      "loss": 0.8652,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964401617762815,
+      "loss": 0.8877,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963769152442983,
+      "loss": 0.9958,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963131128274557,
+      "loss": 1.293,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996248754561349,
+      "loss": 0.9232,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996183840481885,
+      "loss": 0.8268,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961183706252786,
+      "loss": 1.3568,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960523450280565,
+      "loss": 1.295,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959857637270547,
+      "loss": 0.8285,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959186267594193,
+      "loss": 1.217,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995850934162606,
+      "loss": 0.7155,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957826859743816,
+      "loss": 0.8696,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995713882232822,
+      "loss": 0.7753,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956445229763126,
+      "loss": 0.9183,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955746082435502,
+      "loss": 1.1415,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955041380735402,
+      "loss": 0.9979,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954331125055988,
+      "loss": 0.8115,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953615315793511,
+      "loss": 0.899,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952893953347327,
+      "loss": 0.8056,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952167038119888,
+      "loss": 0.8602,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951434570516746,
+      "loss": 0.7165,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995069655094655,
+      "loss": 0.9685,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949952979821044,
+      "loss": 0.8333,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949203857555067,
+      "loss": 1.311,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019948449184566567,
+      "loss": 1.1102,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019947688961276575,
+      "loss": 0.7992,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994692318810923,
+      "loss": 0.8995,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019946151865491755,
+      "loss": 0.8893,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994537499385448,
+      "loss": 0.7274,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019944592573630822,
+      "loss": 1.1428,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943804605257305,
+      "loss": 1.1486,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943011089173536,
+      "loss": 0.7594,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942212025822223,
+      "loss": 1.1127,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941407415649168,
+      "loss": 0.9051,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940597259103271,
+      "loss": 0.9405,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939781556636523,
+      "loss": 0.8527,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993896030870401,
+      "loss": 0.782,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019938133515763905,
+      "loss": 0.9562,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937301178277487,
+      "loss": 0.7662,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019936463296709123,
+      "loss": 0.9841,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019935619871526268,
+      "loss": 0.7998,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019934770903199477,
+      "loss": 0.916,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933916392202394,
+      "loss": 1.0772,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933056339011755,
+      "loss": 0.7625,
+      "step": 286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993219074410739,
+      "loss": 0.7359,
+      "step": 287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019931319607972218,
+      "loss": 1.1514,
+      "step": 288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930442931092256,
+      "loss": 0.7518,
+      "step": 289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929560713956605,
+      "loss": 1.0342,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928672957057457,
+      "loss": 0.8803,
+      "step": 291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019927779660890098,
+      "loss": 0.8159,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019926880825952905,
+      "loss": 0.884,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925976452747344,
+      "loss": 0.7979,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925066541777968,
+      "loss": 1.089,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019924151093552426,
+      "loss": 0.9221,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923230108581446,
+      "loss": 0.8572,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019922303587378853,
+      "loss": 1.0085,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019921371530461567,
+      "loss": 0.9501,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920433938349578,
+      "loss": 0.6836,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991949081156598,
+      "loss": 0.8456,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019918542150636947,
+      "loss": 0.7528,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917587956091746,
+      "loss": 0.883,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916628228462723,
+      "loss": 1.0066,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019915662968285317,
+      "loss": 1.2758,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019914692176098057,
+      "loss": 0.8249,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019913715852442552,
+      "loss": 0.9097,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912733997863496,
+      "loss": 0.8596,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019911746612908676,
+      "loss": 1.1086,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019910753698128958,
+      "loss": 1.0976,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019909755254078295,
+      "loss": 1.2327,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019908751281313725,
+      "loss": 0.7876,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990774178039537,
+      "loss": 0.7474,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990672675188644,
+      "loss": 0.8396,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019905706196353228,
+      "loss": 1.1276,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019904680114365096,
+      "loss": 1.2545,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019903648506494514,
+      "loss": 0.7636,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990261137331702,
+      "loss": 0.8134,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019901568715411232,
+      "loss": 1.0041,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900520533358858,
+      "loss": 1.1382,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019899466827744686,
+      "loss": 1.2478,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019898407599156585,
+      "loss": 0.7872,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019897342848185505,
+      "loss": 0.8877,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019896272575425477,
+      "loss": 1.0388,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989519678147361,
+      "loss": 0.9082,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019894115466930103,
+      "loss": 0.8315,
+      "step": 326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019893028632398223,
+      "loss": 0.8299,
+      "step": 327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019891936278484322,
+      "loss": 0.8246,
+      "step": 328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019890838405797832,
+      "loss": 0.8126,
+      "step": 329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019889735014951262,
+      "loss": 0.9161,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198886261065602,
+      "loss": 0.7962,
+      "step": 331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019887511681243315,
+      "loss": 0.8136,
+      "step": 332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988639173962235,
+      "loss": 0.8017,
+      "step": 333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988526628232213,
+      "loss": 0.7724,
+      "step": 334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019884135309970554,
+      "loss": 0.9275,
+      "step": 335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019882998823198597,
+      "loss": 0.8199,
+      "step": 336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019881856822640307,
+      "loss": 0.8828,
+      "step": 337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019880709308932825,
+      "loss": 1.2037,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019879556282716347,
+      "loss": 1.1236,
+      "step": 339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019878397744634157,
+      "loss": 0.9634,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019877233695332608,
+      "loss": 0.7548,
+      "step": 341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019876064135461135,
+      "loss": 1.1535,
+      "step": 342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019874889065672234,
+      "loss": 0.7822,
+      "step": 343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019873708486621493,
+      "loss": 1.2502,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019872522398967562,
+      "loss": 0.7318,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871330803372165,
+      "loss": 1.0321,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198701337005001,
+      "loss": 0.7208,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986893109101924,
+      "loss": 0.749,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019867722975600523,
+      "loss": 1.2187,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866509354917974,
+      "loss": 0.7941,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019865290229648672,
+      "loss": 0.9278,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019864065600472776,
+      "loss": 1.1671,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019862835468073518,
+      "loss": 0.7631,
+      "step": 353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019861599833137188,
+      "loss": 1.1263,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019860358696353162,
+      "loss": 1.0701,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019859112058413878,
+      "loss": 0.8312,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019857859920014836,
+      "loss": 0.7714,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985660228185462,
+      "loss": 0.7718,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985533914463487,
+      "loss": 0.7403,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854070509060296,
+      "loss": 1.0102,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019852796375838684,
+      "loss": 0.9027,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019851516745680875,
+      "loss": 0.7969,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019850231619300784,
+      "loss": 1.0061,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019848940997415393,
+      "loss": 0.8356,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019847644880744745,
+      "loss": 1.1732,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019846343270011954,
+      "loss": 0.658,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019845036165943195,
+      "loss": 0.805,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019843723569267708,
+      "loss": 0.8707,
+      "step": 368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019842405480717808,
+      "loss": 0.9523,
+      "step": 369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019841081901028854,
+      "loss": 0.765,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019839752830939282,
+      "loss": 1.2218,
+      "step": 371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983841827119059,
+      "loss": 0.7314,
+      "step": 372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983707822252734,
+      "loss": 1.2364,
+      "step": 373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019835732685697145,
+      "loss": 0.8267,
+      "step": 374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000198343816614507,
+      "loss": 0.7219,
+      "step": 375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019833025150541738,
+      "loss": 0.6986,
+      "step": 376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019831663153727072,
+      "loss": 0.9211,
+      "step": 377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019830295671766565,
+      "loss": 0.7074,
+      "step": 378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019828922705423147,
+      "loss": 0.9359,
+      "step": 379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019827544255462804,
+      "loss": 1.3302,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019826160322654579,
+      "loss": 0.7465,
+      "step": 381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982477090777058,
+      "loss": 1.0179,
+      "step": 382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019823376011585966,
+      "loss": 1.0322,
+      "step": 383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982197563487896,
+      "loss": 1.0571,
+      "step": 384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019820569778430843,
+      "loss": 0.7983,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019819158443025954,
+      "loss": 0.7376,
+      "step": 386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019817741629451675,
+      "loss": 0.8892,
+      "step": 387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019816319338498465,
+      "loss": 0.8173,
+      "step": 388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019814891570959825,
+      "loss": 1.2744,
+      "step": 389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019813458327632314,
+      "loss": 0.8638,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019812019609315547,
+      "loss": 0.7957,
+      "step": 391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019810575416812204,
+      "loss": 0.7734,
+      "step": 392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019809125750927992,
+      "loss": 0.6754,
+      "step": 393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019807670612471704,
+      "loss": 1.3262,
+      "step": 394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980621000225516,
+      "loss": 1.1231,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019804743921093249,
+      "loss": 0.8486,
+      "step": 396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019803272369803905,
+      "loss": 0.9065,
+      "step": 397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980179534920812,
+      "loss": 0.7992,
+      "step": 398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019800312860129924,
+      "loss": 0.7934,
+      "step": 399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019798824903396415,
+      "loss": 0.9729,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019797331479837728,
+      "loss": 0.744,
+      "step": 401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019795832590287056,
+      "loss": 0.8752,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019794328235580636,
+      "loss": 0.8761,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019792818416557766,
+      "loss": 1.0826,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979130313406077,
+      "loss": 0.7547,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019789782388935043,
+      "loss": 0.7422,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019788256182029013,
+      "loss": 0.7482,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019786724514194167,
+      "loss": 0.852,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019785187386285022,
+      "loss": 0.845,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019783644799159159,
+      "loss": 0.7778,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019782096753677192,
+      "loss": 0.806,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978054325070279,
+      "loss": 0.8539,
+      "step": 412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977898429110266,
+      "loss": 0.8426,
+      "step": 413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019777419875746554,
+      "loss": 0.8122,
+      "step": 414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977585000550727,
+      "loss": 0.7802,
+      "step": 415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019774274681260648,
+      "loss": 0.982,
+      "step": 416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019772693903885572,
+      "loss": 0.7977,
+      "step": 417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019771107674263966,
+      "loss": 1.056,
+      "step": 418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019769515993280797,
+      "loss": 0.6789,
+      "step": 419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976791886182408,
+      "loss": 0.9158,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019766316280784855,
+      "loss": 0.8205,
+      "step": 421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019764708251057215,
+      "loss": 0.6775,
+      "step": 422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019763094773538295,
+      "loss": 1.0084,
+      "step": 423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976147584912826,
+      "loss": 0.9318,
+      "step": 424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019759851478730315,
+      "loss": 0.818,
+      "step": 425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019758221663250712,
+      "loss": 0.8088,
+      "step": 426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975658640359873,
+      "loss": 1.2447,
+      "step": 427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019754945700686696,
+      "loss": 1.2573,
+      "step": 428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019753299555429966,
+      "loss": 0.8221,
+      "step": 429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975164796874693,
+      "loss": 1.032,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019749990941559027,
+      "loss": 1.2792,
+      "step": 431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019748328474790717,
+      "loss": 0.7828,
+      "step": 432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197466605693695,
+      "loss": 1.2029,
+      "step": 433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019744987226225913,
+      "loss": 1.2488,
+      "step": 434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019743308446293524,
+      "loss": 0.8048,
+      "step": 435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019741624230508936,
+      "loss": 0.8388,
+      "step": 436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019739934579811785,
+      "loss": 1.0528,
+      "step": 437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019738239495144732,
+      "loss": 1.2781,
+      "step": 438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019736538977453477,
+      "loss": 0.9294,
+      "step": 439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019734833027686754,
+      "loss": 0.9181,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019733121646796318,
+      "loss": 0.9305,
+      "step": 441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019731404835736957,
+      "loss": 0.8277,
+      "step": 442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019729682595466498,
+      "loss": 0.8469,
+      "step": 443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019727954926945784,
+      "loss": 0.9038,
+      "step": 444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019726221831138695,
+      "loss": 0.8041,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972448330901213,
+      "loss": 1.0682,
+      "step": 446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019722739361536032,
+      "loss": 0.7675,
+      "step": 447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019720989989683348,
+      "loss": 0.7285,
+      "step": 448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019719235194430072,
+      "loss": 1.1822,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019717474976755215,
+      "loss": 1.0036,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019715709337640806,
+      "loss": 0.7835,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019713938278071913,
+      "loss": 1.26,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019712161799036618,
+      "loss": 0.9031,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019710379901526033,
+      "loss": 0.7482,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019708592586534284,
+      "loss": 0.911,
+      "step": 455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001970679985505853,
+      "loss": 1.2133,
+      "step": 456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019705001708098945,
+      "loss": 0.9758,
+      "step": 457
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019703198146658725,
+      "loss": 0.8193,
+      "step": 458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019701389171744094,
+      "loss": 1.0601,
+      "step": 459
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019699574784364283,
+      "loss": 0.7599,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969775498553155,
+      "loss": 0.7955,
+      "step": 461
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969592977626118,
+      "loss": 0.767,
+      "step": 462
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019694099157571465,
+      "loss": 1.0102,
+      "step": 463
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969226313048371,
+      "loss": 0.8788,
+      "step": 464
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969042169602226,
+      "loss": 1.1459,
+      "step": 465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019688574855214453,
+      "loss": 0.8838,
+      "step": 466
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968672260909066,
+      "loss": 0.7823,
+      "step": 467
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019684864958684247,
+      "loss": 0.9344,
+      "step": 468
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019683001905031623,
+      "loss": 1.252,
+      "step": 469
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019681133449172194,
+      "loss": 0.9055,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967925959214838,
+      "loss": 1.0438,
+      "step": 471
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019677380335005618,
+      "loss": 1.212,
+      "step": 472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967549567879236,
+      "loss": 1.239,
+      "step": 473
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019673605624560063,
+      "loss": 0.7648,
+      "step": 474
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019671710173363205,
+      "loss": 0.805,
+      "step": 475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019669809326259264,
+      "loss": 0.9149,
+      "step": 476
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966790308430874,
+      "loss": 0.9006,
+      "step": 477
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019665991448575133,
+      "loss": 0.8414,
+      "step": 478
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966407442012496,
+      "loss": 0.9258,
+      "step": 479
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019662152000027742,
+      "loss": 0.8916,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019660224189356007,
+      "loss": 1.034,
+      "step": 481
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019658290989185294,
+      "loss": 0.7244,
+      "step": 482
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965635240059415,
+      "loss": 0.7655,
+      "step": 483
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019654408424664118,
+      "loss": 0.7886,
+      "step": 484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965245906247976,
+      "loss": 0.9284,
+      "step": 485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965050431512864,
+      "loss": 0.7476,
+      "step": 486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964854418370132,
+      "loss": 1.3088,
+      "step": 487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964657866929137,
+      "loss": 1.0284,
+      "step": 488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019644607772995357,
+      "loss": 0.8506,
+      "step": 489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019642631495912864,
+      "loss": 0.8788,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019640649839146467,
+      "loss": 1.0119,
+      "step": 491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019638662803801745,
+      "loss": 0.7801,
+      "step": 492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019636670390987276,
+      "loss": 0.7888,
+      "step": 493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019634672601814634,
+      "loss": 0.9194,
+      "step": 494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019632669437398404,
+      "loss": 0.9316,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963066089885617,
+      "loss": 0.8659,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019628646987308494,
+      "loss": 1.1441,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001962662770387896,
+      "loss": 0.7294,
+      "step": 498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019624603049694133,
+      "loss": 1.1228,
+      "step": 499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019622573025883586,
+      "loss": 0.9146,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019620537633579876,
+      "loss": 0.7872,
+      "step": 501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019618496873918565,
+      "loss": 0.9351,
+      "step": 502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019616450748038205,
+      "loss": 0.8774,
+      "step": 503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019614399257080338,
+      "loss": 1.2214,
+      "step": 504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961234240218951,
+      "loss": 1.0441,
+      "step": 505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019610280184513247,
+      "loss": 0.8549,
+      "step": 506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019608212605202077,
+      "loss": 0.7308,
+      "step": 507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019606139665409518,
+      "loss": 1.165,
+      "step": 508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960406136629207,
+      "loss": 1.1357,
+      "step": 509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019601977709009233,
+      "loss": 0.7616,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019599888694723486,
+      "loss": 0.9183,
+      "step": 511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959779432460031,
+      "loss": 1.326,
+      "step": 512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019595694599808163,
+      "loss": 0.8369,
+      "step": 513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195935895215185,
+      "loss": 0.77,
+      "step": 514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959147909090575,
+      "loss": 0.8827,
+      "step": 515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019589363309147339,
+      "loss": 0.8357,
+      "step": 516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019587242177423678,
+      "loss": 1.0133,
+      "step": 517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019585115696918153,
+      "loss": 0.8825,
+      "step": 518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019582983868817144,
+      "loss": 0.8473,
+      "step": 519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958084669431001,
+      "loss": 0.8162,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019578704174589095,
+      "loss": 0.8316,
+      "step": 521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019576556310849728,
+      "loss": 0.8857,
+      "step": 522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019574403104290206,
+      "loss": 0.7637,
+      "step": 523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019572244556111823,
+      "loss": 0.9068,
+      "step": 524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019570080667518847,
+      "loss": 1.0564,
+      "step": 525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019567911439718517,
+      "loss": 1.2553,
+      "step": 526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956573687392107,
+      "loss": 0.9131,
+      "step": 527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019563556971339706,
+      "loss": 0.7132,
+      "step": 528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019561371733190598,
+      "loss": 0.9604,
+      "step": 529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019559181160692916,
+      "loss": 0.8298,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019556985255068785,
+      "loss": 0.8976,
+      "step": 531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019554784017543318,
+      "loss": 0.777,
+      "step": 532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195525774493446,
+      "loss": 1.296,
+      "step": 533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019550365551703687,
+      "loss": 0.8586,
+      "step": 534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954814832585461,
+      "loss": 1.0451,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019545925773034375,
+      "loss": 1.2724,
+      "step": 536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019543697894482958,
+      "loss": 0.8045,
+      "step": 537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019541464691443303,
+      "loss": 0.7609,
+      "step": 538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019539226165161334,
+      "loss": 0.7402,
+      "step": 539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019536982316885934,
+      "loss": 1.1307,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953473314786896,
+      "loss": 0.8778,
+      "step": 541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019532478659365236,
+      "loss": 0.8129,
+      "step": 542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019530218852632564,
+      "loss": 0.896,
+      "step": 543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952795372893169,
+      "loss": 0.8262,
+      "step": 544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019525683289526354,
+      "loss": 0.8404,
+      "step": 545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019523407535683237,
+      "loss": 0.8606,
+      "step": 546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019521126468672004,
+      "loss": 0.8855,
+      "step": 547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019518840089765275,
+      "loss": 0.8748,
+      "step": 548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019516548400238634,
+      "loss": 0.8412,
+      "step": 549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019514251401370626,
+      "loss": 0.9118,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019511949094442766,
+      "loss": 0.7992,
+      "step": 551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950964148073952,
+      "loss": 1.3221,
+      "step": 552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019507328561548328,
+      "loss": 1.1478,
+      "step": 553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019505010338159574,
+      "loss": 1.2623,
+      "step": 554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019502686811866612,
+      "loss": 0.8033,
+      "step": 555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950035798396575,
+      "loss": 0.7954,
+      "step": 556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019498023855756263,
+      "loss": 0.8727,
+      "step": 557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019495684428540364,
+      "loss": 0.7063,
+      "step": 558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019493339703623242,
+      "loss": 0.7898,
+      "step": 559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019490989682313033,
+      "loss": 1.0672,
+      "step": 560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019488634365920823,
+      "loss": 0.8212,
+      "step": 561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019486273755760664,
+      "loss": 0.7595,
+      "step": 562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019483907853149555,
+      "loss": 0.7281,
+      "step": 563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019481536659407442,
+      "loss": 0.7732,
+      "step": 564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019479160175857235,
+      "loss": 0.755,
+      "step": 565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947677840382478,
+      "loss": 0.7823,
+      "step": 566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019474391344638894,
+      "loss": 0.9015,
+      "step": 567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019471998999631324,
+      "loss": 0.6957,
+      "step": 568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019469601370136777,
+      "loss": 0.7812,
+      "step": 569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019467198457492907,
+      "loss": 1.2257,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019464790263040306,
+      "loss": 1.2484,
+      "step": 571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019462376788122532,
+      "loss": 0.8638,
+      "step": 572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945995803408607,
+      "loss": 0.8551,
+      "step": 573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945753400228036,
+      "loss": 0.9228,
+      "step": 574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019455104694057784,
+      "loss": 1.1859,
+      "step": 575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945267011077367,
+      "loss": 0.6843,
+      "step": 576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019450230253786283,
+      "loss": 1.2145,
+      "step": 577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019447785124456838,
+      "loss": 1.2083,
+      "step": 578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944533472414949,
+      "loss": 0.9576,
+      "step": 579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019442879054231321,
+      "loss": 0.8434,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019440418116072378,
+      "loss": 1.1739,
+      "step": 581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943795191104563,
+      "loss": 0.7465,
+      "step": 582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019435480440526984,
+      "loss": 0.8801,
+      "step": 583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019433003705895293,
+      "loss": 0.9207,
+      "step": 584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019430521708532335,
+      "loss": 0.7798,
+      "step": 585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019428034449822846,
+      "loss": 0.9339,
+      "step": 586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019425541931154466,
+      "loss": 0.9389,
+      "step": 587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000194230441539178,
+      "loss": 0.8092,
+      "step": 588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019420541119506368,
+      "loss": 0.9032,
+      "step": 589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941803282931663,
+      "loss": 0.8612,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941551928474797,
+      "loss": 0.983,
+      "step": 591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019413000487202725,
+      "loss": 0.9531,
+      "step": 592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019410476438086132,
+      "loss": 0.779,
+      "step": 593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940794713880638,
+      "loss": 0.7856,
+      "step": 594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019405412590774583,
+      "loss": 1.0216,
+      "step": 595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940287279540478,
+      "loss": 0.7323,
+      "step": 596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019400327754113934,
+      "loss": 0.7745,
+      "step": 597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019397777468321945,
+      "loss": 1.2279,
+      "step": 598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939522193945163,
+      "loss": 0.7155,
+      "step": 599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019392661168928736,
+      "loss": 0.9291,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019390095158181933,
+      "loss": 0.7314,
+      "step": 601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019387523908642816,
+      "loss": 0.8454,
+      "step": 602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019384947421745893,
+      "loss": 0.9141,
+      "step": 603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938236569892861,
+      "loss": 0.9407,
+      "step": 604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019379778741631325,
+      "loss": 0.7939,
+      "step": 605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019377186551297317,
+      "loss": 0.6055,
+      "step": 606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937458912937278,
+      "loss": 0.7844,
+      "step": 607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019371986477306841,
+      "loss": 0.844,
+      "step": 608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936937859655153,
+      "loss": 0.959,
+      "step": 609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193667654885618,
+      "loss": 0.7629,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936414715479552,
+      "loss": 0.7917,
+      "step": 611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019361523596713476,
+      "loss": 0.9305,
+      "step": 612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019358894815779367,
+      "loss": 0.9376,
+      "step": 613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001935626081345981,
+      "loss": 1.1899,
+      "step": 614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001935362159122432,
+      "loss": 0.8525,
+      "step": 615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019350977150545346,
+      "loss": 0.9707,
+      "step": 616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019348327492898235,
+      "loss": 0.852,
+      "step": 617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934567261976125,
+      "loss": 0.7572,
+      "step": 618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934301253261555,
+      "loss": 0.8025,
+      "step": 619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934034723294523,
+      "loss": 0.9823,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019337676722237267,
+      "loss": 0.7223,
+      "step": 621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019335001001981555,
+      "loss": 0.9449,
+      "step": 622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019332320073670904,
+      "loss": 0.8621,
+      "step": 623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001932963393880101,
+      "loss": 0.8772,
+      "step": 624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019326942598870493,
+      "loss": 0.8056,
+      "step": 625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019324246055380862,
+      "loss": 0.9217,
+      "step": 626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019321544309836534,
+      "loss": 0.8897,
+      "step": 627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019318837363744838,
+      "loss": 0.9014,
+      "step": 628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019316125218615988,
+      "loss": 1.0135,
+      "step": 629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001931340787596311,
+      "loss": 0.7273,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019310685337302225,
+      "loss": 1.2205,
+      "step": 631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019307957604152256,
+      "loss": 1.0962,
+      "step": 632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930522467803502,
+      "loss": 0.9125,
+      "step": 633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019302486560475237,
+      "loss": 0.8432,
+      "step": 634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019299743253000515,
+      "loss": 0.9843,
+      "step": 635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929699475714137,
+      "loss": 0.8269,
+      "step": 636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019294241074431195,
+      "loss": 0.7596,
+      "step": 637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929148220640629,
+      "loss": 1.3,
+      "step": 638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928871815460585,
+      "loss": 1.1972,
+      "step": 639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019285948920571948,
+      "loss": 0.8521,
+      "step": 640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019283174505849564,
+      "loss": 0.761,
+      "step": 641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019280394911986562,
+      "loss": 1.1071,
+      "step": 642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019277610140533687,
+      "loss": 0.8365,
+      "step": 643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019274820193044583,
+      "loss": 0.6364,
+      "step": 644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927202507107578,
+      "loss": 0.7958,
+      "step": 645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000192692247761867,
+      "loss": 0.7083,
+      "step": 646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019266419309939633,
+      "loss": 0.9081,
+      "step": 647
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019263608673899776,
+      "loss": 0.9205,
+      "step": 648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000192607928696352,
+      "loss": 0.7543,
+      "step": 649
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019257971898716848,
+      "loss": 0.8191,
+      "step": 650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925514576271857,
+      "loss": 1.0235,
+      "step": 651
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019252314463217077,
+      "loss": 0.8619,
+      "step": 652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019249478001791975,
+      "loss": 0.8556,
+      "step": 653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019246636380025739,
+      "loss": 0.8153,
+      "step": 654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001924378959950373,
+      "loss": 1.178,
+      "step": 655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001924093766181418,
+      "loss": 0.7266,
+      "step": 656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019238080568548206,
+      "loss": 1.0015,
+      "step": 657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000192352183212998,
+      "loss": 0.7283,
+      "step": 658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019232350921665826,
+      "loss": 0.8698,
+      "step": 659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001922947837124602,
+      "loss": 1.1796,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019226600671643,
+      "loss": 0.788,
+      "step": 661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019223717824462254,
+      "loss": 1.1232,
+      "step": 662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019220829831312136,
+      "loss": 0.7243,
+      "step": 663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019217936693803874,
+      "loss": 1.234,
+      "step": 664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019215038413551575,
+      "loss": 0.8195,
+      "step": 665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019212134992172198,
+      "loss": 0.9404,
+      "step": 666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920922643128559,
+      "loss": 0.8566,
+      "step": 667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019206312732514445,
+      "loss": 1.1095,
+      "step": 668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019203393897484342,
+      "loss": 0.8317,
+      "step": 669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920046992782371,
+      "loss": 0.8844,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001919754082516386,
+      "loss": 0.8865,
+      "step": 671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019194606591138946,
+      "loss": 0.7205,
+      "step": 672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019191667227386,
+      "loss": 0.7194,
+      "step": 673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918872273554491,
+      "loss": 0.8417,
+      "step": 674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019185773117258426,
+      "loss": 0.5987,
+      "step": 675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918281837417216,
+      "loss": 0.9765,
+      "step": 676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019179858507934583,
+      "loss": 0.8032,
+      "step": 677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019176893520197016,
+      "loss": 0.8371,
+      "step": 678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917392341261365,
+      "loss": 0.7644,
+      "step": 679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019170948186841526,
+      "loss": 0.9322,
+      "step": 680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916796784454054,
+      "loss": 0.7386,
+      "step": 681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019164982387373443,
+      "loss": 0.8466,
+      "step": 682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019161991817005837,
+      "loss": 1.0487,
+      "step": 683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915899613510619,
+      "loss": 1.2142,
+      "step": 684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019155995343345797,
+      "loss": 1.1719,
+      "step": 685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019152989443398832,
+      "loss": 0.8768,
+      "step": 686
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000191499784369423,
+      "loss": 1.2531,
+      "step": 687
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019146962325656053,
+      "loss": 0.8962,
+      "step": 688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019143941111222806,
+      "loss": 0.909,
+      "step": 689
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019140914795328113,
+      "loss": 0.8765,
+      "step": 690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019137883379660372,
+      "loss": 1.0028,
+      "step": 691
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019134846865910833,
+      "loss": 0.7574,
+      "step": 692
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913180525577358,
+      "loss": 0.8078,
+      "step": 693
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019128758550945548,
+      "loss": 0.8102,
+      "step": 694
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019125706753126518,
+      "loss": 1.2579,
+      "step": 695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019122649864019098,
+      "loss": 0.9694,
+      "step": 696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911958788532875,
+      "loss": 0.856,
+      "step": 697
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019116520818763773,
+      "loss": 0.8775,
+      "step": 698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019113448666035297,
+      "loss": 0.9279,
+      "step": 699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019110371428857298,
+      "loss": 0.7594,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019107289108946586,
+      "loss": 0.773,
+      "step": 701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001910420170802281,
+      "loss": 1.2051,
+      "step": 702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019101109227808442,
+      "loss": 0.8201,
+      "step": 703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019098011670028801,
+      "loss": 0.7431,
+      "step": 704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001909490903641203,
+      "loss": 0.8309,
+      "step": 705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019091801328689116,
+      "loss": 1.1606,
+      "step": 706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019088688548593862,
+      "loss": 1.1138,
+      "step": 707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019085570697862904,
+      "loss": 0.7258,
+      "step": 708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019082447778235716,
+      "loss": 1.1912,
+      "step": 709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019079319791454595,
+      "loss": 0.618,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001907618673926466,
+      "loss": 1.1366,
+      "step": 711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019073048623413863,
+      "loss": 0.8962,
+      "step": 712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019069905445652974,
+      "loss": 1.2416,
+      "step": 713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019066757207735597,
+      "loss": 0.7765,
+      "step": 714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019063603911418153,
+      "loss": 0.7907,
+      "step": 715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001906044555845988,
+      "loss": 1.267,
+      "step": 716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019057282150622849,
+      "loss": 1.2014,
+      "step": 717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019054113689671943,
+      "loss": 1.0028,
+      "step": 718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001905094017737486,
+      "loss": 0.8772,
+      "step": 719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019047761615502133,
+      "loss": 0.7907,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019044578005827093,
+      "loss": 0.8861,
+      "step": 721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000190413893501259,
+      "loss": 0.9266,
+      "step": 722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001903819565017752,
+      "loss": 1.1514,
+      "step": 723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019034996907763745,
+      "loss": 1.193,
+      "step": 724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019031793124669167,
+      "loss": 0.7525,
+      "step": 725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000190285843026812,
+      "loss": 0.9566,
+      "step": 726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019025370443590068,
+      "loss": 0.7745,
+      "step": 727
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019022151549188797,
+      "loss": 1.0209,
+      "step": 728
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019018927621273227,
+      "loss": 0.8073,
+      "step": 729
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019015698661642015,
+      "loss": 0.7195,
+      "step": 730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019012464672096612,
+      "loss": 0.8682,
+      "step": 731
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001900922565444128,
+      "loss": 0.9674,
+      "step": 732
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019005981610483095,
+      "loss": 0.8283,
+      "step": 733
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001900273254203192,
+      "loss": 0.8894,
+      "step": 734
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018999478450900433,
+      "loss": 0.9476,
+      "step": 735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018996219338904117,
+      "loss": 0.8763,
+      "step": 736
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018992955207861246,
+      "loss": 0.8214,
+      "step": 737
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018989686059592896,
+      "loss": 0.7835,
+      "step": 738
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018986411895922952,
+      "loss": 1.0257,
+      "step": 739
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018983132718678085,
+      "loss": 1.0944,
+      "step": 740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018979848529687772,
+      "loss": 0.9374,
+      "step": 741
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897655933078428,
+      "loss": 1.1382,
+      "step": 742
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897326512380267,
+      "loss": 0.7005,
+      "step": 743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896996591058081,
+      "loss": 0.7202,
+      "step": 744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896666169295934,
+      "loss": 0.7891,
+      "step": 745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018963352472781714,
+      "loss": 0.8588,
+      "step": 746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896003825189416,
+      "loss": 0.9114,
+      "step": 747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018956719032145708,
+      "loss": 0.8302,
+      "step": 748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018953394815388164,
+      "loss": 0.7658,
+      "step": 749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018950065603476132,
+      "loss": 0.7431,
+      "step": 750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018946731398267004,
+      "loss": 0.9631,
+      "step": 751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018943392201620948,
+      "loss": 1.0935,
+      "step": 752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018940048015400927,
+      "loss": 0.9248,
+      "step": 753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001893669884147268,
+      "loss": 0.7334,
+      "step": 754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018933344681704734,
+      "loss": 0.9748,
+      "step": 755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018929985537968395,
+      "loss": 1.0854,
+      "step": 756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018926621412137748,
+      "loss": 0.8709,
+      "step": 757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018923252306089663,
+      "loss": 0.8704,
+      "step": 758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018919878221703782,
+      "loss": 0.7703,
+      "step": 759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001891649916086253,
+      "loss": 0.7603,
+      "step": 760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018913115125451107,
+      "loss": 0.8603,
+      "step": 761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018909726117357484,
+      "loss": 0.7508,
+      "step": 762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018906332138472412,
+      "loss": 1.0953,
+      "step": 763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001890293319068941,
+      "loss": 0.8196,
+      "step": 764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018899529275904776,
+      "loss": 1.092,
+      "step": 765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001889612039601757,
+      "loss": 1.0597,
+      "step": 766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018892706552929632,
+      "loss": 0.8564,
+      "step": 767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888928774854557,
+      "loss": 0.78,
+      "step": 768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018885863984772745,
+      "loss": 0.8258,
+      "step": 769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018882435263521306,
+      "loss": 1.1033,
+      "step": 770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001887900158670415,
+      "loss": 0.9392,
+      "step": 771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018875562956236953,
+      "loss": 0.7998,
+      "step": 772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018872119374038144,
+      "loss": 0.7668,
+      "step": 773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886867084202892,
+      "loss": 1.1363,
+      "step": 774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886521736213324,
+      "loss": 0.8753,
+      "step": 775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018861758936277823,
+      "loss": 1.0005,
+      "step": 776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018858295566392142,
+      "loss": 0.7236,
+      "step": 777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885482725440843,
+      "loss": 0.7619,
+      "step": 778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018851354002261692,
+      "loss": 0.6708,
+      "step": 779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018847875811889666,
+      "loss": 0.9209,
+      "step": 780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018844392685232855,
+      "loss": 0.7264,
+      "step": 781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018840904624234524,
+      "loss": 0.8659,
+      "step": 782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018837411630840686,
+      "loss": 1.2368,
+      "step": 783
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018833913707000095,
+      "loss": 1.2405,
+      "step": 784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018830410854664265,
+      "loss": 0.7374,
+      "step": 785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018826903075787465,
+      "loss": 0.8916,
+      "step": 786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018823390372326704,
+      "loss": 0.8267,
+      "step": 787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881987274624174,
+      "loss": 0.895,
+      "step": 788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881635019949508,
+      "loss": 0.8401,
+      "step": 789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018812822734051973,
+      "loss": 0.7871,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018809290351880415,
+      "loss": 0.751,
+      "step": 791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001880575305495114,
+      "loss": 0.9827,
+      "step": 792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018802210845237638,
+      "loss": 1.1364,
+      "step": 793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018798663724716117,
+      "loss": 0.8747,
+      "step": 794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018795111695365545,
+      "loss": 0.8284,
+      "step": 795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018791554759167625,
+      "loss": 1.0535,
+      "step": 796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018787992918106784,
+      "loss": 1.0331,
+      "step": 797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018784426174170201,
+      "loss": 0.7169,
+      "step": 798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018780854529347785,
+      "loss": 0.77,
+      "step": 799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018777277985632175,
+      "loss": 0.8857,
+      "step": 800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001877369654501875,
+      "loss": 0.9018,
+      "step": 801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018770110209505618,
+      "loss": 0.7237,
+      "step": 802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018766518981093615,
+      "loss": 0.8863,
+      "step": 803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018762922861786314,
+      "loss": 0.6113,
+      "step": 804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018759321853590007,
+      "loss": 0.7622,
+      "step": 805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018755715958513724,
+      "loss": 0.762,
+      "step": 806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018752105178569212,
+      "loss": 0.7994,
+      "step": 807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001874848951577095,
+      "loss": 0.7333,
+      "step": 808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018744868972136138,
+      "loss": 0.9383,
+      "step": 809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018741243549684703,
+      "loss": 0.713,
+      "step": 810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018737613250439282,
+      "loss": 1.0946,
+      "step": 811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873397807642525,
+      "loss": 1.0076,
+      "step": 812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018730338029670688,
+      "loss": 1.1272,
+      "step": 813
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018726693112206406,
+      "loss": 1.0308,
+      "step": 814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001872304332606592,
+      "loss": 0.7424,
+      "step": 815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018719388673285473,
+      "loss": 0.7813,
+      "step": 816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018715729155904017,
+      "loss": 0.9649,
+      "step": 817
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018712064775963216,
+      "loss": 0.6178,
+      "step": 818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018708395535507453,
+      "loss": 0.8899,
+      "step": 819
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018704721436583826,
+      "loss": 1.0072,
+      "step": 820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018701042481242126,
+      "loss": 0.807,
+      "step": 821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018697358671534872,
+      "loss": 0.6548,
+      "step": 822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018693670009517285,
+      "loss": 0.7831,
+      "step": 823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001868997649724729,
+      "loss": 0.7152,
+      "step": 824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018686278136785517,
+      "loss": 0.8799,
+      "step": 825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018682574930195314,
+      "loss": 0.8156,
+      "step": 826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018678866879542715,
+      "loss": 0.7716,
+      "step": 827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018675153986896467,
+      "loss": 0.8243,
+      "step": 828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867143625432801,
+      "loss": 1.1379,
+      "step": 829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018667713683911502,
+      "loss": 0.8228,
+      "step": 830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018663986277723778,
+      "loss": 0.7806,
+      "step": 831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018660254037844388,
+      "loss": 0.9013,
+      "step": 832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018656516966355567,
+      "loss": 0.8739,
+      "step": 833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865277506534225,
+      "loss": 1.3261,
+      "step": 834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018649028336892068,
+      "loss": 1.0373,
+      "step": 835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018645276783095348,
+      "loss": 0.8204,
+      "step": 836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186415204060451,
+      "loss": 1.2178,
+      "step": 837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018637759207837032,
+      "loss": 0.8818,
+      "step": 838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018633993190569537,
+      "loss": 0.9419,
+      "step": 839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018630222356343703,
+      "loss": 0.949,
+      "step": 840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018626446707263302,
+      "loss": 0.8191,
+      "step": 841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001862266624543479,
+      "loss": 0.7543,
+      "step": 842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001861888097296731,
+      "loss": 0.8827,
+      "step": 843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018615090891972692,
+      "loss": 0.8868,
+      "step": 844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001861129600456544,
+      "loss": 0.7618,
+      "step": 845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018607496312862752,
+      "loss": 0.8195,
+      "step": 846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018603691818984492,
+      "loss": 1.0934,
+      "step": 847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018599882525053218,
+      "loss": 0.8049,
+      "step": 848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018596068433194155,
+      "loss": 0.8028,
+      "step": 849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001859224954553521,
+      "loss": 0.8277,
+      "step": 850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018588425864206963,
+      "loss": 1.3975,
+      "step": 851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858459739134267,
+      "loss": 1.1244,
+      "step": 852
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858076412907826,
+      "loss": 0.8393,
+      "step": 853
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018576926079552332,
+      "loss": 0.7982,
+      "step": 854
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018573083244906159,
+      "loss": 0.7915,
+      "step": 855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018569235627283683,
+      "loss": 0.86,
+      "step": 856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018565383228831515,
+      "loss": 0.8648,
+      "step": 857
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018561526051698925,
+      "loss": 1.1242,
+      "step": 858
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018557664098037865,
+      "loss": 0.8392,
+      "step": 859
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018553797370002937,
+      "loss": 0.8578,
+      "step": 860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018549925869751417,
+      "loss": 1.1891,
+      "step": 861
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018546049599443235,
+      "loss": 1.0554,
+      "step": 862
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018542168561240985,
+      "loss": 0.8322,
+      "step": 863
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853828275730993,
+      "loss": 0.8619,
+      "step": 864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018534392189817976,
+      "loss": 0.8288,
+      "step": 865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018530496860935696,
+      "loss": 0.9531,
+      "step": 866
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018526596772836323,
+      "loss": 0.9301,
+      "step": 867
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018522691927695735,
+      "loss": 1.1874,
+      "step": 868
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018518782327692473,
+      "loss": 0.7518,
+      "step": 869
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018514867975007727,
+      "loss": 1.1569,
+      "step": 870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018510948871825335,
+      "loss": 0.9468,
+      "step": 871
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018507025020331793,
+      "loss": 1.2472,
+      "step": 872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018503096422716237,
+      "loss": 0.9213,
+      "step": 873
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018499163081170465,
+      "loss": 1.0249,
+      "step": 874
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018495224997888902,
+      "loss": 1.1541,
+      "step": 875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018491282175068637,
+      "loss": 1.1721,
+      "step": 876
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001848733461490939,
+      "loss": 0.7213,
+      "step": 877
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018483382319613533,
+      "loss": 0.7785,
+      "step": 878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018479425291386073,
+      "loss": 0.9132,
+      "step": 879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018475463532434668,
+      "loss": 0.8495,
+      "step": 880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018471497044969597,
+      "loss": 0.8689,
+      "step": 881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018467525831203794,
+      "loss": 0.7155,
+      "step": 882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018463549893352822,
+      "loss": 0.689,
+      "step": 883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018459569233634886,
+      "loss": 0.6814,
+      "step": 884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018455583854270815,
+      "loss": 0.9616,
+      "step": 885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018451593757484077,
+      "loss": 0.8292,
+      "step": 886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018447598945500775,
+      "loss": 0.8145,
+      "step": 887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018443599420549638,
+      "loss": 0.9012,
+      "step": 888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018439595184862025,
+      "loss": 0.9587,
+      "step": 889
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018435586240671926,
+      "loss": 1.1108,
+      "step": 890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018431572590215956,
+      "loss": 0.7444,
+      "step": 891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018427554235733352,
+      "loss": 0.8102,
+      "step": 892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842353117946598,
+      "loss": 0.7013,
+      "step": 893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018419503423658328,
+      "loss": 0.7211,
+      "step": 894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018415470970557508,
+      "loss": 1.2538,
+      "step": 895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018411433822413252,
+      "loss": 0.9405,
+      "step": 896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018407391981477907,
+      "loss": 0.65,
+      "step": 897
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018403345450006438,
+      "loss": 0.9641,
+      "step": 898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018399294230256436,
+      "loss": 1.255,
+      "step": 899
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018395238324488098,
+      "loss": 0.707,
+      "step": 900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018391177734964243,
+      "loss": 0.7742,
+      "step": 901
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018387112463950293,
+      "loss": 0.6631,
+      "step": 902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018383042513714294,
+      "loss": 1.2275,
+      "step": 903
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018378967886526892,
+      "loss": 0.6871,
+      "step": 904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018374888584661347,
+      "loss": 0.7866,
+      "step": 905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001837080461039353,
+      "loss": 0.7481,
+      "step": 906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018366715966001916,
+      "loss": 0.7226,
+      "step": 907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018362622653767582,
+      "loss": 0.8211,
+      "step": 908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001835852467597421,
+      "loss": 0.8991,
+      "step": 909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001835442203490809,
+      "loss": 0.8461,
+      "step": 910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018350314732858115,
+      "loss": 1.1904,
+      "step": 911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018346202772115766,
+      "loss": 1.0999,
+      "step": 912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001834208615497514,
+      "loss": 0.6776,
+      "step": 913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018337964883732917,
+      "loss": 0.6549,
+      "step": 914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018333838960688382,
+      "loss": 0.9346,
+      "step": 915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018329708388143415,
+      "loss": 0.9335,
+      "step": 916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018325573168402484,
+      "loss": 1.1565,
+      "step": 917
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018321433303772656,
+      "loss": 1.1414,
+      "step": 918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018317288796563586,
+      "loss": 0.995,
+      "step": 919
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018313139649087524,
+      "loss": 1.05,
+      "step": 920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000183089858636593,
+      "loss": 1.1844,
+      "step": 921
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001830482744259634,
+      "loss": 0.7004,
+      "step": 922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001830066438821865,
+      "loss": 1.3182,
+      "step": 923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018296496702848827,
+      "loss": 0.7766,
+      "step": 924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018292324388812046,
+      "loss": 0.8426,
+      "step": 925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018288147448436068,
+      "loss": 1.0324,
+      "step": 926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018283965884051235,
+      "loss": 0.972,
+      "step": 927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018279779697990466,
+      "loss": 0.9418,
+      "step": 928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018275588892589263,
+      "loss": 0.7835,
+      "step": 929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000182713934701857,
+      "loss": 0.9306,
+      "step": 930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018267193433120426,
+      "loss": 0.8263,
+      "step": 931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018262988783736674,
+      "loss": 1.0205,
+      "step": 932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001825877952438024,
+      "loss": 0.8446,
+      "step": 933
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018254565657399496,
+      "loss": 1.2698,
+      "step": 934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018250347185145386,
+      "loss": 0.7955,
+      "step": 935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018246124109971422,
+      "loss": 1.2737,
+      "step": 936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018241896434233686,
+      "loss": 0.7881,
+      "step": 937
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018237664160290816,
+      "loss": 0.7551,
+      "step": 938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018233427290504031,
+      "loss": 1.17,
+      "step": 939
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018229185827237107,
+      "loss": 0.915,
+      "step": 940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001822493977285638,
+      "loss": 0.8124,
+      "step": 941
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018220689129730752,
+      "loss": 0.8087,
+      "step": 942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018216433900231682,
+      "loss": 0.9861,
+      "step": 943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018212174086733188,
+      "loss": 0.867,
+      "step": 944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001820790969161185,
+      "loss": 0.8329,
+      "step": 945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018203640717246797,
+      "loss": 0.8455,
+      "step": 946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018199367166019716,
+      "loss": 1.3677,
+      "step": 947
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018195089040314852,
+      "loss": 0.7308,
+      "step": 948
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001819080634251899,
+      "loss": 0.6444,
+      "step": 949
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018186519075021483,
+      "loss": 0.7601,
+      "step": 950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018182227240214218,
+      "loss": 1.0946,
+      "step": 951
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018177930840491637,
+      "loss": 1.2936,
+      "step": 952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001817362987825073,
+      "loss": 0.7296,
+      "step": 953
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018169324355891031,
+      "loss": 0.8125,
+      "step": 954
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018165014275814615,
+      "loss": 0.9922,
+      "step": 955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018160699640426103,
+      "loss": 1.1033,
+      "step": 956
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001815638045213266,
+      "loss": 0.9574,
+      "step": 957
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001815205671334398,
+      "loss": 0.88,
+      "step": 958
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018147728426472313,
+      "loss": 0.9139,
+      "step": 959
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018143395593932433,
+      "loss": 1.3666,
+      "step": 960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018139058218141654,
+      "loss": 0.7359,
+      "step": 961
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018134716301519823,
+      "loss": 0.7048,
+      "step": 962
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018130369846489325,
+      "loss": 1.3398,
+      "step": 963
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018126018855475073,
+      "loss": 1.0966,
+      "step": 964
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018121663330904517,
+      "loss": 0.9221,
+      "step": 965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018117303275207622,
+      "loss": 0.876,
+      "step": 966
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018112938690816895,
+      "loss": 0.9349,
+      "step": 967
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018108569580167365,
+      "loss": 0.7877,
+      "step": 968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018104195945696584,
+      "loss": 0.8738,
+      "step": 969
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018099817789844633,
+      "loss": 0.8358,
+      "step": 970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018095435115054112,
+      "loss": 0.5896,
+      "step": 971
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018091047923770134,
+      "loss": 1.1552,
+      "step": 972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018086656218440353,
+      "loss": 0.6735,
+      "step": 973
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018082260001514922,
+      "loss": 0.7837,
+      "step": 974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018077859275446518,
+      "loss": 0.7476,
+      "step": 975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001807345404269033,
+      "loss": 1.105,
+      "step": 976
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018069044305704071,
+      "loss": 0.7346,
+      "step": 977
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001806463006694796,
+      "loss": 0.7085,
+      "step": 978
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018060211328884725,
+      "loss": 0.7481,
+      "step": 979
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018055788093979607,
+      "loss": 0.8372,
+      "step": 980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018051360364700357,
+      "loss": 0.8691,
+      "step": 981
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018046928143517236,
+      "loss": 0.8323,
+      "step": 982
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018042491432903007,
+      "loss": 0.7827,
+      "step": 983
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001803805023533293,
+      "loss": 1.0576,
+      "step": 984
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018033604553284785,
+      "loss": 0.7029,
+      "step": 985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018029154389238844,
+      "loss": 0.6926,
+      "step": 986
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018024699745677883,
+      "loss": 0.7634,
+      "step": 987
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018020240625087172,
+      "loss": 0.8089,
+      "step": 988
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018015777029954482,
+      "loss": 0.7396,
+      "step": 989
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018011308962770082,
+      "loss": 1.0392,
+      "step": 990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018006836426026735,
+      "loss": 0.7445,
+      "step": 991
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018002359422219694,
+      "loss": 0.9733,
+      "step": 992
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017997877953846708,
+      "loss": 0.8706,
+      "step": 993
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001799339202340802,
+      "loss": 0.8731,
+      "step": 994
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017988901633406355,
+      "loss": 1.1162,
+      "step": 995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001798440678634693,
+      "loss": 0.8191,
+      "step": 996
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017979907484737444,
+      "loss": 0.6713,
+      "step": 997
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001797540373108809,
+      "loss": 0.7968,
+      "step": 998
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001797089552791154,
+      "loss": 1.0307,
+      "step": 999
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001796638287772295,
+      "loss": 0.6392,
+      "step": 1000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017961865783039951,
+      "loss": 0.7386,
+      "step": 1001
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017957344246382663,
+      "loss": 1.1037,
+      "step": 1002
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017952818270273674,
+      "loss": 0.6614,
+      "step": 1003
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017948287857238058,
+      "loss": 0.7981,
+      "step": 1004
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001794375300980336,
+      "loss": 0.7061,
+      "step": 1005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017939213730499597,
+      "loss": 0.8082,
+      "step": 1006
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017934670021859262,
+      "loss": 0.7921,
+      "step": 1007
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017930121886417321,
+      "loss": 0.8962,
+      "step": 1008
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017925569326711206,
+      "loss": 0.7873,
+      "step": 1009
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017921012345280816,
+      "loss": 0.7909,
+      "step": 1010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017916450944668522,
+      "loss": 0.8408,
+      "step": 1011
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017911885127419157,
+      "loss": 0.8655,
+      "step": 1012
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017907314896080015,
+      "loss": 0.8107,
+      "step": 1013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017902740253200863,
+      "loss": 0.8153,
+      "step": 1014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017898161201333914,
+      "loss": 0.8115,
+      "step": 1015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017893577743033858,
+      "loss": 1.0721,
+      "step": 1016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017888989880857832,
+      "loss": 0.8559,
+      "step": 1017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017884397617365432,
+      "loss": 0.6862,
+      "step": 1018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017879800955118708,
+      "loss": 0.6784,
+      "step": 1019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001787519989668217,
+      "loss": 0.7706,
+      "step": 1020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017870594444622774,
+      "loss": 0.6798,
+      "step": 1021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017865984601509935,
+      "loss": 0.9224,
+      "step": 1022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017861370369915512,
+      "loss": 1.1769,
+      "step": 1023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017856751752413807,
+      "loss": 0.8042,
+      "step": 1024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001785212875158158,
+      "loss": 0.8167,
+      "step": 1025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017847501369998035,
+      "loss": 0.9939,
+      "step": 1026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017842869610244812,
+      "loss": 0.7204,
+      "step": 1027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017838233474905997,
+      "loss": 1.0586,
+      "step": 1028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017833592966568128,
+      "loss": 0.9607,
+      "step": 1029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017828948087820164,
+      "loss": 0.9898,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017824298841253512,
+      "loss": 0.8041,
+      "step": 1031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001781964522946202,
+      "loss": 0.6818,
+      "step": 1032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017814987255041967,
+      "loss": 0.9165,
+      "step": 1033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017810324920592061,
+      "loss": 0.7775,
+      "step": 1034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017805658228713453,
+      "loss": 0.6994,
+      "step": 1035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017800987182009718,
+      "loss": 0.9052,
+      "step": 1036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017796311783086855,
+      "loss": 0.7336,
+      "step": 1037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017791632034553308,
+      "loss": 0.7305,
+      "step": 1038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001778694793901993,
+      "loss": 0.9161,
+      "step": 1039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017782259499100013,
+      "loss": 0.724,
+      "step": 1040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001777756671740926,
+      "loss": 0.7643,
+      "step": 1041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001777286959656581,
+      "loss": 0.7094,
+      "step": 1042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017768168139190214,
+      "loss": 0.8283,
+      "step": 1043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001776346234790544,
+      "loss": 0.9304,
+      "step": 1044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017758752225336885,
+      "loss": 0.8892,
+      "step": 1045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001775403777411235,
+      "loss": 1.1047,
+      "step": 1046
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017749318996862056,
+      "loss": 0.8218,
+      "step": 1047
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017744595896218644,
+      "loss": 0.8479,
+      "step": 1048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017739868474817155,
+      "loss": 0.7091,
+      "step": 1049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017735136735295048,
+      "loss": 0.7839,
+      "step": 1050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017730400680292192,
+      "loss": 0.7897,
+      "step": 1051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017725660312450863,
+      "loss": 0.8386,
+      "step": 1052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017720915634415735,
+      "loss": 0.8171,
+      "step": 1053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000177161666488339,
+      "loss": 0.8983,
+      "step": 1054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017711413358354844,
+      "loss": 0.7345,
+      "step": 1055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017706655765630457,
+      "loss": 1.0389,
+      "step": 1056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001770189387331503,
+      "loss": 1.0126,
+      "step": 1057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001769712768406525,
+      "loss": 0.7894,
+      "step": 1058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017692357200540208,
+      "loss": 0.7014,
+      "step": 1059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017687582425401384,
+      "loss": 0.853,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001768280336131265,
+      "loss": 0.9014,
+      "step": 1061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017678020010940284,
+      "loss": 0.9339,
+      "step": 1062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001767323237695294,
+      "loss": 0.7658,
+      "step": 1063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017668440462021673,
+      "loss": 0.9859,
+      "step": 1064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017663644268819923,
+      "loss": 1.0104,
+      "step": 1065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001765884380002351,
+      "loss": 0.7192,
+      "step": 1066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001765403905831065,
+      "loss": 0.7017,
+      "step": 1067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017649230046361936,
+      "loss": 0.8086,
+      "step": 1068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017644416766860347,
+      "loss": 0.7409,
+      "step": 1069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001763959922249124,
+      "loss": 0.6884,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017634777415942356,
+      "loss": 0.8594,
+      "step": 1071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001762995134990381,
+      "loss": 0.8932,
+      "step": 1072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017625121027068088,
+      "loss": 0.9334,
+      "step": 1073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017620286450130063,
+      "loss": 0.6501,
+      "step": 1074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017615447621786975,
+      "loss": 0.7798,
+      "step": 1075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017610604544738437,
+      "loss": 0.9686,
+      "step": 1076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001760575722168643,
+      "loss": 0.7503,
+      "step": 1077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017600905655335304,
+      "loss": 0.8803,
+      "step": 1078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017596049848391782,
+      "loss": 0.7921,
+      "step": 1079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017591189803564948,
+      "loss": 0.762,
+      "step": 1080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017586325523566247,
+      "loss": 0.9039,
+      "step": 1081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017581457011109494,
+      "loss": 0.691,
+      "step": 1082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017576584268910863,
+      "loss": 0.9576,
+      "step": 1083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017571707299688887,
+      "loss": 0.6806,
+      "step": 1084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017566826106164455,
+      "loss": 0.8244,
+      "step": 1085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017561940691060817,
+      "loss": 1.0197,
+      "step": 1086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017557051057103578,
+      "loss": 0.6547,
+      "step": 1087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001755215720702069,
+      "loss": 0.689,
+      "step": 1088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017547259143542467,
+      "loss": 0.8464,
+      "step": 1089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017542356869401567,
+      "loss": 1.2358,
+      "step": 1090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017537450387333,
+      "loss": 0.896,
+      "step": 1091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017532539700074127,
+      "loss": 0.8574,
+      "step": 1092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001752762481036464,
+      "loss": 0.857,
+      "step": 1093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017522705720946595,
+      "loss": 1.1983,
+      "step": 1094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001751778243456438,
+      "loss": 0.8727,
+      "step": 1095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017512854953964728,
+      "loss": 0.7436,
+      "step": 1096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017507923281896706,
+      "loss": 0.7684,
+      "step": 1097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017502987421111734,
+      "loss": 0.8123,
+      "step": 1098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017498047374363553,
+      "loss": 0.8877,
+      "step": 1099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017493103144408244,
+      "loss": 0.814,
+      "step": 1100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748815473400423,
+      "loss": 0.7948,
+      "step": 1101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017483202145912255,
+      "loss": 0.8498,
+      "step": 1102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017478245382895397,
+      "loss": 0.6163,
+      "step": 1103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017473284447719068,
+      "loss": 0.6478,
+      "step": 1104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017468319343151006,
+      "loss": 1.2627,
+      "step": 1105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017463350071961274,
+      "loss": 0.7449,
+      "step": 1106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017458376636922252,
+      "loss": 0.742,
+      "step": 1107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001745339904080866,
+      "loss": 0.7594,
+      "step": 1108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017448417286397517,
+      "loss": 0.7634,
+      "step": 1109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001744343137646819,
+      "loss": 0.8821,
+      "step": 1110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017438441313802341,
+      "loss": 0.8685,
+      "step": 1111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017433447101183953,
+      "loss": 0.6018,
+      "step": 1112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017428448741399336,
+      "loss": 1.3248,
+      "step": 1113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017423446237237102,
+      "loss": 0.7236,
+      "step": 1114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017418439591488177,
+      "loss": 1.084,
+      "step": 1115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017413428806945807,
+      "loss": 0.7876,
+      "step": 1116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017408413886405531,
+      "loss": 0.7795,
+      "step": 1117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017403394832665213,
+      "loss": 0.6834,
+      "step": 1118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017398371648525007,
+      "loss": 0.6042,
+      "step": 1119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739334433678738,
+      "loss": 0.6571,
+      "step": 1120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017388312900257105,
+      "loss": 0.7418,
+      "step": 1121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017383277341741243,
+      "loss": 0.7436,
+      "step": 1122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017378237664049173,
+      "loss": 1.0476,
+      "step": 1123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017373193869992556,
+      "loss": 0.6601,
+      "step": 1124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001736814596238536,
+      "loss": 0.8281,
+      "step": 1125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001736309394404384,
+      "loss": 0.7074,
+      "step": 1126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001735803781778655,
+      "loss": 0.7873,
+      "step": 1127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017352977586434335,
+      "loss": 1.0252,
+      "step": 1128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017347913252810333,
+      "loss": 0.6605,
+      "step": 1129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017342844819739957,
+      "loss": 0.8207,
+      "step": 1130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017337772290050926,
+      "loss": 0.7445,
+      "step": 1131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017332695666573232,
+      "loss": 0.6191,
+      "step": 1132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017327614952139156,
+      "loss": 0.7592,
+      "step": 1133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017322530149583264,
+      "loss": 0.8371,
+      "step": 1134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001731744126174239,
+      "loss": 0.6686,
+      "step": 1135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001731234829145566,
+      "loss": 0.7199,
+      "step": 1136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001730725124156448,
+      "loss": 0.749,
+      "step": 1137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017302150114912516,
+      "loss": 1.1516,
+      "step": 1138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017297044914345723,
+      "loss": 0.9373,
+      "step": 1139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017291935642712325,
+      "loss": 1.0609,
+      "step": 1140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017286822302862813,
+      "loss": 0.7351,
+      "step": 1141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017281704897649953,
+      "loss": 0.7428,
+      "step": 1142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017276583429928782,
+      "loss": 0.7945,
+      "step": 1143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017271457902556586,
+      "loss": 1.1341,
+      "step": 1144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001726632831839294,
+      "loss": 0.8453,
+      "step": 1145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017261194680299666,
+      "loss": 0.8274,
+      "step": 1146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017256056991140855,
+      "loss": 0.8166,
+      "step": 1147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017250915253782853,
+      "loss": 0.7769,
+      "step": 1148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017245769471094268,
+      "loss": 0.8077,
+      "step": 1149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017240619645945962,
+      "loss": 0.9531,
+      "step": 1150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017235465781211063,
+      "loss": 0.8138,
+      "step": 1151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017230307879764932,
+      "loss": 0.7116,
+      "step": 1152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000172251459444852,
+      "loss": 0.8119,
+      "step": 1153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017219979978251744,
+      "loss": 1.1044,
+      "step": 1154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017214809983946689,
+      "loss": 0.8437,
+      "step": 1155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017209635964454405,
+      "loss": 1.1665,
+      "step": 1156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017204457922661507,
+      "loss": 0.8641,
+      "step": 1157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017199275861456864,
+      "loss": 0.8545,
+      "step": 1158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017194089783731574,
+      "loss": 0.7653,
+      "step": 1159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017188899692378978,
+      "loss": 0.7888,
+      "step": 1160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001718370559029467,
+      "loss": 0.8215,
+      "step": 1161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017178507480376467,
+      "loss": 0.7499,
+      "step": 1162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017173305365524426,
+      "loss": 1.0713,
+      "step": 1163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017168099248640842,
+      "loss": 0.7618,
+      "step": 1164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017162889132630235,
+      "loss": 0.7571,
+      "step": 1165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017157675020399365,
+      "loss": 0.8323,
+      "step": 1166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017152456914857218,
+      "loss": 0.6701,
+      "step": 1167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017147234818915008,
+      "loss": 0.7981,
+      "step": 1168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001714200873548617,
+      "loss": 0.9859,
+      "step": 1169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001713677866748637,
+      "loss": 0.8093,
+      "step": 1170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017131544617833498,
+      "loss": 0.8685,
+      "step": 1171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001712630658944766,
+      "loss": 0.8973,
+      "step": 1172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001712106458525119,
+      "loss": 0.9637,
+      "step": 1173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017115818608168622,
+      "loss": 0.8861,
+      "step": 1174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017110568661126734,
+      "loss": 1.0755,
+      "step": 1175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017105314747054498,
+      "loss": 1.0714,
+      "step": 1176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000171000568688831,
+      "loss": 0.6956,
+      "step": 1177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017094795029545953,
+      "loss": 0.6567,
+      "step": 1178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001708952923197866,
+      "loss": 0.6914,
+      "step": 1179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017084259479119046,
+      "loss": 1.153,
+      "step": 1180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017078985773907147,
+      "loss": 0.9566,
+      "step": 1181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001707370811928518,
+      "loss": 0.8087,
+      "step": 1182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017068426518197593,
+      "loss": 0.728,
+      "step": 1183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001706314097359102,
+      "loss": 0.8632,
+      "step": 1184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000170578514884143,
+      "loss": 0.9656,
+      "step": 1185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017052558065618466,
+      "loss": 0.8409,
+      "step": 1186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017047260708156757,
+      "loss": 0.7354,
+      "step": 1187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000170419594189846,
+      "loss": 0.6723,
+      "step": 1188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017036654201059616,
+      "loss": 0.7283,
+      "step": 1189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017031345057341618,
+      "loss": 0.7662,
+      "step": 1190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001702603199079261,
+      "loss": 0.8887,
+      "step": 1191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017020715004376788,
+      "loss": 0.7521,
+      "step": 1192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001701539410106053,
+      "loss": 0.6605,
+      "step": 1193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000170100692838124,
+      "loss": 1.1209,
+      "step": 1194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017004740555603149,
+      "loss": 0.6323,
+      "step": 1195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016999407919405705,
+      "loss": 0.7898,
+      "step": 1196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001699407137819518,
+      "loss": 0.8385,
+      "step": 1197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016988730934948862,
+      "loss": 0.8345,
+      "step": 1198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016983386592646217,
+      "loss": 1.2248,
+      "step": 1199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001697803835426889,
+      "loss": 0.7007,
+      "step": 1200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016972686222800696,
+      "loss": 0.6496,
+      "step": 1201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016967330201227618,
+      "loss": 0.8306,
+      "step": 1202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001696197029253782,
+      "loss": 0.8774,
+      "step": 1203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016956606499721622,
+      "loss": 0.794,
+      "step": 1204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001695123882577152,
+      "loss": 0.8986,
+      "step": 1205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016945867273682178,
+      "loss": 0.8157,
+      "step": 1206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016940491846450408,
+      "loss": 0.7293,
+      "step": 1207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000169351125470752,
+      "loss": 1.1711,
+      "step": 1208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016929729378557698,
+      "loss": 1.178,
+      "step": 1209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016924342343901204,
+      "loss": 0.945,
+      "step": 1210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016918951446111174,
+      "loss": 0.8757,
+      "step": 1211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016913556688195235,
+      "loss": 0.7307,
+      "step": 1212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016908158073163145,
+      "loss": 1.2374,
+      "step": 1213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016902755604026833,
+      "loss": 0.8055,
+      "step": 1214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001689734928380036,
+      "loss": 0.637,
+      "step": 1215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016891939115499954,
+      "loss": 0.9276,
+      "step": 1216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016886525102143983,
+      "loss": 1.0233,
+      "step": 1217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001688110724675295,
+      "loss": 0.7553,
+      "step": 1218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016875685552349517,
+      "loss": 0.7383,
+      "step": 1219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016870260021958474,
+      "loss": 0.8091,
+      "step": 1220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016864830658606765,
+      "loss": 0.9609,
+      "step": 1221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016859397465323462,
+      "loss": 0.7546,
+      "step": 1222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001685396044513978,
+      "loss": 0.841,
+      "step": 1223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016848519601089063,
+      "loss": 1.0615,
+      "step": 1224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016843074936206796,
+      "loss": 0.8505,
+      "step": 1225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016837626453530583,
+      "loss": 0.7382,
+      "step": 1226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016832174156100176,
+      "loss": 0.8695,
+      "step": 1227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001682671804695744,
+      "loss": 0.7711,
+      "step": 1228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016821258129146374,
+      "loss": 1.1022,
+      "step": 1229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000168157944057131,
+      "loss": 0.795,
+      "step": 1230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001681032687970586,
+      "loss": 0.8947,
+      "step": 1231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016804855554175023,
+      "loss": 0.707,
+      "step": 1232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001679938043217308,
+      "loss": 0.761,
+      "step": 1233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016793901516754626,
+      "loss": 0.7434,
+      "step": 1234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016788418810976388,
+      "loss": 0.7306,
+      "step": 1235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016782932317897204,
+      "loss": 0.6983,
+      "step": 1236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016777442040578016,
+      "loss": 1.014,
+      "step": 1237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016771947982081889,
+      "loss": 0.8535,
+      "step": 1238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016766450145473986,
+      "loss": 0.7517,
+      "step": 1239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016760948533821592,
+      "loss": 0.6431,
+      "step": 1240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016755443150194088,
+      "loss": 0.9754,
+      "step": 1241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016749933997662958,
+      "loss": 0.7508,
+      "step": 1242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016744421079301796,
+      "loss": 1.0657,
+      "step": 1243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016738904398186288,
+      "loss": 0.7881,
+      "step": 1244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016733383957394233,
+      "loss": 1.1563,
+      "step": 1245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016727859760005511,
+      "loss": 0.7387,
+      "step": 1246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016722331809102111,
+      "loss": 0.7873,
+      "step": 1247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016716800107768109,
+      "loss": 0.7251,
+      "step": 1248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016711264659089681,
+      "loss": 0.7495,
+      "step": 1249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016705725466155077,
+      "loss": 0.734,
+      "step": 1250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001670018253205466,
+      "loss": 0.8924,
+      "step": 1251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001669463585988085,
+      "loss": 0.7084,
+      "step": 1252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016689085452728194,
+      "loss": 1.2171,
+      "step": 1253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016683531313693278,
+      "loss": 0.6227,
+      "step": 1254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016677973445874798,
+      "loss": 0.6979,
+      "step": 1255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016672411852373522,
+      "loss": 0.7538,
+      "step": 1256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000166668465362923,
+      "loss": 0.777,
+      "step": 1257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016661277500736052,
+      "loss": 1.102,
+      "step": 1258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016655704748811783,
+      "loss": 0.8362,
+      "step": 1259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001665012828362856,
+      "loss": 0.8481,
+      "step": 1260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001664454810829753,
+      "loss": 0.7781,
+      "step": 1261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001663896422593191,
+      "loss": 0.8093,
+      "step": 1262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016633376639646981,
+      "loss": 0.7489,
+      "step": 1263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001662778535256009,
+      "loss": 0.688,
+      "step": 1264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016622190367790652,
+      "loss": 0.8231,
+      "step": 1265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016616591688460144,
+      "loss": 1.0839,
+      "step": 1266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016610989317692102,
+      "loss": 0.7779,
+      "step": 1267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001660538325861213,
+      "loss": 0.766,
+      "step": 1268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016599773514347873,
+      "loss": 0.777,
+      "step": 1269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001659416008802905,
+      "loss": 0.7331,
+      "step": 1270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016588542982787422,
+      "loss": 1.1696,
+      "step": 1271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001658292220175681,
+      "loss": 0.7233,
+      "step": 1272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001657729774807308,
+      "loss": 1.2115,
+      "step": 1273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016571669624874152,
+      "loss": 0.7684,
+      "step": 1274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001656603783529999,
+      "loss": 0.8427,
+      "step": 1275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016560402382492602,
+      "loss": 0.8492,
+      "step": 1276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001655476326959605,
+      "loss": 0.6848,
+      "step": 1277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016549120499756418,
+      "loss": 0.7025,
+      "step": 1278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001654347407612185,
+      "loss": 0.7141,
+      "step": 1279
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016537824001842518,
+      "loss": 0.6646,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016532170280070634,
+      "loss": 0.6822,
+      "step": 1281
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016526512913960447,
+      "loss": 0.6465,
+      "step": 1282
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016520851906668234,
+      "loss": 0.7332,
+      "step": 1283
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016515187261352304,
+      "loss": 0.8378,
+      "step": 1284
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016509518981173003,
+      "loss": 0.7317,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016503847069292696,
+      "loss": 0.8755,
+      "step": 1286
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016498171528875777,
+      "loss": 0.7592,
+      "step": 1287
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016492492363088675,
+      "loss": 0.9662,
+      "step": 1288
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016486809575099818,
+      "loss": 1.2188,
+      "step": 1289
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016481123168079676,
+      "loss": 0.7209,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016475433145200728,
+      "loss": 0.7371,
+      "step": 1291
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016469739509637475,
+      "loss": 0.9152,
+      "step": 1292
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001646404226456643,
+      "loss": 1.2457,
+      "step": 1293
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016458341413166122,
+      "loss": 0.6817,
+      "step": 1294
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016452636958617093,
+      "loss": 0.8631,
+      "step": 1295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016446928904101892,
+      "loss": 0.6925,
+      "step": 1296
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001644121725280508,
+      "loss": 0.7308,
+      "step": 1297
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016435502007913216,
+      "loss": 0.9012,
+      "step": 1298
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001642978317261488,
+      "loss": 1.109,
+      "step": 1299
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016424060750100638,
+      "loss": 0.7394,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001641833474356307,
+      "loss": 0.6854,
+      "step": 1301
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016412605156196748,
+      "loss": 0.7256,
+      "step": 1302
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016406871991198247,
+      "loss": 0.7891,
+      "step": 1303
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016401135251766134,
+      "loss": 0.7263,
+      "step": 1304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016395394941100966,
+      "loss": 0.7569,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016389651062405304,
+      "loss": 0.7096,
+      "step": 1306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016383903618883694,
+      "loss": 0.7712,
+      "step": 1307
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016378152613742667,
+      "loss": 1.1087,
+      "step": 1308
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016372398050190748,
+      "loss": 1.1161,
+      "step": 1309
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001636663993143844,
+      "loss": 0.7769,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016360878260698235,
+      "loss": 0.7697,
+      "step": 1311
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001635511304118461,
+      "loss": 0.794,
+      "step": 1312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016349344276114005,
+      "loss": 0.7875,
+      "step": 1313
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016343571968704859,
+      "loss": 0.9212,
+      "step": 1314
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016337796122177574,
+      "loss": 0.8813,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016332016739754532,
+      "loss": 0.8694,
+      "step": 1316
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016326233824660088,
+      "loss": 0.7368,
+      "step": 1317
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016320447380120562,
+      "loss": 0.896,
+      "step": 1318
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016314657409364252,
+      "loss": 0.7308,
+      "step": 1319
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016308863915621413,
+      "loss": 0.8139,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016303066902124277,
+      "loss": 0.7579,
+      "step": 1321
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016297266372107032,
+      "loss": 0.7818,
+      "step": 1322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001629146232880583,
+      "loss": 0.7342,
+      "step": 1323
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001628565477545878,
+      "loss": 0.8335,
+      "step": 1324
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016279843715305952,
+      "loss": 0.7774,
+      "step": 1325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016274029151589373,
+      "loss": 0.7463,
+      "step": 1326
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016268211087553029,
+      "loss": 0.8322,
+      "step": 1327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016262389526442846,
+      "loss": 0.8269,
+      "step": 1328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016256564471506718,
+      "loss": 0.8845,
+      "step": 1329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016250735925994473,
+      "loss": 0.7703,
+      "step": 1330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001624490389315789,
+      "loss": 0.9478,
+      "step": 1331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016239068376250709,
+      "loss": 1.0622,
+      "step": 1332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016233229378528587,
+      "loss": 0.7906,
+      "step": 1333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016227386903249147,
+      "loss": 0.8459,
+      "step": 1334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016221540953671935,
+      "loss": 0.6515,
+      "step": 1335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001621569153305844,
+      "loss": 0.7436,
+      "step": 1336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162098386446721,
+      "loss": 1.1662,
+      "step": 1337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016203982291778277,
+      "loss": 1.0177,
+      "step": 1338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001619812247764426,
+      "loss": 1.081,
+      "step": 1339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016192259205539275,
+      "loss": 0.772,
+      "step": 1340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016186392478734484,
+      "loss": 0.8454,
+      "step": 1341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016180522300502967,
+      "loss": 0.7464,
+      "step": 1342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016174648674119733,
+      "loss": 0.6795,
+      "step": 1343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016168771602861716,
+      "loss": 1.033,
+      "step": 1344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001616289109000777,
+      "loss": 0.7961,
+      "step": 1345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016157007138838667,
+      "loss": 0.8435,
+      "step": 1346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016151119752637105,
+      "loss": 0.6263,
+      "step": 1347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001614522893468769,
+      "loss": 0.7673,
+      "step": 1348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016139334688276938,
+      "loss": 1.1573,
+      "step": 1349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016133437016693303,
+      "loss": 0.7089,
+      "step": 1350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016127535923227118,
+      "loss": 0.9058,
+      "step": 1351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016121631411170645,
+      "loss": 0.842,
+      "step": 1352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001611572348381805,
+      "loss": 0.8301,
+      "step": 1353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016109812144465398,
+      "loss": 0.833,
+      "step": 1354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016103897396410667,
+      "loss": 0.8095,
+      "step": 1355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016097979242953728,
+      "loss": 0.7296,
+      "step": 1356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016092057687396353,
+      "loss": 0.5905,
+      "step": 1357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001608613273304222,
+      "loss": 0.8206,
+      "step": 1358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016080204383196893,
+      "loss": 0.7934,
+      "step": 1359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001607427264116784,
+      "loss": 0.7155,
+      "step": 1360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001606833751026441,
+      "loss": 0.6779,
+      "step": 1361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016062398993797858,
+      "loss": 1.1252,
+      "step": 1362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001605645709508131,
+      "loss": 0.7764,
+      "step": 1363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016050511817429798,
+      "loss": 0.834,
+      "step": 1364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016044563164160222,
+      "loss": 1.0998,
+      "step": 1365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016038611138591372,
+      "loss": 0.7793,
+      "step": 1366
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016032655744043926,
+      "loss": 0.893,
+      "step": 1367
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001602669698384043,
+      "loss": 0.706,
+      "step": 1368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016020734861305319,
+      "loss": 0.6823,
+      "step": 1369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001601476937976489,
+      "loss": 0.8082,
+      "step": 1370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001600880054254733,
+      "loss": 0.7805,
+      "step": 1371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016002828352982683,
+      "loss": 0.6804,
+      "step": 1372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015996852814402875,
+      "loss": 0.6841,
+      "step": 1373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015990873930141702,
+      "loss": 0.755,
+      "step": 1374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015984891703534805,
+      "loss": 0.7169,
+      "step": 1375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015978906137919721,
+      "loss": 0.6714,
+      "step": 1376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015972917236635828,
+      "loss": 0.792,
+      "step": 1377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015966925003024364,
+      "loss": 0.8336,
+      "step": 1378
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015960929440428444,
+      "loss": 0.6639,
+      "step": 1379
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015954930552193028,
+      "loss": 0.8294,
+      "step": 1380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001594892834166493,
+      "loss": 0.9311,
+      "step": 1381
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001594292281219282,
+      "loss": 0.8043,
+      "step": 1382
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001593691396712722,
+      "loss": 0.8389,
+      "step": 1383
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015930901809820504,
+      "loss": 0.8919,
+      "step": 1384
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015924886343626893,
+      "loss": 0.7495,
+      "step": 1385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015918867571902448,
+      "loss": 0.8505,
+      "step": 1386
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001591284549800508,
+      "loss": 1.1558,
+      "step": 1387
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001590682012529455,
+      "loss": 0.9466,
+      "step": 1388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001590079145713244,
+      "loss": 0.7638,
+      "step": 1389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015894759496882184,
+      "loss": 0.6409,
+      "step": 1390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015888724247909054,
+      "loss": 0.7845,
+      "step": 1391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001588268571358015,
+      "loss": 0.7978,
+      "step": 1392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015876643897264406,
+      "loss": 0.9266,
+      "step": 1393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015870598802332592,
+      "loss": 1.2092,
+      "step": 1394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015864550432157307,
+      "loss": 0.6648,
+      "step": 1395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015858498790112963,
+      "loss": 1.1195,
+      "step": 1396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001585244387957582,
+      "loss": 0.7832,
+      "step": 1397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015846385703923943,
+      "loss": 0.7713,
+      "step": 1398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015840324266537233,
+      "loss": 0.7596,
+      "step": 1399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015834259570797395,
+      "loss": 0.6797,
+      "step": 1400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015828191620087965,
+      "loss": 1.068,
+      "step": 1401
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001582212041779429,
+      "loss": 0.6269,
+      "step": 1402
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015816045967303533,
+      "loss": 0.8136,
+      "step": 1403
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015809968272004663,
+      "loss": 1.1065,
+      "step": 1404
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001580388733528847,
+      "loss": 0.7888,
+      "step": 1405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001579780316054754,
+      "loss": 0.6451,
+      "step": 1406
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015791715751176276,
+      "loss": 1.1557,
+      "step": 1407
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015785625110570876,
+      "loss": 0.7709,
+      "step": 1408
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015779531242129355,
+      "loss": 0.6711,
+      "step": 1409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015773434149251515,
+      "loss": 0.8918,
+      "step": 1410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015767333835338962,
+      "loss": 0.8382,
+      "step": 1411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015761230303795097,
+      "loss": 0.8705,
+      "step": 1412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001575512355802512,
+      "loss": 0.7811,
+      "step": 1413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015749013601436027,
+      "loss": 0.718,
+      "step": 1414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015742900437436594,
+      "loss": 0.6653,
+      "step": 1415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015736784069437398,
+      "loss": 0.741,
+      "step": 1416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015730664500850794,
+      "loss": 0.7837,
+      "step": 1417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015724541735090926,
+      "loss": 0.6298,
+      "step": 1418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015718415775573728,
+      "loss": 0.666,
+      "step": 1419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015712286625716912,
+      "loss": 0.8654,
+      "step": 1420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015706154288939965,
+      "loss": 1.1059,
+      "step": 1421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015700018768664156,
+      "loss": 0.796,
+      "step": 1422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015693880068312526,
+      "loss": 0.6837,
+      "step": 1423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015687738191309906,
+      "loss": 0.8152,
+      "step": 1424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015681593141082874,
+      "loss": 0.8757,
+      "step": 1425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000156754449210598,
+      "loss": 0.7891,
+      "step": 1426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001566929353467081,
+      "loss": 0.7982,
+      "step": 1427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015663138985347802,
+      "loss": 0.6741,
+      "step": 1428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015656981276524436,
+      "loss": 0.8701,
+      "step": 1429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015650820411636137,
+      "loss": 1.1657,
+      "step": 1430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015644656394120092,
+      "loss": 0.7298,
+      "step": 1431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001563848922741524,
+      "loss": 0.7046,
+      "step": 1432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015632318914962285,
+      "loss": 0.7787,
+      "step": 1433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015626145460203683,
+      "loss": 1.1264,
+      "step": 1434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001561996886658364,
+      "loss": 0.7404,
+      "step": 1435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015613789137548117,
+      "loss": 0.7549,
+      "step": 1436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015607606276544825,
+      "loss": 1.1564,
+      "step": 1437
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015601420287023217,
+      "loss": 0.9689,
+      "step": 1438
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015595231172434496,
+      "loss": 0.6903,
+      "step": 1439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015589038936231607,
+      "loss": 1.0629,
+      "step": 1440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001558284358186924,
+      "loss": 0.9363,
+      "step": 1441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001557664511280382,
+      "loss": 0.8291,
+      "step": 1442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015570443532493512,
+      "loss": 0.7134,
+      "step": 1443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015564238844398214,
+      "loss": 0.8021,
+      "step": 1444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001555803105197956,
+      "loss": 0.6942,
+      "step": 1445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015551820158700918,
+      "loss": 0.8319,
+      "step": 1446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015545606168027381,
+      "loss": 0.82,
+      "step": 1447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015539389083425777,
+      "loss": 0.9128,
+      "step": 1448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015533168908364651,
+      "loss": 0.5676,
+      "step": 1449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015526945646314287,
+      "loss": 0.8987,
+      "step": 1450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015520719300746668,
+      "loss": 0.7113,
+      "step": 1451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015514489875135515,
+      "loss": 0.7513,
+      "step": 1452
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001550825737295627,
+      "loss": 0.7862,
+      "step": 1453
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001550202179768608,
+      "loss": 0.764,
+      "step": 1454
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001549578315280381,
+      "loss": 0.7513,
+      "step": 1455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015489541441790041,
+      "loss": 0.6705,
+      "step": 1456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015483296668127058,
+      "loss": 0.7029,
+      "step": 1457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015477048835298867,
+      "loss": 1.1344,
+      "step": 1458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015470797946791164,
+      "loss": 1.1088,
+      "step": 1459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015464544006091365,
+      "loss": 0.7481,
+      "step": 1460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015458287016688576,
+      "loss": 0.6969,
+      "step": 1461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001545202698207362,
+      "loss": 1.1885,
+      "step": 1462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015445763905738998,
+      "loss": 1.0946,
+      "step": 1463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015439497791178925,
+      "loss": 0.8163,
+      "step": 1464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015433228641889302,
+      "loss": 0.9288,
+      "step": 1465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015426956461367733,
+      "loss": 0.6443,
+      "step": 1466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015420681253113498,
+      "loss": 1.0396,
+      "step": 1467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015414403020627576,
+      "loss": 0.6928,
+      "step": 1468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001540812176741264,
+      "loss": 1.2441,
+      "step": 1469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001540183749697303,
+      "loss": 1.1232,
+      "step": 1470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001539555021281479,
+      "loss": 0.7034,
+      "step": 1471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001538925991844562,
+      "loss": 0.8273,
+      "step": 1472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015382966617374932,
+      "loss": 0.7665,
+      "step": 1473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015376670313113785,
+      "loss": 1.1153,
+      "step": 1474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015370371009174925,
+      "loss": 0.6727,
+      "step": 1475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015364068709072782,
+      "loss": 0.608,
+      "step": 1476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015357763416323444,
+      "loss": 0.7372,
+      "step": 1477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001535145513444467,
+      "loss": 0.6697,
+      "step": 1478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015345143866955896,
+      "loss": 0.6863,
+      "step": 1479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001533882961737821,
+      "loss": 0.7554,
+      "step": 1480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015332512389234376,
+      "loss": 0.9858,
+      "step": 1481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015326192186048814,
+      "loss": 0.7291,
+      "step": 1482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015319869011347606,
+      "loss": 0.8398,
+      "step": 1483
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001531354286865848,
+      "loss": 0.9041,
+      "step": 1484
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001530721376151085,
+      "loss": 0.7609,
+      "step": 1485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015300881693435745,
+      "loss": 1.2032,
+      "step": 1486
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001529454666796587,
+      "loss": 1.1469,
+      "step": 1487
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015288208688635585,
+      "loss": 0.8269,
+      "step": 1488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001528186775898088,
+      "loss": 0.9444,
+      "step": 1489
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015275523882539392,
+      "loss": 0.7951,
+      "step": 1490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015269177062850425,
+      "loss": 0.9445,
+      "step": 1491
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015262827303454897,
+      "loss": 0.7803,
+      "step": 1492
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015256474607895382,
+      "loss": 1.1368,
+      "step": 1493
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015250118979716088,
+      "loss": 1.0131,
+      "step": 1494
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001524376042246286,
+      "loss": 0.7328,
+      "step": 1495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015237398939683172,
+      "loss": 0.7857,
+      "step": 1496
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015231034534926146,
+      "loss": 0.9951,
+      "step": 1497
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015224667211742512,
+      "loss": 0.7919,
+      "step": 1498
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015218296973684638,
+      "loss": 0.668,
+      "step": 1499
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015211923824306532,
+      "loss": 0.6087,
+      "step": 1500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015205547767163798,
+      "loss": 0.8868,
+      "step": 1501
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015199168805813692,
+      "loss": 0.9968,
+      "step": 1502
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015192786943815062,
+      "loss": 0.7753,
+      "step": 1503
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000151864021847284,
+      "loss": 0.6523,
+      "step": 1504
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015180014532115797,
+      "loss": 0.7597,
+      "step": 1505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001517362398954097,
+      "loss": 0.6545,
+      "step": 1506
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001516723056056923,
+      "loss": 0.8231,
+      "step": 1507
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015160834248767525,
+      "loss": 0.7709,
+      "step": 1508
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001515443505770439,
+      "loss": 1.0117,
+      "step": 1509
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001514803299094997,
+      "loss": 0.7166,
+      "step": 1510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015141628052076024,
+      "loss": 1.0412,
+      "step": 1511
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015135220244655905,
+      "loss": 0.791,
+      "step": 1512
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015128809572264565,
+      "loss": 0.9898,
+      "step": 1513
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015122396038478567,
+      "loss": 1.0854,
+      "step": 1514
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001511597964687605,
+      "loss": 0.6639,
+      "step": 1515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015109560401036767,
+      "loss": 0.658,
+      "step": 1516
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001510313830454205,
+      "loss": 0.736,
+      "step": 1517
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015096713360974827,
+      "loss": 0.7042,
+      "step": 1518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015090285573919618,
+      "loss": 0.8601,
+      "step": 1519
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015083854946962524,
+      "loss": 1.0543,
+      "step": 1520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001507742148369123,
+      "loss": 0.9105,
+      "step": 1521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015070985187695004,
+      "loss": 0.8109,
+      "step": 1522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015064546062564695,
+      "loss": 0.8614,
+      "step": 1523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015058104111892736,
+      "loss": 0.7089,
+      "step": 1524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015051659339273123,
+      "loss": 1.2022,
+      "step": 1525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015045211748301445,
+      "loss": 0.7487,
+      "step": 1526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015038761342574844,
+      "loss": 0.9716,
+      "step": 1527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015032308125692047,
+      "loss": 0.9616,
+      "step": 1528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001502585210125334,
+      "loss": 0.8871,
+      "step": 1529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001501939327286058,
+      "loss": 0.8378,
+      "step": 1530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015012931644117186,
+      "loss": 7.8827,
+      "step": 1531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015006467218628145,
+      "loss": 0.8948,
+      "step": 1532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.1433,
+      "step": 1533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001499352999184085,
+      "loss": 0.7745,
+      "step": 1534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001498705719776035,
+      "loss": 0.7307,
+      "step": 1535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014980581621369717,
+      "loss": 1.2757,
+      "step": 1536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014974103266281711,
+      "loss": 1.5306,
+      "step": 1537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014967622136110648,
+      "loss": 3.2369,
+      "step": 1538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014961138234472396,
+      "loss": 2.42,
+      "step": 1539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014954651564984354,
+      "loss": 5.4162,
+      "step": 1540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001494816213126548,
+      "loss": 3.6403,
+      "step": 1541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014941669936936272,
+      "loss": 0.9805,
+      "step": 1542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014935174985618768,
+      "loss": 6.4043,
+      "step": 1543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001492867728093653,
+      "loss": 1.5563,
+      "step": 1544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014922176826514676,
+      "loss": 6.0213,
+      "step": 1545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014915673625979847,
+      "loss": 1.403,
+      "step": 1546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014909167682960222,
+      "loss": 4.8716,
+      "step": 1547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014902659001085503,
+      "loss": 2.757,
+      "step": 1548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014896147583986927,
+      "loss": 2.3939,
+      "step": 1549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014889633435297253,
+      "loss": 0.9625,
+      "step": 1550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014883116558650765,
+      "loss": 1.7601,
+      "step": 1551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001487659695768327,
+      "loss": 1.2222,
+      "step": 1552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014870074636032091,
+      "loss": 2.7204,
+      "step": 1553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014863549597336073,
+      "loss": 1.2115,
+      "step": 1554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001485702184523558,
+      "loss": 1.5653,
+      "step": 1555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014850491383372477,
+      "loss": 1.7238,
+      "step": 1556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014843958215390158,
+      "loss": 8.2243,
+      "step": 1557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001483742234493351,
+      "loss": 6.2993,
+      "step": 1558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001483088377564894,
+      "loss": 6.8062,
+      "step": 1559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014824342511184354,
+      "loss": 6.8574,
+      "step": 1560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014817798555189166,
+      "loss": 6.6152,
+      "step": 1561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014811251911314287,
+      "loss": 6.8459,
+      "step": 1562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014804702583212135,
+      "loss": 6.8507,
+      "step": 1563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014798150574536618,
+      "loss": 6.3845,
+      "step": 1564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001479159588894314,
+      "loss": 6.6917,
+      "step": 1565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014785038530088607,
+      "loss": 6.4867,
+      "step": 1566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014778478501631406,
+      "loss": 6.4823,
+      "step": 1567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014771915807231412,
+      "loss": 6.4259,
+      "step": 1568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014765350450550004,
+      "loss": 6.476,
+      "step": 1569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001475878243525003,
+      "loss": 6.5804,
+      "step": 1570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014752211764995824,
+      "loss": 6.5111,
+      "step": 1571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014745638443453205,
+      "loss": 6.7354,
+      "step": 1572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014739062474289474,
+      "loss": 6.85,
+      "step": 1573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014732483861173398,
+      "loss": 12.6848,
+      "step": 1574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014725902607775235,
+      "loss": 16.019,
+      "step": 1575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014719318717766697,
+      "loss": 12.8834,
+      "step": 1576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014712732194820983,
+      "loss": 10.9559,
+      "step": 1577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014706143042612755,
+      "loss": 12.1284,
+      "step": 1578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014699551264818144,
+      "loss": 9.0057,
+      "step": 1579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014692956865114737,
+      "loss": 9.4675,
+      "step": 1580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014686359847181595,
+      "loss": 8.459,
+      "step": 1581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014679760214699234,
+      "loss": 7.9017,
+      "step": 1582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001467315797134963,
+      "loss": 8.8052,
+      "step": 1583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014666553120816215,
+      "loss": 8.1576,
+      "step": 1584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014659945666783877,
+      "loss": 7.7302,
+      "step": 1585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014653335612938956,
+      "loss": 7.606,
+      "step": 1586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001464672296296924,
+      "loss": 7.4739,
+      "step": 1587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014640107720563972,
+      "loss": 7.189,
+      "step": 1588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001463348988941383,
+      "loss": 7.3493,
+      "step": 1589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001462686947321095,
+      "loss": 6.9407,
+      "step": 1590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014620246475648897,
+      "loss": 7.4235,
+      "step": 1591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014613620900422687,
+      "loss": 7.1747,
+      "step": 1592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014606992751228765,
+      "loss": 6.9437,
+      "step": 1593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014600362031765024,
+      "loss": 7.0596,
+      "step": 1594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014593728745730775,
+      "loss": 7.4948,
+      "step": 1595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014587092896826775,
+      "loss": 7.4567,
+      "step": 1596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014580454488755202,
+      "loss": 6.7082,
+      "step": 1597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014573813525219667,
+      "loss": 7.2623,
+      "step": 1598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014567170009925203,
+      "loss": 7.3477,
+      "step": 1599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001456052394657827,
+      "loss": 6.8472,
+      "step": 1600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001455387533888675,
+      "loss": 6.8933,
+      "step": 1601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001454722419055994,
+      "loss": 6.7426,
+      "step": 1602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014540570505308556,
+      "loss": 6.5627,
+      "step": 1603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001453391428684473,
+      "loss": 6.5685,
+      "step": 1604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014527255538882009,
+      "loss": 6.59,
+      "step": 1605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001452059426513535,
+      "loss": 6.3929,
+      "step": 1606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014513930469321117,
+      "loss": 7.1671,
+      "step": 1607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014507264155157084,
+      "loss": 6.5558,
+      "step": 1608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001450059532636243,
+      "loss": 6.58,
+      "step": 1609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014493923986657738,
+      "loss": 6.5764,
+      "step": 1610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014487250139764984,
+      "loss": 6.3884,
+      "step": 1611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014480573789407556,
+      "loss": 6.5248,
+      "step": 1612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014473894939310224,
+      "loss": 6.5298,
+      "step": 1613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014467213593199161,
+      "loss": 6.5084,
+      "step": 1614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014460529754801936,
+      "loss": 6.5015,
+      "step": 1615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014453843427847502,
+      "loss": 6.4341,
+      "step": 1616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014447154616066202,
+      "loss": 6.3925,
+      "step": 1617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001444046332318976,
+      "loss": 6.4818,
+      "step": 1618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000144337695529513,
+      "loss": 6.4613,
+      "step": 1619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014427073309085308,
+      "loss": 6.6576,
+      "step": 1620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001442037459532767,
+      "loss": 6.4058,
+      "step": 1621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014413673415415628,
+      "loss": 6.488,
+      "step": 1622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014406969773087823,
+      "loss": 6.5194,
+      "step": 1623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014400263672084256,
+      "loss": 6.4882,
+      "step": 1624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000143935551161463,
+      "loss": 6.6054,
+      "step": 1625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014386844109016698,
+      "loss": 6.1953,
+      "step": 1626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014380130654439565,
+      "loss": 6.6525,
+      "step": 1627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014373414756160377,
+      "loss": 6.4639,
+      "step": 1628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001436669641792598,
+      "loss": 6.5639,
+      "step": 1629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014359975643484573,
+      "loss": 6.4019,
+      "step": 1630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001435325243658572,
+      "loss": 6.2497,
+      "step": 1631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014346526800980335,
+      "loss": 6.4747,
+      "step": 1632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014339798740420698,
+      "loss": 6.0207,
+      "step": 1633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014333068258660432,
+      "loss": 6.4693,
+      "step": 1634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014326335359454514,
+      "loss": 6.2633,
+      "step": 1635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014319600046559274,
+      "loss": 6.4282,
+      "step": 1636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001431286232373238,
+      "loss": 6.5141,
+      "step": 1637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001430612219473285,
+      "loss": 6.4598,
+      "step": 1638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014299379663321052,
+      "loss": 6.2354,
+      "step": 1639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014292634733258672,
+      "loss": 6.4186,
+      "step": 1640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014285887408308761,
+      "loss": 6.6149,
+      "step": 1641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014279137692235685,
+      "loss": 6.2925,
+      "step": 1642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014272385588805158,
+      "loss": 6.4493,
+      "step": 1643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014265631101784216,
+      "loss": 6.1315,
+      "step": 1644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014258874234941233,
+      "loss": 6.69,
+      "step": 1645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014252114992045905,
+      "loss": 6.5247,
+      "step": 1646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014245353376869256,
+      "loss": 6.48,
+      "step": 1647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014238589393183632,
+      "loss": 6.4322,
+      "step": 1648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014231823044762702,
+      "loss": 6.4262,
+      "step": 1649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014225054335381454,
+      "loss": 6.5269,
+      "step": 1650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014218283268816191,
+      "loss": 6.6034,
+      "step": 1651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001421150984884454,
+      "loss": 6.296,
+      "step": 1652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014204734079245425,
+      "loss": 6.4431,
+      "step": 1653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014197955963799093,
+      "loss": 6.4739,
+      "step": 1654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000141911755062871,
+      "loss": 6.5614,
+      "step": 1655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000141843927104923,
+      "loss": 6.1447,
+      "step": 1656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014177607580198863,
+      "loss": 6.6652,
+      "step": 1657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001417082011919225,
+      "loss": 6.268,
+      "step": 1658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014164030331259228,
+      "loss": 6.5092,
+      "step": 1659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014157238220187867,
+      "loss": 6.2098,
+      "step": 1660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001415044378976752,
+      "loss": 6.4735,
+      "step": 1661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014143647043788848,
+      "loss": 6.2498,
+      "step": 1662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014136847986043796,
+      "loss": 6.4193,
+      "step": 1663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014130046620325598,
+      "loss": 6.253,
+      "step": 1664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001412324295042878,
+      "loss": 6.4437,
+      "step": 1665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014116436980149156,
+      "loss": 6.7946,
+      "step": 1666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001410962871328381,
+      "loss": 6.419,
+      "step": 1667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014102818153631126,
+      "loss": 6.3317,
+      "step": 1668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014096005304990746,
+      "loss": 6.3917,
+      "step": 1669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014089190171163613,
+      "loss": 6.5424,
+      "step": 1670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014082372755951927,
+      "loss": 6.2093,
+      "step": 1671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014075553063159165,
+      "loss": 6.3339,
+      "step": 1672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014068731096590072,
+      "loss": 6.3484,
+      "step": 1673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014061906860050676,
+      "loss": 6.4111,
+      "step": 1674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014055080357348252,
+      "loss": 6.8257,
+      "step": 1675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001404825159229135,
+      "loss": 6.3456,
+      "step": 1676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014041420568689783,
+      "loss": 6.288,
+      "step": 1677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001403458729035462,
+      "loss": 6.2676,
+      "step": 1678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001402775176109819,
+      "loss": 6.2699,
+      "step": 1679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014020913984734076,
+      "loss": 6.373,
+      "step": 1680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014014073965077116,
+      "loss": 6.5894,
+      "step": 1681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014007231705943403,
+      "loss": 6.1406,
+      "step": 1682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014000387211150268,
+      "loss": 6.5136,
+      "step": 1683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013993540484516306,
+      "loss": 6.2389,
+      "step": 1684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001398669152986134,
+      "loss": 6.0989,
+      "step": 1685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001397984035100645,
+      "loss": 6.1335,
+      "step": 1686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013972986951773947,
+      "loss": 6.3129,
+      "step": 1687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001396613133598739,
+      "loss": 6.4895,
+      "step": 1688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013959273507471563,
+      "loss": 6.0957,
+      "step": 1689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000139524134700525,
+      "loss": 6.2414,
+      "step": 1690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001394555122755745,
+      "loss": 6.1398,
+      "step": 1691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013938686783814905,
+      "loss": 6.3526,
+      "step": 1692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013931820142654578,
+      "loss": 6.4797,
+      "step": 1693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013924951307907416,
+      "loss": 6.3098,
+      "step": 1694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013918080283405578,
+      "loss": 6.4593,
+      "step": 1695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013911207072982455,
+      "loss": 6.3346,
+      "step": 1696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013904331680472654,
+      "loss": 6.3288,
+      "step": 1697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013897454109711992,
+      "loss": 6.2668,
+      "step": 1698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013890574364537522,
+      "loss": 6.5235,
+      "step": 1699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013883692448787483,
+      "loss": 6.3011,
+      "step": 1700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013876808366301347,
+      "loss": 6.3974,
+      "step": 1701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013869922120919782,
+      "loss": 6.3371,
+      "step": 1702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001386303371648467,
+      "loss": 6.4275,
+      "step": 1703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013856143156839093,
+      "loss": 6.5808,
+      "step": 1704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013849250445827335,
+      "loss": 6.2704,
+      "step": 1705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013842355587294884,
+      "loss": 6.3062,
+      "step": 1706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013835458585088428,
+      "loss": 6.387,
+      "step": 1707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013828559443055841,
+      "loss": 6.2549,
+      "step": 1708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000138216581650462,
+      "loss": 6.5657,
+      "step": 1709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001381475475490977,
+      "loss": 6.4466,
+      "step": 1710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001380784921649801,
+      "loss": 6.2688,
+      "step": 1711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001380094155366356,
+      "loss": 6.3419,
+      "step": 1712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013794031770260245,
+      "loss": 6.6256,
+      "step": 1713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013787119870143077,
+      "loss": 6.3281,
+      "step": 1714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013780205857168254,
+      "loss": 6.2076,
+      "step": 1715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001377328973519314,
+      "loss": 6.206,
+      "step": 1716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013766371508076289,
+      "loss": 6.3082,
+      "step": 1717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013759451179677412,
+      "loss": 6.4032,
+      "step": 1718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013752528753857414,
+      "loss": 6.4992,
+      "step": 1719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013745604234478352,
+      "loss": 6.3242,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013738677625403463,
+      "loss": 6.358,
+      "step": 1721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001373174893049714,
+      "loss": 6.4612,
+      "step": 1722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013724818153624955,
+      "loss": 6.2113,
+      "step": 1723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001371788529865362,
+      "loss": 6.1986,
+      "step": 1724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001371095036945102,
+      "loss": 6.3698,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013704013369886202,
+      "loss": 6.3267,
+      "step": 1726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013697074303829356,
+      "loss": 6.1734,
+      "step": 1727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013690133175151827,
+      "loss": 6.3958,
+      "step": 1728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001368318998772612,
+      "loss": 6.2618,
+      "step": 1729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001367624474542588,
+      "loss": 6.3302,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013669297452125902,
+      "loss": 6.4165,
+      "step": 1731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013662348111702116,
+      "loss": 6.4292,
+      "step": 1732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013655396728031613,
+      "loss": 6.288,
+      "step": 1733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013648443304992605,
+      "loss": 6.1921,
+      "step": 1734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013641487846464457,
+      "loss": 6.101,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001363453035632766,
+      "loss": 6.2503,
+      "step": 1736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013627570838463838,
+      "loss": 6.2244,
+      "step": 1737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013620609296755747,
+      "loss": 6.2938,
+      "step": 1738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001361364573508728,
+      "loss": 6.5181,
+      "step": 1739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001360668015734345,
+      "loss": 6.1849,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013599712567410397,
+      "loss": 6.1714,
+      "step": 1741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001359274296917538,
+      "loss": 6.2501,
+      "step": 1742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001358577136652678,
+      "loss": 6.2268,
+      "step": 1743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013578797763354102,
+      "loss": 6.2704,
+      "step": 1744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013571822163547959,
+      "loss": 6.1218,
+      "step": 1745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013564844571000082,
+      "loss": 6.4821,
+      "step": 1746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013557864989603312,
+      "loss": 6.2434,
+      "step": 1747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013550883423251602,
+      "loss": 5.9646,
+      "step": 1748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001354389987584001,
+      "loss": 6.0927,
+      "step": 1749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013536914351264706,
+      "loss": 6.2453,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013529926853422951,
+      "loss": 6.3013,
+      "step": 1751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001352293738621312,
+      "loss": 6.4283,
+      "step": 1752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001351594595353467,
+      "loss": 6.2198,
+      "step": 1753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013508952559288174,
+      "loss": 6.2004,
+      "step": 1754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001350195720737528,
+      "loss": 6.2782,
+      "step": 1755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001349495990169875,
+      "loss": 6.2479,
+      "step": 1756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013487960646162413,
+      "loss": 6.405,
+      "step": 1757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001348095944467121,
+      "loss": 6.2503,
+      "step": 1758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013473956301131144,
+      "loss": 6.264,
+      "step": 1759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013466951219449317,
+      "loss": 6.2464,
+      "step": 1760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013459944203533903,
+      "loss": 6.0091,
+      "step": 1761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013452935257294167,
+      "loss": 6.0367,
+      "step": 1762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013445924384640435,
+      "loss": 6.0201,
+      "step": 1763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013438911589484125,
+      "loss": 6.1824,
+      "step": 1764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013431896875737716,
+      "loss": 6.1951,
+      "step": 1765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013424880247314756,
+      "loss": 6.3949,
+      "step": 1766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013417861708129867,
+      "loss": 6.051,
+      "step": 1767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013410841262098737,
+      "loss": 6.0454,
+      "step": 1768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013403818913138116,
+      "loss": 6.1892,
+      "step": 1769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013396794665165819,
+      "loss": 6.2491,
+      "step": 1770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013389768522100714,
+      "loss": 6.1155,
+      "step": 1771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001338274048786273,
+      "loss": 6.0341,
+      "step": 1772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013375710566372853,
+      "loss": 6.0171,
+      "step": 1773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013368678761553117,
+      "loss": 6.3091,
+      "step": 1774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001336164507732661,
+      "loss": 6.0905,
+      "step": 1775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001335460951761747,
+      "loss": 5.7609,
+      "step": 1776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001334757208635088,
+      "loss": 6.061,
+      "step": 1777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013340532787453063,
+      "loss": 6.0727,
+      "step": 1778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013333491624851291,
+      "loss": 6.0051,
+      "step": 1779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001332644860247387,
+      "loss": 6.2068,
+      "step": 1780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013319403724250145,
+      "loss": 6.2266,
+      "step": 1781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000133123569941105,
+      "loss": 6.3613,
+      "step": 1782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001330530841598635,
+      "loss": 5.951,
+      "step": 1783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013298257993810136,
+      "loss": 6.2434,
+      "step": 1784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013291205731515333,
+      "loss": 5.9283,
+      "step": 1785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013284151633036445,
+      "loss": 6.0682,
+      "step": 1786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013277095702309,
+      "loss": 6.0069,
+      "step": 1787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013270037943269536,
+      "loss": 6.1625,
+      "step": 1788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001326297835985563,
+      "loss": 6.1059,
+      "step": 1789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001325591695600586,
+      "loss": 5.7657,
+      "step": 1790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013248853735659834,
+      "loss": 5.7797,
+      "step": 1791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001324178870275816,
+      "loss": 5.82,
+      "step": 1792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013234721861242468,
+      "loss": 5.8784,
+      "step": 1793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013227653215055392,
+      "loss": 6.3732,
+      "step": 1794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001322058276814057,
+      "loss": 5.7555,
+      "step": 1795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001321351052444265,
+      "loss": 5.8261,
+      "step": 1796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013206436487907288,
+      "loss": 5.7316,
+      "step": 1797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013199360662481118,
+      "loss": 5.8632,
+      "step": 1798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013192283052111805,
+      "loss": 6.0123,
+      "step": 1799
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013185203660747975,
+      "loss": 5.7268,
+      "step": 1800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013178122492339274,
+      "loss": 5.8105,
+      "step": 1801
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013171039550836325,
+      "loss": 5.8517,
+      "step": 1802
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013163954840190748,
+      "loss": 6.1373,
+      "step": 1803
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013156868364355145,
+      "loss": 6.1187,
+      "step": 1804
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013149780127283108,
+      "loss": 6.1972,
+      "step": 1805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013142690132929204,
+      "loss": 6.1036,
+      "step": 1806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001313559838524899,
+      "loss": 5.9634,
+      "step": 1807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001312850488819899,
+      "loss": 6.1682,
+      "step": 1808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013121409645736718,
+      "loss": 5.7636,
+      "step": 1809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013114312661820647,
+      "loss": 5.7283,
+      "step": 1810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013107213940410228,
+      "loss": 5.7696,
+      "step": 1811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013100113485465883,
+      "loss": 5.8777,
+      "step": 1812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013093011300949005,
+      "loss": 6.1405,
+      "step": 1813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013085907390821938,
+      "loss": 5.819,
+      "step": 1814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013078801759048005,
+      "loss": 5.7884,
+      "step": 1815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013071694409591477,
+      "loss": 5.9401,
+      "step": 1816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013064585346417594,
+      "loss": 5.9613,
+      "step": 1817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001305747457349254,
+      "loss": 5.7591,
+      "step": 1818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013050362094783462,
+      "loss": 5.7406,
+      "step": 1819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013043247914258456,
+      "loss": 6.4132,
+      "step": 1820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001303613203588657,
+      "loss": 6.2558,
+      "step": 1821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001302901446363779,
+      "loss": 5.622,
+      "step": 1822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001302189520148306,
+      "loss": 6.097,
+      "step": 1823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013014774253394262,
+      "loss": 5.795,
+      "step": 1824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013007651623344208,
+      "loss": 6.0564,
+      "step": 1825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013000527315306667,
+      "loss": 5.7375,
+      "step": 1826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012993401333256328,
+      "loss": 5.7209,
+      "step": 1827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012986273681168823,
+      "loss": 5.7938,
+      "step": 1828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012979144363020715,
+      "loss": 5.7286,
+      "step": 1829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012972013382789494,
+      "loss": 5.6946,
+      "step": 1830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012964880744453577,
+      "loss": 5.791,
+      "step": 1831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012957746451992307,
+      "loss": 5.3453,
+      "step": 1832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012950610509385947,
+      "loss": 5.7815,
+      "step": 1833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012943472920615693,
+      "loss": 5.8071,
+      "step": 1834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012936333689663642,
+      "loss": 5.9901,
+      "step": 1835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012929192820512818,
+      "loss": 5.7457,
+      "step": 1836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012922050317147154,
+      "loss": 5.7468,
+      "step": 1837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012914906183551499,
+      "loss": 5.6209,
+      "step": 1838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001290776042371161,
+      "loss": 6.0,
+      "step": 1839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001290061304161415,
+      "loss": 5.9195,
+      "step": 1840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012893464041246683,
+      "loss": 5.8736,
+      "step": 1841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012886313426597692,
+      "loss": 5.9502,
+      "step": 1842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012879161201656536,
+      "loss": 5.7172,
+      "step": 1843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012872007370413497,
+      "loss": 5.7135,
+      "step": 1844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012864851936859734,
+      "loss": 5.8102,
+      "step": 1845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012857694904987312,
+      "loss": 5.546,
+      "step": 1846
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001285053627878918,
+      "loss": 5.9395,
+      "step": 1847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012843376062259186,
+      "loss": 6.1356,
+      "step": 1848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001283621425939205,
+      "loss": 6.2066,
+      "step": 1849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012829050874183396,
+      "loss": 5.9948,
+      "step": 1850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012821885910629713,
+      "loss": 6.061,
+      "step": 1851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012814719372728384,
+      "loss": 5.8723,
+      "step": 1852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012807551264477663,
+      "loss": 6.3026,
+      "step": 1853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001280038158987668,
+      "loss": 5.8661,
+      "step": 1854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012793210352925443,
+      "loss": 5.7865,
+      "step": 1855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012786037557624834,
+      "loss": 6.112,
+      "step": 1856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001277886320797659,
+      "loss": 5.9369,
+      "step": 1857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012771687307983333,
+      "loss": 6.5297,
+      "step": 1858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012764509861648538,
+      "loss": 5.9764,
+      "step": 1859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012757330872976551,
+      "loss": 5.8026,
+      "step": 1860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001275015034597257,
+      "loss": 6.0599,
+      "step": 1861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012742968284642662,
+      "loss": 5.8627,
+      "step": 1862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012735784692993732,
+      "loss": 5.8837,
+      "step": 1863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001272859957503356,
+      "loss": 5.7127,
+      "step": 1864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012721412934770765,
+      "loss": 6.0241,
+      "step": 1865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012714224776214824,
+      "loss": 5.7936,
+      "step": 1866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012707035103376042,
+      "loss": 5.9739,
+      "step": 1867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012699843920265593,
+      "loss": 5.739,
+      "step": 1868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012692651230895478,
+      "loss": 6.008,
+      "step": 1869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012685457039278544,
+      "loss": 5.9717,
+      "step": 1870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012678261349428479,
+      "loss": 6.0103,
+      "step": 1871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012671064165359794,
+      "loss": 5.7655,
+      "step": 1872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001266386549108785,
+      "loss": 5.5927,
+      "step": 1873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001265666533062883,
+      "loss": 5.819,
+      "step": 1874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001264946368799975,
+      "loss": 5.6942,
+      "step": 1875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012642260567218445,
+      "loss": 6.3965,
+      "step": 1876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012635055972303589,
+      "loss": 5.6408,
+      "step": 1877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012627849907274663,
+      "loss": 5.8059,
+      "step": 1878
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012620642376151982,
+      "loss": 6.0999,
+      "step": 1879
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001261343338295667,
+      "loss": 5.563,
+      "step": 1880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012606222931710662,
+      "loss": 5.6492,
+      "step": 1881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012599011026436722,
+      "loss": 5.6152,
+      "step": 1882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012591797671158416,
+      "loss": 5.9175,
+      "step": 1883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012584582869900116,
+      "loss": 5.7753,
+      "step": 1884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012577366626687005,
+      "loss": 5.4416,
+      "step": 1885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001257014894554507,
+      "loss": 5.9947,
+      "step": 1886
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012562929830501105,
+      "loss": 5.8081,
+      "step": 1887
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012555709285582691,
+      "loss": 5.7274,
+      "step": 1888
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001254848731481822,
+      "loss": 5.7364,
+      "step": 1889
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012541263922236873,
+      "loss": 5.3902,
+      "step": 1890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012534039111868626,
+      "loss": 6.001,
+      "step": 1891
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012526812887744243,
+      "loss": 5.9476,
+      "step": 1892
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012519585253895286,
+      "loss": 5.6836,
+      "step": 1893
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012512356214354085,
+      "loss": 5.8577,
+      "step": 1894
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001250512577315378,
+      "loss": 5.8482,
+      "step": 1895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012497893934328266,
+      "loss": 5.6658,
+      "step": 1896
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001249066070191224,
+      "loss": 5.584,
+      "step": 1897
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012483426079941163,
+      "loss": 5.8876,
+      "step": 1898
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012476190072451278,
+      "loss": 5.8251,
+      "step": 1899
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012468952683479593,
+      "loss": 5.7291,
+      "step": 1900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000124617139170639,
+      "loss": 5.4948,
+      "step": 1901
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012454473777242747,
+      "loss": 5.6823,
+      "step": 1902
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012447232268055451,
+      "loss": 5.7211,
+      "step": 1903
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000124399893935421,
+      "loss": 5.4988,
+      "step": 1904
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012432745157743536,
+      "loss": 5.6122,
+      "step": 1905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001242549956470136,
+      "loss": 5.9317,
+      "step": 1906
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012418252618457939,
+      "loss": 5.5455,
+      "step": 1907
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012411004323056388,
+      "loss": 5.9802,
+      "step": 1908
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012403754682540572,
+      "loss": 6.0323,
+      "step": 1909
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012396503700955113,
+      "loss": 5.6384,
+      "step": 1910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001238925138234538,
+      "loss": 5.608,
+      "step": 1911
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012381997730757485,
+      "loss": 5.6278,
+      "step": 1912
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012374742750238284,
+      "loss": 5.2892,
+      "step": 1913
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012367486444835377,
+      "loss": 5.3749,
+      "step": 1914
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012360228818597105,
+      "loss": 5.7651,
+      "step": 1915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001235296987557254,
+      "loss": 5.8204,
+      "step": 1916
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012345709619811484,
+      "loss": 5.5542,
+      "step": 1917
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012338448055364487,
+      "loss": 5.861,
+      "step": 1918
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012331185186282824,
+      "loss": 5.5009,
+      "step": 1919
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012323921016618486,
+      "loss": 6.1888,
+      "step": 1920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012316655550424205,
+      "loss": 5.5862,
+      "step": 1921
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012309388791753424,
+      "loss": 5.7021,
+      "step": 1922
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012302120744660318,
+      "loss": 5.6234,
+      "step": 1923
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012294851413199772,
+      "loss": 5.5921,
+      "step": 1924
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012287580801427392,
+      "loss": 5.5504,
+      "step": 1925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000122803089133995,
+      "loss": 5.6034,
+      "step": 1926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012273035753173123,
+      "loss": 5.6783,
+      "step": 1927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012265761324806005,
+      "loss": 5.6386,
+      "step": 1928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012258485632356592,
+      "loss": 5.2511,
+      "step": 1929
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012251208679884044,
+      "loss": 5.4675,
+      "step": 1930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012243930471448212,
+      "loss": 5.634,
+      "step": 1931
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012236651011109654,
+      "loss": 5.3982,
+      "step": 1932
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012229370302929625,
+      "loss": 5.5768,
+      "step": 1933
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012222088350970087,
+      "loss": 5.4685,
+      "step": 1934
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012214805159293672,
+      "loss": 5.6282,
+      "step": 1935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001220752073196373,
+      "loss": 5.6636,
+      "step": 1936
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012200235073044278,
+      "loss": 5.4652,
+      "step": 1937
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001219294818660004,
+      "loss": 5.6762,
+      "step": 1938
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012185660076696408,
+      "loss": 5.5786,
+      "step": 1939
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012178370747399464,
+      "loss": 5.3471,
+      "step": 1940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012171080202775976,
+      "loss": 5.458,
+      "step": 1941
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012163788446893378,
+      "loss": 5.4634,
+      "step": 1942
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012156495483819788,
+      "loss": 5.7569,
+      "step": 1943
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012149201317623995,
+      "loss": 5.5668,
+      "step": 1944
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012141905952375459,
+      "loss": 5.4752,
+      "step": 1945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012134609392144312,
+      "loss": 5.3098,
+      "step": 1946
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012127311641001347,
+      "loss": 5.8015,
+      "step": 1947
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012120012703018027,
+      "loss": 5.6698,
+      "step": 1948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012112712582266473,
+      "loss": 5.5392,
+      "step": 1949
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001210541128281947,
+      "loss": 5.6476,
+      "step": 1950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012098108808750457,
+      "loss": 5.5411,
+      "step": 1951
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012090805164133528,
+      "loss": 5.5406,
+      "step": 1952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012083500353043433,
+      "loss": 5.6789,
+      "step": 1953
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001207619437955557,
+      "loss": 5.3815,
+      "step": 1954
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001206888724774599,
+      "loss": 5.453,
+      "step": 1955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012061578961691384,
+      "loss": 5.7491,
+      "step": 1956
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012054269525469094,
+      "loss": 5.4906,
+      "step": 1957
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012046958943157094,
+      "loss": 5.4009,
+      "step": 1958
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001203964721883401,
+      "loss": 5.4194,
+      "step": 1959
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012032334356579091,
+      "loss": 5.905,
+      "step": 1960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012025020360472237,
+      "loss": 5.5942,
+      "step": 1961
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012017705234593963,
+      "loss": 5.9204,
+      "step": 1962
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012010388983025429,
+      "loss": 5.6862,
+      "step": 1963
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012003071609848413,
+      "loss": 5.8047,
+      "step": 1964
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011995753119145325,
+      "loss": 5.6499,
+      "step": 1965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011988433514999195,
+      "loss": 5.7418,
+      "step": 1966
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011981112801493681,
+      "loss": 5.6442,
+      "step": 1967
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011973790982713047,
+      "loss": 5.3239,
+      "step": 1968
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011966468062742187,
+      "loss": 5.6214,
+      "step": 1969
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011959144045666597,
+      "loss": 5.379,
+      "step": 1970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011951818935572399,
+      "loss": 5.5679,
+      "step": 1971
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011944492736546311,
+      "loss": 5.8959,
+      "step": 1972
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011937165452675672,
+      "loss": 5.2506,
+      "step": 1973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001192983708804841,
+      "loss": 5.5242,
+      "step": 1974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011922507646753071,
+      "loss": 5.5984,
+      "step": 1975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011915177132878792,
+      "loss": 5.468,
+      "step": 1976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001190784555051532,
+      "loss": 5.4182,
+      "step": 1977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011900512903752981,
+      "loss": 5.5114,
+      "step": 1978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011893179196682708,
+      "loss": 5.3334,
+      "step": 1979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001188584443339602,
+      "loss": 5.8111,
+      "step": 1980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011878508617985033,
+      "loss": 5.4782,
+      "step": 1981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011871171754542439,
+      "loss": 5.5,
+      "step": 1982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011863833847161518,
+      "loss": 5.4946,
+      "step": 1983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011856494899936135,
+      "loss": 5.5847,
+      "step": 1984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011849154916960739,
+      "loss": 5.5883,
+      "step": 1985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011841813902330345,
+      "loss": 5.3837,
+      "step": 1986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011834471860140554,
+      "loss": 5.4433,
+      "step": 1987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011827128794487539,
+      "loss": 5.2318,
+      "step": 1988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011819784709468038,
+      "loss": 5.4027,
+      "step": 1989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011812439609179364,
+      "loss": 5.4125,
+      "step": 1990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011805093497719393,
+      "loss": 5.3197,
+      "step": 1991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011797746379186567,
+      "loss": 5.636,
+      "step": 1992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011790398257679884,
+      "loss": 5.6198,
+      "step": 1993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011783049137298916,
+      "loss": 5.4409,
+      "step": 1994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011775699022143775,
+      "loss": 5.4198,
+      "step": 1995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011768347916315141,
+      "loss": 5.3604,
+      "step": 1996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011760995823914238,
+      "loss": 5.5775,
+      "step": 1997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011753642749042844,
+      "loss": 5.5246,
+      "step": 1998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011746288695803285,
+      "loss": 5.358,
+      "step": 1999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011738933668298437,
+      "loss": 5.327,
+      "step": 2000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011731577670631709,
+      "loss": 5.656,
+      "step": 2001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011724220706907063,
+      "loss": 5.5356,
+      "step": 2002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011716862781228992,
+      "loss": 5.5582,
+      "step": 2003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011709503897702529,
+      "loss": 5.3516,
+      "step": 2004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011702144060433243,
+      "loss": 5.3443,
+      "step": 2005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001169478327352723,
+      "loss": 5.4858,
+      "step": 2006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011687421541091117,
+      "loss": 5.1576,
+      "step": 2007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011680058867232067,
+      "loss": 5.6428,
+      "step": 2008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011672695256057753,
+      "loss": 5.3761,
+      "step": 2009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011665330711676386,
+      "loss": 5.414,
+      "step": 2010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011657965238196684,
+      "loss": 5.2476,
+      "step": 2011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000116505988397279,
+      "loss": 5.1554,
+      "step": 2012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011643231520379782,
+      "loss": 5.437,
+      "step": 2013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011635863284262616,
+      "loss": 5.4906,
+      "step": 2014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011628494135487174,
+      "loss": 5.3166,
+      "step": 2015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001162112407816476,
+      "loss": 5.1841,
+      "step": 2016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011613753116407167,
+      "loss": 5.3659,
+      "step": 2017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011606381254326706,
+      "loss": 5.7864,
+      "step": 2018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011599008496036179,
+      "loss": 5.3991,
+      "step": 2019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011591634845648902,
+      "loss": 5.1903,
+      "step": 2020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011584260307278669,
+      "loss": 5.148,
+      "step": 2021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001157688488503979,
+      "loss": 5.4441,
+      "step": 2022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011569508583047053,
+      "loss": 5.285,
+      "step": 2023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011562131405415749,
+      "loss": 5.932,
+      "step": 2024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011554753356261642,
+      "loss": 5.0768,
+      "step": 2025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011547374439701001,
+      "loss": 5.0414,
+      "step": 2026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011539994659850562,
+      "loss": 5.027,
+      "step": 2027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011532614020827555,
+      "loss": 5.2198,
+      "step": 2028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011525232526749679,
+      "loss": 5.1485,
+      "step": 2029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011517850181735121,
+      "loss": 5.274,
+      "step": 2030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011510466989902536,
+      "loss": 5.3166,
+      "step": 2031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011503082955371047,
+      "loss": 5.4804,
+      "step": 2032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001149569808226026,
+      "loss": 5.2274,
+      "step": 2033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011488312374690235,
+      "loss": 5.5538,
+      "step": 2034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001148092583678151,
+      "loss": 5.1501,
+      "step": 2035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011473538472655074,
+      "loss": 5.3138,
+      "step": 2036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011466150286432392,
+      "loss": 5.0166,
+      "step": 2037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011458761282235369,
+      "loss": 5.9526,
+      "step": 2038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011451371464186383,
+      "loss": 5.5414,
+      "step": 2039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011443980836408257,
+      "loss": 5.407,
+      "step": 2040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011436589403024267,
+      "loss": 5.2332,
+      "step": 2041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011429197168158141,
+      "loss": 5.223,
+      "step": 2042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011421804135934058,
+      "loss": 5.3991,
+      "step": 2043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011414410310476624,
+      "loss": 5.193,
+      "step": 2044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011407015695910912,
+      "loss": 5.4026,
+      "step": 2045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011399620296362415,
+      "loss": 5.2777,
+      "step": 2046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011392224115957076,
+      "loss": 5.5588,
+      "step": 2047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011384827158821269,
+      "loss": 5.3843,
+      "step": 2048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011377429429081805,
+      "loss": 5.3966,
+      "step": 2049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011370030930865918,
+      "loss": 5.3262,
+      "step": 2050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011362631668301282,
+      "loss": 5.1651,
+      "step": 2051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011355231645515987,
+      "loss": 4.8893,
+      "step": 2052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011347830866638556,
+      "loss": 5.4239,
+      "step": 2053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011340429335797927,
+      "loss": 5.2002,
+      "step": 2054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011333027057123455,
+      "loss": 5.0707,
+      "step": 2055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011325624034744923,
+      "loss": 5.4757,
+      "step": 2056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011318220272792526,
+      "loss": 5.4874,
+      "step": 2057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001131081577539686,
+      "loss": 5.1644,
+      "step": 2058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011303410546688946,
+      "loss": 5.1072,
+      "step": 2059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011296004590800204,
+      "loss": 5.2465,
+      "step": 2060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011288597911862468,
+      "loss": 5.8931,
+      "step": 2061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011281190514007962,
+      "loss": 5.5762,
+      "step": 2062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011273782401369329,
+      "loss": 5.4374,
+      "step": 2063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001126637357807959,
+      "loss": 4.9889,
+      "step": 2064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011258964048272182,
+      "loss": 5.1574,
+      "step": 2065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011251553816080927,
+      "loss": 5.2789,
+      "step": 2066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011244142885640032,
+      "loss": 5.3039,
+      "step": 2067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001123673126108411,
+      "loss": 5.0442,
+      "step": 2068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011229318946548146,
+      "loss": 5.4445,
+      "step": 2069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011221905946167521,
+      "loss": 5.1922,
+      "step": 2070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011214492264077989,
+      "loss": 5.2595,
+      "step": 2071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011207077904415695,
+      "loss": 5.4255,
+      "step": 2072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001119966287131715,
+      "loss": 5.4981,
+      "step": 2073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011192247168919252,
+      "loss": 5.4042,
+      "step": 2074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011184830801359265,
+      "loss": 5.4922,
+      "step": 2075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011177413772774828,
+      "loss": 5.7683,
+      "step": 2076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011169996087303945,
+      "loss": 5.4848,
+      "step": 2077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001116257774908499,
+      "loss": 5.2768,
+      "step": 2078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011155158762256698,
+      "loss": 5.4852,
+      "step": 2079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011147739130958172,
+      "loss": 5.0833,
+      "step": 2080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011140318859328864,
+      "loss": 5.2956,
+      "step": 2081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011132897951508599,
+      "loss": 5.4122,
+      "step": 2082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011125476411637536,
+      "loss": 5.4259,
+      "step": 2083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011118054243856206,
+      "loss": 5.5007,
+      "step": 2084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001111063145230548,
+      "loss": 4.9879,
+      "step": 2085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011103208041126582,
+      "loss": 5.1759,
+      "step": 2086
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011095784014461074,
+      "loss": 5.2889,
+      "step": 2087
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011088359376450871,
+      "loss": 5.0753,
+      "step": 2088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011080934131238225,
+      "loss": 5.2163,
+      "step": 2089
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011073508282965724,
+      "loss": 5.3708,
+      "step": 2090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011066081835776292,
+      "loss": 5.1635,
+      "step": 2091
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011058654793813195,
+      "loss": 5.1818,
+      "step": 2092
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011051227161220021,
+      "loss": 5.121,
+      "step": 2093
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000110437989421407,
+      "loss": 5.4157,
+      "step": 2094
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011036370140719472,
+      "loss": 5.0666,
+      "step": 2095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011028940761100918,
+      "loss": 5.1464,
+      "step": 2096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011021510807429933,
+      "loss": 4.9469,
+      "step": 2097
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011014080283851736,
+      "loss": 5.2467,
+      "step": 2098
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011006649194511855,
+      "loss": 5.3007,
+      "step": 2099
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001099921754355615,
+      "loss": 4.9361,
+      "step": 2100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010991785335130784,
+      "loss": 5.3561,
+      "step": 2101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010984352573382227,
+      "loss": 5.2066,
+      "step": 2102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010976919262457264,
+      "loss": 5.0449,
+      "step": 2103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010969485406502989,
+      "loss": 5.2406,
+      "step": 2104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010962051009666789,
+      "loss": 5.1916,
+      "step": 2105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010954616076096372,
+      "loss": 4.9326,
+      "step": 2106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001094718060993972,
+      "loss": 5.352,
+      "step": 2107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010939744615345138,
+      "loss": 5.1277,
+      "step": 2108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010932308096461203,
+      "loss": 5.2155,
+      "step": 2109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010924871057436803,
+      "loss": 5.211,
+      "step": 2110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010917433502421102,
+      "loss": 5.0565,
+      "step": 2111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010909995435563565,
+      "loss": 5.35,
+      "step": 2112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001090255686101393,
+      "loss": 5.2761,
+      "step": 2113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010895117782922223,
+      "loss": 5.1442,
+      "step": 2114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010887678205438758,
+      "loss": 5.097,
+      "step": 2115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010880238132714116,
+      "loss": 5.1622,
+      "step": 2116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010872797568899163,
+      "loss": 5.1246,
+      "step": 2117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010865356518145028,
+      "loss": 5.1215,
+      "step": 2118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010857914984603133,
+      "loss": 5.0179,
+      "step": 2119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010850472972425143,
+      "loss": 5.1776,
+      "step": 2120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010843030485763012,
+      "loss": 5.1414,
+      "step": 2121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010835587528768943,
+      "loss": 5.1621,
+      "step": 2122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001082814410559541,
+      "loss": 5.0264,
+      "step": 2123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010820700220395147,
+      "loss": 4.9995,
+      "step": 2124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010813255877321142,
+      "loss": 5.2157,
+      "step": 2125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001080581108052664,
+      "loss": 5.2887,
+      "step": 2126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010798365834165139,
+      "loss": 5.0837,
+      "step": 2127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010790920142390386,
+      "loss": 5.0769,
+      "step": 2128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010783474009356385,
+      "loss": 5.2646,
+      "step": 2129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010776027439217371,
+      "loss": 5.2979,
+      "step": 2130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010768580436127841,
+      "loss": 5.0926,
+      "step": 2131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010761133004242516,
+      "loss": 5.2711,
+      "step": 2132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010753685147716373,
+      "loss": 5.1218,
+      "step": 2133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010746236870704609,
+      "loss": 4.721,
+      "step": 2134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010738788177362668,
+      "loss": 5.332,
+      "step": 2135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010731339071846223,
+      "loss": 5.4181,
+      "step": 2136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010723889558311176,
+      "loss": 5.1604,
+      "step": 2137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010716439640913652,
+      "loss": 5.2302,
+      "step": 2138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010708989323810011,
+      "loss": 5.0168,
+      "step": 2139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001070153861115683,
+      "loss": 5.0986,
+      "step": 2140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010694087507110908,
+      "loss": 5.1658,
+      "step": 2141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010686636015829258,
+      "loss": 5.2025,
+      "step": 2142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010679184141469123,
+      "loss": 4.9841,
+      "step": 2143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010671731888187939,
+      "loss": 5.0597,
+      "step": 2144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010664279260143373,
+      "loss": 5.0292,
+      "step": 2145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010656826261493285,
+      "loss": 4.9571,
+      "step": 2146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010649372896395755,
+      "loss": 5.099,
+      "step": 2147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001064191916900906,
+      "loss": 5.1225,
+      "step": 2148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010634465083491679,
+      "loss": 5.2088,
+      "step": 2149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010627010644002296,
+      "loss": 4.8165,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001061955585469978,
+      "loss": 5.2179,
+      "step": 2151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010612100719743217,
+      "loss": 4.8948,
+      "step": 2152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010604645243291864,
+      "loss": 5.139,
+      "step": 2153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010597189429505184,
+      "loss": 4.7754,
+      "step": 2154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010589733282542815,
+      "loss": 5.2204,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001058227680656459,
+      "loss": 5.0672,
+      "step": 2156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010574820005730525,
+      "loss": 4.9097,
+      "step": 2157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010567362884200816,
+      "loss": 4.8816,
+      "step": 2158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010559905446135833,
+      "loss": 5.1841,
+      "step": 2159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010552447695696133,
+      "loss": 5.1294,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010544989637042435,
+      "loss": 4.8983,
+      "step": 2161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010537531274335641,
+      "loss": 4.9008,
+      "step": 2162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010530072611736812,
+      "loss": 5.1846,
+      "step": 2163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010522613653407189,
+      "loss": 5.1147,
+      "step": 2164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010515154403508167,
+      "loss": 5.0365,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010507694866201312,
+      "loss": 5.0303,
+      "step": 2166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001050023504564834,
+      "loss": 5.1024,
+      "step": 2167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010492774946011134,
+      "loss": 5.0464,
+      "step": 2168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001048531457145173,
+      "loss": 5.04,
+      "step": 2169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010477853926132317,
+      "loss": 4.8314,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010470393014215239,
+      "loss": 4.9606,
+      "step": 2171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010462931839862977,
+      "loss": 5.151,
+      "step": 2172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010455470407238172,
+      "loss": 5.5782,
+      "step": 2173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010448008720503604,
+      "loss": 4.9281,
+      "step": 2174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010440546783822191,
+      "loss": 5.082,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010433084601356998,
+      "loss": 5.1515,
+      "step": 2176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010425622177271215,
+      "loss": 5.2052,
+      "step": 2177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010418159515728185,
+      "loss": 5.0718,
+      "step": 2178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010410696620891367,
+      "loss": 5.19,
+      "step": 2179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010403233496924356,
+      "loss": 5.067,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010395770147990878,
+      "loss": 4.9891,
+      "step": 2181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010388306578254783,
+      "loss": 4.9502,
+      "step": 2182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010380842791880036,
+      "loss": 5.273,
+      "step": 2183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010373378793030736,
+      "loss": 5.0941,
+      "step": 2184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010365914585871091,
+      "loss": 5.4155,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010358450174565428,
+      "loss": 5.3562,
+      "step": 2186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001035098556327819,
+      "loss": 5.1829,
+      "step": 2187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010343520756173935,
+      "loss": 5.0781,
+      "step": 2188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010336055757417314,
+      "loss": 4.9675,
+      "step": 2189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010328590571173105,
+      "loss": 5.0741,
+      "step": 2190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010321125201606177,
+      "loss": 5.1012,
+      "step": 2191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010313659652881508,
+      "loss": 4.8107,
+      "step": 2192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010306193929164173,
+      "loss": 5.2464,
+      "step": 2193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010298728034619352,
+      "loss": 5.0688,
+      "step": 2194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010291261973412302,
+      "loss": 5.2515,
+      "step": 2195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010283795749708393,
+      "loss": 5.017,
+      "step": 2196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010276329367673077,
+      "loss": 5.3024,
+      "step": 2197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010268862831471889,
+      "loss": 4.9233,
+      "step": 2198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010261396145270463,
+      "loss": 5.0355,
+      "step": 2199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010253929313234505,
+      "loss": 5.0326,
+      "step": 2200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010246462339529812,
+      "loss": 4.9686,
+      "step": 2201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010238995228322248,
+      "loss": 5.1409,
+      "step": 2202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010231527983777765,
+      "loss": 4.8924,
+      "step": 2203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010224060610062383,
+      "loss": 4.9972,
+      "step": 2204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010216593111342204,
+      "loss": 4.7736,
+      "step": 2205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010209125491783377,
+      "loss": 5.025,
+      "step": 2206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010201657755552145,
+      "loss": 4.914,
+      "step": 2207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010194189906814799,
+      "loss": 5.3443,
+      "step": 2208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010186721949737701,
+      "loss": 4.9102,
+      "step": 2209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010179253888487268,
+      "loss": 4.8983,
+      "step": 2210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010171785727229981,
+      "loss": 4.824,
+      "step": 2211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001016431747013237,
+      "loss": 4.8926,
+      "step": 2212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010156849121361025,
+      "loss": 5.0299,
+      "step": 2213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010149380685082582,
+      "loss": 5.0338,
+      "step": 2214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010141912165463728,
+      "loss": 5.0945,
+      "step": 2215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010134443566671195,
+      "loss": 5.2583,
+      "step": 2216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010126974892871768,
+      "loss": 5.572,
+      "step": 2217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010119506148232257,
+      "loss": 4.8467,
+      "step": 2218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010112037336919527,
+      "loss": 5.2996,
+      "step": 2219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010104568463100467,
+      "loss": 4.781,
+      "step": 2220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010097099530942019,
+      "loss": 4.782,
+      "step": 2221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010089630544611135,
+      "loss": 4.9131,
+      "step": 2222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010082161508274819,
+      "loss": 5.0252,
+      "step": 2223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010074692426100085,
+      "loss": 4.7785,
+      "step": 2224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010067223302253985,
+      "loss": 4.9819,
+      "step": 2225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010059754140903588,
+      "loss": 5.0495,
+      "step": 2226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010052284946215984,
+      "loss": 5.3168,
+      "step": 2227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010044815722358285,
+      "loss": 4.9089,
+      "step": 2228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010037346473497618,
+      "loss": 5.1307,
+      "step": 2229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010029877203801122,
+      "loss": 4.8795,
+      "step": 2230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010022407917435943,
+      "loss": 5.0554,
+      "step": 2231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010014938618569253,
+      "loss": 4.7961,
+      "step": 2232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010007469311368214,
+      "loss": 4.6947,
+      "step": 2233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001,
+      "loss": 5.1526,
+      "step": 2234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992530688631789e-05,
+      "loss": 4.9397,
+      "step": 2235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985061381430747e-05,
+      "loss": 4.9638,
+      "step": 2236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.977592082564058e-05,
+      "loss": 5.1175,
+      "step": 2237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.970122796198882e-05,
+      "loss": 4.9208,
+      "step": 2238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962653526502386e-05,
+      "loss": 4.7494,
+      "step": 2239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.955184277641716e-05,
+      "loss": 5.0913,
+      "step": 2240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.947715053784018e-05,
+      "loss": 5.2166,
+      "step": 2241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940245859096413e-05,
+      "loss": 4.8968,
+      "step": 2242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.932776697746017e-05,
+      "loss": 5.0003,
+      "step": 2243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925307573899914e-05,
+      "loss": 4.7831,
+      "step": 2244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917838491725183e-05,
+      "loss": 4.9058,
+      "step": 2245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.910369455388869e-05,
+      "loss": 5.4679,
+      "step": 2246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.902900469057984e-05,
+      "loss": 4.8061,
+      "step": 2247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895431536899536e-05,
+      "loss": 5.3939,
+      "step": 2248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887962663080477e-05,
+      "loss": 4.907,
+      "step": 2249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880493851767748e-05,
+      "loss": 4.8924,
+      "step": 2250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.873025107128235e-05,
+      "loss": 5.0048,
+      "step": 2251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.865556433328806e-05,
+      "loss": 4.9126,
+      "step": 2252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.858087834536273e-05,
+      "loss": 4.9105,
+      "step": 2253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850619314917422e-05,
+      "loss": 4.732,
+      "step": 2254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.843150878638975e-05,
+      "loss": 4.8195,
+      "step": 2255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835682529867633e-05,
+      "loss": 4.7132,
+      "step": 2256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.828214272770019e-05,
+      "loss": 5.1971,
+      "step": 2257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.820746111512734e-05,
+      "loss": 4.8838,
+      "step": 2258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.813278050262304e-05,
+      "loss": 4.4306,
+      "step": 2259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805810093185204e-05,
+      "loss": 5.0948,
+      "step": 2260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79834224444786e-05,
+      "loss": 5.0393,
+      "step": 2261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.790874508216625e-05,
+      "loss": 5.0072,
+      "step": 2262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783406888657803e-05,
+      "loss": 5.0735,
+      "step": 2263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775939389937617e-05,
+      "loss": 5.8258,
+      "step": 2264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.768472016222236e-05,
+      "loss": 4.5821,
+      "step": 2265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761004771677753e-05,
+      "loss": 5.013,
+      "step": 2266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.753537660470191e-05,
+      "loss": 5.0631,
+      "step": 2267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746070686765495e-05,
+      "loss": 4.5676,
+      "step": 2268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73860385472954e-05,
+      "loss": 4.7859,
+      "step": 2269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731137168528112e-05,
+      "loss": 5.2783,
+      "step": 2270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.723670632326927e-05,
+      "loss": 5.4088,
+      "step": 2271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716204250291609e-05,
+      "loss": 5.059,
+      "step": 2272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.708738026587699e-05,
+      "loss": 4.8639,
+      "step": 2273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.701271965380652e-05,
+      "loss": 5.2594,
+      "step": 2274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.693806070835825e-05,
+      "loss": 5.1352,
+      "step": 2275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.686340347118493e-05,
+      "loss": 4.964,
+      "step": 2276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.678874798393824e-05,
+      "loss": 4.7904,
+      "step": 2277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671409428826897e-05,
+      "loss": 4.9325,
+      "step": 2278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663944242582687e-05,
+      "loss": 4.7724,
+      "step": 2279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656479243826069e-05,
+      "loss": 4.7177,
+      "step": 2280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649014436721812e-05,
+      "loss": 4.7991,
+      "step": 2281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.641549825434573e-05,
+      "loss": 4.8368,
+      "step": 2282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634085414128915e-05,
+      "loss": 4.4968,
+      "step": 2283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.626621206969266e-05,
+      "loss": 5.1123,
+      "step": 2284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619157208119969e-05,
+      "loss": 4.985,
+      "step": 2285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611693421745219e-05,
+      "loss": 4.8275,
+      "step": 2286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604229852009124e-05,
+      "loss": 4.8851,
+      "step": 2287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596766503075643e-05,
+      "loss": 4.6306,
+      "step": 2288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.589303379108636e-05,
+      "loss": 4.9313,
+      "step": 2289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.581840484271814e-05,
+      "loss": 4.8506,
+      "step": 2290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.574377822728786e-05,
+      "loss": 5.1137,
+      "step": 2291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.566915398643006e-05,
+      "loss": 5.003,
+      "step": 2292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559453216177811e-05,
+      "loss": 5.2056,
+      "step": 2293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551991279496401e-05,
+      "loss": 5.4428,
+      "step": 2294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54452959276183e-05,
+      "loss": 5.0995,
+      "step": 2295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537068160137026e-05,
+      "loss": 4.6788,
+      "step": 2296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529606985784765e-05,
+      "loss": 4.7266,
+      "step": 2297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.522146073867685e-05,
+      "loss": 5.4772,
+      "step": 2298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.514685428548271e-05,
+      "loss": 4.7043,
+      "step": 2299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507225053988868e-05,
+      "loss": 5.2481,
+      "step": 2300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499764954351661e-05,
+      "loss": 4.8794,
+      "step": 2301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492305133798691e-05,
+      "loss": 4.8964,
+      "step": 2302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.484845596491833e-05,
+      "loss": 4.821,
+      "step": 2303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477386346592812e-05,
+      "loss": 4.627,
+      "step": 2304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46992738826319e-05,
+      "loss": 4.992,
+      "step": 2305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462468725664362e-05,
+      "loss": 4.968,
+      "step": 2306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.455010362957569e-05,
+      "loss": 4.5275,
+      "step": 2307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44755230430387e-05,
+      "loss": 5.2811,
+      "step": 2308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.440094553864168e-05,
+      "loss": 4.9623,
+      "step": 2309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432637115799185e-05,
+      "loss": 4.9603,
+      "step": 2310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425179994269476e-05,
+      "loss": 4.8355,
+      "step": 2311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.417723193435409e-05,
+      "loss": 4.7826,
+      "step": 2312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.410266717457187e-05,
+      "loss": 5.2518,
+      "step": 2313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402810570494817e-05,
+      "loss": 4.8633,
+      "step": 2314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395354756708138e-05,
+      "loss": 4.7686,
+      "step": 2315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.387899280256787e-05,
+      "loss": 4.8729,
+      "step": 2316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38044414530022e-05,
+      "loss": 4.5588,
+      "step": 2317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37298935599771e-05,
+      "loss": 4.9615,
+      "step": 2318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.365534916508324e-05,
+      "loss": 4.9145,
+      "step": 2319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358080830990944e-05,
+      "loss": 5.1345,
+      "step": 2320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350627103604245e-05,
+      "loss": 4.6743,
+      "step": 2321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.343173738506716e-05,
+      "loss": 5.1294,
+      "step": 2322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.335720739856629e-05,
+      "loss": 5.067,
+      "step": 2323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.328268111812062e-05,
+      "loss": 4.9218,
+      "step": 2324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32081585853088e-05,
+      "loss": 4.9438,
+      "step": 2325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313363984170743e-05,
+      "loss": 4.864,
+      "step": 2326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.305912492889097e-05,
+      "loss": 4.7595,
+      "step": 2327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298461388843174e-05,
+      "loss": 5.0221,
+      "step": 2328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.291010676189993e-05,
+      "loss": 4.6497,
+      "step": 2329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28356035908635e-05,
+      "loss": 4.7184,
+      "step": 2330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276110441688829e-05,
+      "loss": 4.6971,
+      "step": 2331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268660928153778e-05,
+      "loss": 4.675,
+      "step": 2332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.261211822637334e-05,
+      "loss": 4.7457,
+      "step": 2333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253763129295392e-05,
+      "loss": 5.1473,
+      "step": 2334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24631485228363e-05,
+      "loss": 4.827,
+      "step": 2335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.238866995757482e-05,
+      "loss": 5.0089,
+      "step": 2336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231419563872161e-05,
+      "loss": 5.1739,
+      "step": 2337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223972560782627e-05,
+      "loss": 4.7765,
+      "step": 2338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216525990643617e-05,
+      "loss": 5.006,
+      "step": 2339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209079857609618e-05,
+      "loss": 5.0262,
+      "step": 2340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201634165834863e-05,
+      "loss": 4.7026,
+      "step": 2341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194188919473365e-05,
+      "loss": 4.6917,
+      "step": 2342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.186744122678859e-05,
+      "loss": 4.4732,
+      "step": 2343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179299779604856e-05,
+      "loss": 4.6513,
+      "step": 2344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171855894404588e-05,
+      "loss": 4.7863,
+      "step": 2345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.164412471231059e-05,
+      "loss": 4.998,
+      "step": 2346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156969514236989e-05,
+      "loss": 4.6338,
+      "step": 2347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.149527027574858e-05,
+      "loss": 4.6184,
+      "step": 2348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.142085015396867e-05,
+      "loss": 4.7695,
+      "step": 2349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134643481854973e-05,
+      "loss": 4.5672,
+      "step": 2350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127202431100844e-05,
+      "loss": 4.6122,
+      "step": 2351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.119761867285888e-05,
+      "loss": 5.0029,
+      "step": 2352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.112321794561246e-05,
+      "loss": 5.0366,
+      "step": 2353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104882217077779e-05,
+      "loss": 4.9236,
+      "step": 2354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.097443138986075e-05,
+      "loss": 5.0191,
+      "step": 2355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090004564436439e-05,
+      "loss": 4.7351,
+      "step": 2356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.082566497578899e-05,
+      "loss": 4.4556,
+      "step": 2357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075128942563199e-05,
+      "loss": 5.0486,
+      "step": 2358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.067691903538799e-05,
+      "loss": 4.5613,
+      "step": 2359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060255384654866e-05,
+      "loss": 4.6734,
+      "step": 2360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.052819390060282e-05,
+      "loss": 4.7827,
+      "step": 2361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.045383923903635e-05,
+      "loss": 4.8477,
+      "step": 2362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.037948990333212e-05,
+      "loss": 5.0421,
+      "step": 2363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.030514593497017e-05,
+      "loss": 4.9573,
+      "step": 2364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.023080737542739e-05,
+      "loss": 4.8453,
+      "step": 2365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015647426617777e-05,
+      "loss": 4.7406,
+      "step": 2366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00821466486922e-05,
+      "loss": 4.7427,
+      "step": 2367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00078245644385e-05,
+      "loss": 4.8062,
+      "step": 2368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.993350805488144e-05,
+      "loss": 5.3919,
+      "step": 2369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.985919716148268e-05,
+      "loss": 5.1993,
+      "step": 2370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978489192570068e-05,
+      "loss": 4.9046,
+      "step": 2371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.971059238899083e-05,
+      "loss": 5.2253,
+      "step": 2372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963629859280527e-05,
+      "loss": 4.5589,
+      "step": 2373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.956201057859304e-05,
+      "loss": 5.0235,
+      "step": 2374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.948772838779983e-05,
+      "loss": 5.1138,
+      "step": 2375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941345206186807e-05,
+      "loss": 4.7247,
+      "step": 2376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.933918164223713e-05,
+      "loss": 4.812,
+      "step": 2377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926491717034278e-05,
+      "loss": 5.0594,
+      "step": 2378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.919065868761779e-05,
+      "loss": 4.6537,
+      "step": 2379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911640623549128e-05,
+      "loss": 4.6653,
+      "step": 2380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904215985538927e-05,
+      "loss": 4.7145,
+      "step": 2381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896791958873417e-05,
+      "loss": 4.7381,
+      "step": 2382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889368547694522e-05,
+      "loss": 4.4969,
+      "step": 2383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.881945756143792e-05,
+      "loss": 4.635,
+      "step": 2384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.874523588362466e-05,
+      "loss": 4.7368,
+      "step": 2385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867102048491406e-05,
+      "loss": 4.7419,
+      "step": 2386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859681140671137e-05,
+      "loss": 4.8661,
+      "step": 2387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852260869041833e-05,
+      "loss": 5.1793,
+      "step": 2388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.844841237743304e-05,
+      "loss": 4.9025,
+      "step": 2389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837422250915013e-05,
+      "loss": 4.8197,
+      "step": 2390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830003912696057e-05,
+      "loss": 4.4873,
+      "step": 2391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.822586227225176e-05,
+      "loss": 5.0499,
+      "step": 2392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.815169198640736e-05,
+      "loss": 4.7104,
+      "step": 2393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80775283108075e-05,
+      "loss": 4.8909,
+      "step": 2394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.800337128682849e-05,
+      "loss": 5.6265,
+      "step": 2395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.792922095584308e-05,
+      "loss": 5.3003,
+      "step": 2396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.785507735922012e-05,
+      "loss": 4.6338,
+      "step": 2397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.778094053832482e-05,
+      "loss": 4.9684,
+      "step": 2398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.770681053451856e-05,
+      "loss": 5.1384,
+      "step": 2399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763268738915892e-05,
+      "loss": 4.3649,
+      "step": 2400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75585711435997e-05,
+      "loss": 4.8313,
+      "step": 2401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.748446183919076e-05,
+      "loss": 4.529,
+      "step": 2402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.741035951727819e-05,
+      "loss": 4.7177,
+      "step": 2403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73362642192041e-05,
+      "loss": 4.9056,
+      "step": 2404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.726217598630675e-05,
+      "loss": 5.3655,
+      "step": 2405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718809485992037e-05,
+      "loss": 4.7964,
+      "step": 2406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.711402088137534e-05,
+      "loss": 5.1476,
+      "step": 2407
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703995409199796e-05,
+      "loss": 4.9152,
+      "step": 2408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.696589453311055e-05,
+      "loss": 5.304,
+      "step": 2409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.689184224603146e-05,
+      "loss": 4.7596,
+      "step": 2410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.681779727207477e-05,
+      "loss": 4.8144,
+      "step": 2411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67437596525508e-05,
+      "loss": 4.7594,
+      "step": 2412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.666972942876546e-05,
+      "loss": 4.3688,
+      "step": 2413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65957066420208e-05,
+      "loss": 4.8495,
+      "step": 2414
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652169133361445e-05,
+      "loss": 4.6022,
+      "step": 2415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644768354484014e-05,
+      "loss": 4.9917,
+      "step": 2416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.637368331698718e-05,
+      "loss": 4.8793,
+      "step": 2417
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.629969069134083e-05,
+      "loss": 4.8649,
+      "step": 2418
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.622570570918195e-05,
+      "loss": 4.644,
+      "step": 2419
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.615172841178733e-05,
+      "loss": 4.5259,
+      "step": 2420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607775884042928e-05,
+      "loss": 4.7908,
+      "step": 2421
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600379703637589e-05,
+      "loss": 4.962,
+      "step": 2422
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592984304089093e-05,
+      "loss": 4.7355,
+      "step": 2423
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.585589689523378e-05,
+      "loss": 4.5852,
+      "step": 2424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.578195864065947e-05,
+      "loss": 4.7592,
+      "step": 2425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570802831841857e-05,
+      "loss": 4.6863,
+      "step": 2426
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.563410596975734e-05,
+      "loss": 4.5579,
+      "step": 2427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556019163591744e-05,
+      "loss": 5.8094,
+      "step": 2428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.548628535813618e-05,
+      "loss": 4.7312,
+      "step": 2429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.541238717764631e-05,
+      "loss": 4.5154,
+      "step": 2430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53384971356761e-05,
+      "loss": 4.722,
+      "step": 2431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526461527344927e-05,
+      "loss": 5.0966,
+      "step": 2432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.519074163218492e-05,
+      "loss": 4.6346,
+      "step": 2433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511687625309767e-05,
+      "loss": 4.8888,
+      "step": 2434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.504301917739741e-05,
+      "loss": 4.7029,
+      "step": 2435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496917044628955e-05,
+      "loss": 4.7027,
+      "step": 2436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489533010097465e-05,
+      "loss": 4.8332,
+      "step": 2437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48214981826488e-05,
+      "loss": 4.4167,
+      "step": 2438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.474767473250321e-05,
+      "loss": 4.6962,
+      "step": 2439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467385979172448e-05,
+      "loss": 4.7657,
+      "step": 2440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.460005340149438e-05,
+      "loss": 5.0822,
+      "step": 2441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.452625560299001e-05,
+      "loss": 5.3616,
+      "step": 2442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445246643738357e-05,
+      "loss": 4.9705,
+      "step": 2443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437868594584254e-05,
+      "loss": 4.8515,
+      "step": 2444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43049141695295e-05,
+      "loss": 4.7522,
+      "step": 2445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.423115114960211e-05,
+      "loss": 4.639,
+      "step": 2446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415739692721335e-05,
+      "loss": 4.8201,
+      "step": 2447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.408365154351101e-05,
+      "loss": 4.92,
+      "step": 2448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.400991503963823e-05,
+      "loss": 4.7201,
+      "step": 2449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.393618745673297e-05,
+      "loss": 5.4807,
+      "step": 2450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.386246883592836e-05,
+      "loss": 5.1874,
+      "step": 2451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.378875921835243e-05,
+      "loss": 4.9036,
+      "step": 2452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.371505864512827e-05,
+      "loss": 4.572,
+      "step": 2453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364136715737387e-05,
+      "loss": 4.9411,
+      "step": 2454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356768479620219e-05,
+      "loss": 4.9857,
+      "step": 2455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349401160272105e-05,
+      "loss": 5.1271,
+      "step": 2456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.342034761803317e-05,
+      "loss": 4.86,
+      "step": 2457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.334669288323619e-05,
+      "loss": 4.7699,
+      "step": 2458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32730474394225e-05,
+      "loss": 4.8615,
+      "step": 2459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319941132767938e-05,
+      "loss": 5.1049,
+      "step": 2460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.312578458908884e-05,
+      "loss": 4.8127,
+      "step": 2461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305216726472773e-05,
+      "loss": 4.923,
+      "step": 2462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.297855939566758e-05,
+      "loss": 4.7137,
+      "step": 2463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290496102297472e-05,
+      "loss": 4.4381,
+      "step": 2464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283137218771007e-05,
+      "loss": 5.0585,
+      "step": 2465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.275779293092938e-05,
+      "loss": 4.648,
+      "step": 2466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.268422329368292e-05,
+      "loss": 5.142,
+      "step": 2467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.261066331701566e-05,
+      "loss": 4.7871,
+      "step": 2468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253711304196716e-05,
+      "loss": 4.3545,
+      "step": 2469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246357250957157e-05,
+      "loss": 4.6464,
+      "step": 2470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.239004176085765e-05,
+      "loss": 4.8674,
+      "step": 2471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23165208368486e-05,
+      "loss": 4.7031,
+      "step": 2472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.224300977856227e-05,
+      "loss": 4.3758,
+      "step": 2473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216950862701084e-05,
+      "loss": 4.3057,
+      "step": 2474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.209601742320117e-05,
+      "loss": 4.4397,
+      "step": 2475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202253620813434e-05,
+      "loss": 4.7974,
+      "step": 2476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19490650228061e-05,
+      "loss": 4.5097,
+      "step": 2477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18756039082064e-05,
+      "loss": 5.024,
+      "step": 2478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180215290531965e-05,
+      "loss": 4.736,
+      "step": 2479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172871205512465e-05,
+      "loss": 4.6788,
+      "step": 2480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165528139859448e-05,
+      "loss": 4.6832,
+      "step": 2481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158186097669658e-05,
+      "loss": 4.7676,
+      "step": 2482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150845083039265e-05,
+      "loss": 4.8322,
+      "step": 2483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.143505100063867e-05,
+      "loss": 4.9833,
+      "step": 2484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136166152838484e-05,
+      "loss": 4.7164,
+      "step": 2485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.128828245457565e-05,
+      "loss": 4.5524,
+      "step": 2486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121491382014969e-05,
+      "loss": 5.0264,
+      "step": 2487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.114155566603981e-05,
+      "loss": 4.7615,
+      "step": 2488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106820803317293e-05,
+      "loss": 4.7684,
+      "step": 2489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099487096247023e-05,
+      "loss": 4.4197,
+      "step": 2490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092154449484685e-05,
+      "loss": 4.4412,
+      "step": 2491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.084822867121209e-05,
+      "loss": 4.6064,
+      "step": 2492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077492353246934e-05,
+      "loss": 4.6365,
+      "step": 2493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.070162911951593e-05,
+      "loss": 5.1831,
+      "step": 2494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062834547324334e-05,
+      "loss": 4.4377,
+      "step": 2495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05550726345369e-05,
+      "loss": 4.4655,
+      "step": 2496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.048181064427603e-05,
+      "loss": 5.521,
+      "step": 2497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.040855954333402e-05,
+      "loss": 4.8323,
+      "step": 2498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033531937257817e-05,
+      "loss": 4.929,
+      "step": 2499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026209017286954e-05,
+      "loss": 4.669,
+      "step": 2500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01888719850632e-05,
+      "loss": 4.46,
+      "step": 2501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.011566485000807e-05,
+      "loss": 4.6634,
+      "step": 2502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.004246880854676e-05,
+      "loss": 4.5859,
+      "step": 2503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.996928390151592e-05,
+      "loss": 4.8602,
+      "step": 2504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989611016974573e-05,
+      "loss": 4.8802,
+      "step": 2505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98229476540604e-05,
+      "loss": 4.7623,
+      "step": 2506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.974979639527764e-05,
+      "loss": 4.9374,
+      "step": 2507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96766564342091e-05,
+      "loss": 4.6957,
+      "step": 2508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96035278116599e-05,
+      "loss": 4.8734,
+      "step": 2509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.953041056842909e-05,
+      "loss": 4.3131,
+      "step": 2510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.945730474530906e-05,
+      "loss": 4.6002,
+      "step": 2511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938421038308618e-05,
+      "loss": 4.8045,
+      "step": 2512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.931112752254014e-05,
+      "loss": 4.5105,
+      "step": 2513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.923805620444433e-05,
+      "loss": 4.9483,
+      "step": 2514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.916499646956572e-05,
+      "loss": 4.7307,
+      "step": 2515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.909194835866476e-05,
+      "loss": 4.6364,
+      "step": 2516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.901891191249548e-05,
+      "loss": 4.9495,
+      "step": 2517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894588717180532e-05,
+      "loss": 4.5654,
+      "step": 2518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887287417733528e-05,
+      "loss": 4.7037,
+      "step": 2519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.879987296981974e-05,
+      "loss": 4.9111,
+      "step": 2520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872688358998656e-05,
+      "loss": 4.487,
+      "step": 2521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865390607855689e-05,
+      "loss": 4.505,
+      "step": 2522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858094047624543e-05,
+      "loss": 4.7491,
+      "step": 2523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.850798682376008e-05,
+      "loss": 4.6972,
+      "step": 2524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.843504516180216e-05,
+      "loss": 4.6586,
+      "step": 2525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836211553106626e-05,
+      "loss": 4.794,
+      "step": 2526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.828919797224027e-05,
+      "loss": 5.0303,
+      "step": 2527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.821629252600537e-05,
+      "loss": 4.6147,
+      "step": 2528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814339923303594e-05,
+      "loss": 4.5489,
+      "step": 2529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807051813399964e-05,
+      "loss": 4.8055,
+      "step": 2530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799764926955722e-05,
+      "loss": 4.523,
+      "step": 2531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792479268036274e-05,
+      "loss": 4.8169,
+      "step": 2532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.785194840706328e-05,
+      "loss": 4.6939,
+      "step": 2533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777911649029917e-05,
+      "loss": 4.3998,
+      "step": 2534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.770629697070373e-05,
+      "loss": 4.5966,
+      "step": 2535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.763348988890348e-05,
+      "loss": 4.4522,
+      "step": 2536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756069528551794e-05,
+      "loss": 4.4232,
+      "step": 2537
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748791320115959e-05,
+      "loss": 4.5244,
+      "step": 2538
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.741514367643412e-05,
+      "loss": 4.5209,
+      "step": 2539
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734238675193997e-05,
+      "loss": 4.6644,
+      "step": 2540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72696424682688e-05,
+      "loss": 4.7317,
+      "step": 2541
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719691086600501e-05,
+      "loss": 4.5704,
+      "step": 2542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71241919857261e-05,
+      "loss": 4.7243,
+      "step": 2543
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.705148586800228e-05,
+      "loss": 4.3015,
+      "step": 2544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697879255339685e-05,
+      "loss": 4.704,
+      "step": 2545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.690611208246576e-05,
+      "loss": 4.5653,
+      "step": 2546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.683344449575799e-05,
+      "loss": 4.7399,
+      "step": 2547
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.676078983381518e-05,
+      "loss": 4.8721,
+      "step": 2548
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66881481371718e-05,
+      "loss": 4.9128,
+      "step": 2549
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.661551944635515e-05,
+      "loss": 4.5,
+      "step": 2550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.65429038018852e-05,
+      "loss": 4.6276,
+      "step": 2551
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647030124427467e-05,
+      "loss": 4.5222,
+      "step": 2552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.639771181402898e-05,
+      "loss": 4.6174,
+      "step": 2553
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632513555164624e-05,
+      "loss": 4.6407,
+      "step": 2554
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625257249761717e-05,
+      "loss": 5.4892,
+      "step": 2555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.618002269242517e-05,
+      "loss": 4.2958,
+      "step": 2556
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.610748617654621e-05,
+      "loss": 4.87,
+      "step": 2557
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.603496299044888e-05,
+      "loss": 4.5434,
+      "step": 2558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.596245317459429e-05,
+      "loss": 4.8543,
+      "step": 2559
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.588995676943616e-05,
+      "loss": 5.2115,
+      "step": 2560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.581747381542064e-05,
+      "loss": 4.7096,
+      "step": 2561
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.574500435298641e-05,
+      "loss": 4.4577,
+      "step": 2562
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567254842256468e-05,
+      "loss": 4.6197,
+      "step": 2563
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.560010606457902e-05,
+      "loss": 4.6103,
+      "step": 2564
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.552767731944551e-05,
+      "loss": 4.5186,
+      "step": 2565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545526222757256e-05,
+      "loss": 4.6615,
+      "step": 2566
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.538286082936102e-05,
+      "loss": 4.4093,
+      "step": 2567
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.531047316520406e-05,
+      "loss": 4.4082,
+      "step": 2568
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523809927548725e-05,
+      "loss": 4.97,
+      "step": 2569
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516573920058837e-05,
+      "loss": 4.8685,
+      "step": 2570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.509339298087761e-05,
+      "loss": 4.389,
+      "step": 2571
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.502106065671737e-05,
+      "loss": 4.3678,
+      "step": 2572
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494874226846222e-05,
+      "loss": 4.3884,
+      "step": 2573
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.487643785645917e-05,
+      "loss": 5.0239,
+      "step": 2574
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480414746104717e-05,
+      "loss": 4.361,
+      "step": 2575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473187112255759e-05,
+      "loss": 4.597,
+      "step": 2576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465960888131374e-05,
+      "loss": 4.4765,
+      "step": 2577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.45873607776313e-05,
+      "loss": 4.5183,
+      "step": 2578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.45151268518178e-05,
+      "loss": 4.7765,
+      "step": 2579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444290714417312e-05,
+      "loss": 4.6786,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.437070169498896e-05,
+      "loss": 4.9315,
+      "step": 2581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429851054454932e-05,
+      "loss": 5.168,
+      "step": 2582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.422633373312999e-05,
+      "loss": 4.7403,
+      "step": 2583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415417130099888e-05,
+      "loss": 4.2922,
+      "step": 2584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408202328841589e-05,
+      "loss": 4.6958,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.400988973563279e-05,
+      "loss": 4.9281,
+      "step": 2586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.39377706828934e-05,
+      "loss": 4.5661,
+      "step": 2587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386566617043335e-05,
+      "loss": 4.8873,
+      "step": 2588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37935762384802e-05,
+      "loss": 4.1272,
+      "step": 2589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372150092725337e-05,
+      "loss": 4.5937,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364944027696414e-05,
+      "loss": 4.435,
+      "step": 2591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357739432781554e-05,
+      "loss": 4.6622,
+      "step": 2592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350536312000254e-05,
+      "loss": 4.6201,
+      "step": 2593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34333466937117e-05,
+      "loss": 4.3808,
+      "step": 2594
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336134508912151e-05,
+      "loss": 4.4148,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32893583464021e-05,
+      "loss": 4.4438,
+      "step": 2596
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.321738650571525e-05,
+      "loss": 4.6843,
+      "step": 2597
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.314542960721458e-05,
+      "loss": 4.7642,
+      "step": 2598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.307348769104523e-05,
+      "loss": 4.7066,
+      "step": 2599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.30015607973441e-05,
+      "loss": 5.4881,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.292964896623959e-05,
+      "loss": 4.7982,
+      "step": 2601
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.28577522378518e-05,
+      "loss": 4.518,
+      "step": 2602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278587065229232e-05,
+      "loss": 4.4044,
+      "step": 2603
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.271400424966439e-05,
+      "loss": 4.7324,
+      "step": 2604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.264215307006266e-05,
+      "loss": 5.0853,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.257031715357341e-05,
+      "loss": 4.5254,
+      "step": 2606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.249849654027432e-05,
+      "loss": 4.8855,
+      "step": 2607
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242669127023451e-05,
+      "loss": 4.746,
+      "step": 2608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.235490138351464e-05,
+      "loss": 4.7479,
+      "step": 2609
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.22831269201667e-05,
+      "loss": 4.6687,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.221136792023413e-05,
+      "loss": 4.7851,
+      "step": 2611
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21396244237517e-05,
+      "loss": 4.5404,
+      "step": 2612
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206789647074559e-05,
+      "loss": 4.5363,
+      "step": 2613
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.199618410123322e-05,
+      "loss": 4.499,
+      "step": 2614
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192448735522341e-05,
+      "loss": 4.5675,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.185280627271618e-05,
+      "loss": 4.7072,
+      "step": 2616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.17811408937029e-05,
+      "loss": 4.5219,
+      "step": 2617
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.170949125816608e-05,
+      "loss": 5.0611,
+      "step": 2618
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.163785740607952e-05,
+      "loss": 4.7112,
+      "step": 2619
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.15662393774082e-05,
+      "loss": 5.1847,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.149463721210821e-05,
+      "loss": 4.866,
+      "step": 2621
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142305095012691e-05,
+      "loss": 4.6452,
+      "step": 2622
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135148063140268e-05,
+      "loss": 4.6589,
+      "step": 2623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.127992629586506e-05,
+      "loss": 4.6721,
+      "step": 2624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120838798343463e-05,
+      "loss": 4.7886,
+      "step": 2625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.113686573402312e-05,
+      "loss": 4.5254,
+      "step": 2626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.106535958753317e-05,
+      "loss": 4.5934,
+      "step": 2627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099386958385854e-05,
+      "loss": 4.5545,
+      "step": 2628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.092239576288391e-05,
+      "loss": 4.6968,
+      "step": 2629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.085093816448502e-05,
+      "loss": 4.5644,
+      "step": 2630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.077949682852851e-05,
+      "loss": 4.7963,
+      "step": 2631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070807179487185e-05,
+      "loss": 4.3668,
+      "step": 2632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.063666310336363e-05,
+      "loss": 4.8182,
+      "step": 2633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056527079384308e-05,
+      "loss": 4.5842,
+      "step": 2634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.049389490614054e-05,
+      "loss": 4.6333,
+      "step": 2635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042253548007694e-05,
+      "loss": 4.3372,
+      "step": 2636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035119255546427e-05,
+      "loss": 4.3628,
+      "step": 2637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027986617210507e-05,
+      "loss": 4.5376,
+      "step": 2638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.020855636979286e-05,
+      "loss": 4.4343,
+      "step": 2639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013726318831176e-05,
+      "loss": 4.5689,
+      "step": 2640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.006598666743676e-05,
+      "loss": 4.9216,
+      "step": 2641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999472684693338e-05,
+      "loss": 4.8716,
+      "step": 2642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992348376655794e-05,
+      "loss": 4.4309,
+      "step": 2643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.985225746605743e-05,
+      "loss": 4.3724,
+      "step": 2644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97810479851694e-05,
+      "loss": 4.4862,
+      "step": 2645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.970985536362213e-05,
+      "loss": 4.2283,
+      "step": 2646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.963867964113433e-05,
+      "loss": 4.5674,
+      "step": 2647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956752085741547e-05,
+      "loss": 4.2891,
+      "step": 2648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.949637905216539e-05,
+      "loss": 3.9629,
+      "step": 2649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942525426507463e-05,
+      "loss": 4.5202,
+      "step": 2650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.935414653582408e-05,
+      "loss": 4.7839,
+      "step": 2651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.928305590408524e-05,
+      "loss": 4.5305,
+      "step": 2652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921198240951998e-05,
+      "loss": 4.4183,
+      "step": 2653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914092609178063e-05,
+      "loss": 4.1484,
+      "step": 2654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906988699051e-05,
+      "loss": 4.3844,
+      "step": 2655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899886514534117e-05,
+      "loss": 4.5152,
+      "step": 2656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.892786059589775e-05,
+      "loss": 4.5473,
+      "step": 2657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.885687338179357e-05,
+      "loss": 4.5146,
+      "step": 2658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878590354263286e-05,
+      "loss": 4.1696,
+      "step": 2659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.871495111801008e-05,
+      "loss": 4.3007,
+      "step": 2660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86440161475101e-05,
+      "loss": 4.8862,
+      "step": 2661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.857309867070795e-05,
+      "loss": 4.5694,
+      "step": 2662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850219872716893e-05,
+      "loss": 4.8313,
+      "step": 2663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.843131635644859e-05,
+      "loss": 4.7839,
+      "step": 2664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.836045159809253e-05,
+      "loss": 4.0188,
+      "step": 2665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82896044916368e-05,
+      "loss": 4.8823,
+      "step": 2666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821877507660728e-05,
+      "loss": 4.4418,
+      "step": 2667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.81479633925203e-05,
+      "loss": 4.384,
+      "step": 2668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.807716947888197e-05,
+      "loss": 4.6984,
+      "step": 2669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.800639337518881e-05,
+      "loss": 4.3741,
+      "step": 2670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.793563512092713e-05,
+      "loss": 4.3672,
+      "step": 2671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78648947555735e-05,
+      "loss": 4.4403,
+      "step": 2672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.77941723185943e-05,
+      "loss": 4.3713,
+      "step": 2673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772346784944612e-05,
+      "loss": 4.7194,
+      "step": 2674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.765278138757531e-05,
+      "loss": 4.4787,
+      "step": 2675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758211297241843e-05,
+      "loss": 4.4813,
+      "step": 2676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.751146264340171e-05,
+      "loss": 4.4396,
+      "step": 2677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744083043994141e-05,
+      "loss": 4.342,
+      "step": 2678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737021640144375e-05,
+      "loss": 4.5254,
+      "step": 2679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729962056730465e-05,
+      "loss": 4.5036,
+      "step": 2680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.722904297691005e-05,
+      "loss": 4.4227,
+      "step": 2681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715848366963555e-05,
+      "loss": 4.426,
+      "step": 2682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708794268484669e-05,
+      "loss": 4.5924,
+      "step": 2683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.701742006189869e-05,
+      "loss": 4.6239,
+      "step": 2684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.694691584013655e-05,
+      "loss": 4.7275,
+      "step": 2685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687643005889502e-05,
+      "loss": 4.4038,
+      "step": 2686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680596275749857e-05,
+      "loss": 4.7696,
+      "step": 2687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.673551397526135e-05,
+      "loss": 4.3634,
+      "step": 2688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66650837514871e-05,
+      "loss": 4.628,
+      "step": 2689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.659467212546938e-05,
+      "loss": 4.4198,
+      "step": 2690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652427913649121e-05,
+      "loss": 4.3827,
+      "step": 2691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.64539048238253e-05,
+      "loss": 4.6431,
+      "step": 2692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63835492267339e-05,
+      "loss": 4.2913,
+      "step": 2693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631321238446883e-05,
+      "loss": 4.3769,
+      "step": 2694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.624289433627148e-05,
+      "loss": 4.4719,
+      "step": 2695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.61725951213727e-05,
+      "loss": 4.4035,
+      "step": 2696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610231477899287e-05,
+      "loss": 4.7687,
+      "step": 2697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603205334834181e-05,
+      "loss": 4.4154,
+      "step": 2698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596181086861887e-05,
+      "loss": 4.4254,
+      "step": 2699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.589158737901264e-05,
+      "loss": 4.3391,
+      "step": 2700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582138291870139e-05,
+      "loss": 4.9645,
+      "step": 2701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.575119752685249e-05,
+      "loss": 4.4657,
+      "step": 2702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56810312426229e-05,
+      "loss": 4.1168,
+      "step": 2703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561088410515876e-05,
+      "loss": 4.5901,
+      "step": 2704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554075615359566e-05,
+      "loss": 4.6979,
+      "step": 2705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.547064742705833e-05,
+      "loss": 4.9531,
+      "step": 2706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540055796466098e-05,
+      "loss": 4.238,
+      "step": 2707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.533048780550683e-05,
+      "loss": 4.4259,
+      "step": 2708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526043698868858e-05,
+      "loss": 4.4761,
+      "step": 2709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.51904055532879e-05,
+      "loss": 4.5177,
+      "step": 2710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.512039353837586e-05,
+      "loss": 4.6234,
+      "step": 2711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.505040098301254e-05,
+      "loss": 4.5114,
+      "step": 2712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498042792624722e-05,
+      "loss": 4.4457,
+      "step": 2713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.491047440711833e-05,
+      "loss": 4.4168,
+      "step": 2714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484054046465333e-05,
+      "loss": 4.4622,
+      "step": 2715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477062613786885e-05,
+      "loss": 4.606,
+      "step": 2716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47007314657705e-05,
+      "loss": 4.5866,
+      "step": 2717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463085648735295e-05,
+      "loss": 4.9829,
+      "step": 2718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.456100124159988e-05,
+      "loss": 4.4634,
+      "step": 2719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4491165767484e-05,
+      "loss": 4.4487,
+      "step": 2720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44213501039669e-05,
+      "loss": 4.1891,
+      "step": 2721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.435155428999922e-05,
+      "loss": 4.6031,
+      "step": 2722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428177836452045e-05,
+      "loss": 4.4461,
+      "step": 2723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4212022366459e-05,
+      "loss": 4.4593,
+      "step": 2724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414228633473222e-05,
+      "loss": 4.591,
+      "step": 2725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.407257030824623e-05,
+      "loss": 4.561,
+      "step": 2726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400287432589604e-05,
+      "loss": 4.4506,
+      "step": 2727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393319842656549e-05,
+      "loss": 5.3523,
+      "step": 2728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.38635426491272e-05,
+      "loss": 4.5791,
+      "step": 2729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.379390703244254e-05,
+      "loss": 4.4062,
+      "step": 2730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372429161536166e-05,
+      "loss": 4.3621,
+      "step": 2731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365469643672342e-05,
+      "loss": 4.5032,
+      "step": 2732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358512153535542e-05,
+      "loss": 4.6992,
+      "step": 2733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.351556695007397e-05,
+      "loss": 4.5259,
+      "step": 2734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.344603271968388e-05,
+      "loss": 4.7183,
+      "step": 2735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.337651888297886e-05,
+      "loss": 4.2802,
+      "step": 2736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3307025478741e-05,
+      "loss": 4.1536,
+      "step": 2737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.323755254574124e-05,
+      "loss": 4.6292,
+      "step": 2738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31681001227388e-05,
+      "loss": 4.4841,
+      "step": 2739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309866824848174e-05,
+      "loss": 4.4641,
+      "step": 2740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302925696170645e-05,
+      "loss": 4.4404,
+      "step": 2741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295986630113802e-05,
+      "loss": 4.8312,
+      "step": 2742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.28904963054898e-05,
+      "loss": 4.4293,
+      "step": 2743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282114701346385e-05,
+      "loss": 4.5701,
+      "step": 2744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275181846375049e-05,
+      "loss": 4.3182,
+      "step": 2745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26825106950286e-05,
+      "loss": 4.5493,
+      "step": 2746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26132237459654e-05,
+      "loss": 4.5713,
+      "step": 2747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254395765521651e-05,
+      "loss": 4.4307,
+      "step": 2748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247471246142591e-05,
+      "loss": 4.2252,
+      "step": 2749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.24054882032259e-05,
+      "loss": 4.453,
+      "step": 2750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233628491923718e-05,
+      "loss": 4.2567,
+      "step": 2751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.226710264806861e-05,
+      "loss": 4.3081,
+      "step": 2752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.219794142831749e-05,
+      "loss": 4.7538,
+      "step": 2753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.212880129856922e-05,
+      "loss": 4.721,
+      "step": 2754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.205968229739759e-05,
+      "loss": 4.1956,
+      "step": 2755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199058446336443e-05,
+      "loss": 3.8165,
+      "step": 2756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19215078350199e-05,
+      "loss": 4.6638,
+      "step": 2757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.185245245090231e-05,
+      "loss": 4.2232,
+      "step": 2758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1783418349538e-05,
+      "loss": 4.7961,
+      "step": 2759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171440556944161e-05,
+      "loss": 4.568,
+      "step": 2760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164541414911574e-05,
+      "loss": 4.5154,
+      "step": 2761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.157644412705117e-05,
+      "loss": 4.4618,
+      "step": 2762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.150749554172665e-05,
+      "loss": 4.1243,
+      "step": 2763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14385684316091e-05,
+      "loss": 4.8025,
+      "step": 2764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13696628351533e-05,
+      "loss": 4.5266,
+      "step": 2765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13007787908022e-05,
+      "loss": 4.5024,
+      "step": 2766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123191633698652e-05,
+      "loss": 4.3698,
+      "step": 2767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.11630755121252e-05,
+      "loss": 4.2227,
+      "step": 2768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109425635462483e-05,
+      "loss": 4.3299,
+      "step": 2769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1025458902880085e-05,
+      "loss": 4.5341,
+      "step": 2770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0956683195273524e-05,
+      "loss": 4.4299,
+      "step": 2771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088792927017548e-05,
+      "loss": 3.9035,
+      "step": 2772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.081919716594425e-05,
+      "loss": 4.3841,
+      "step": 2773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.075048692092586e-05,
+      "loss": 4.561,
+      "step": 2774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.068179857345424e-05,
+      "loss": 4.4067,
+      "step": 2775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.061313216185096e-05,
+      "loss": 4.3693,
+      "step": 2776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.054448772442551e-05,
+      "loss": 4.684,
+      "step": 2777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.047586529947502e-05,
+      "loss": 4.1178,
+      "step": 2778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0407264925284367e-05,
+      "loss": 4.2131,
+      "step": 2779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0338686640126116e-05,
+      "loss": 4.4698,
+      "step": 2780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027013048226056e-05,
+      "loss": 4.5544,
+      "step": 2781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.020159648993555e-05,
+      "loss": 4.3229,
+      "step": 2782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.013308470138663e-05,
+      "loss": 4.1449,
+      "step": 2783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0064595154836986e-05,
+      "loss": 4.3525,
+      "step": 2784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.999612788849733e-05,
+      "loss": 4.9818,
+      "step": 2785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.992768294056601e-05,
+      "loss": 4.4686,
+      "step": 2786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985926034922884e-05,
+      "loss": 4.1854,
+      "step": 2787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9790860152659245e-05,
+      "loss": 4.2646,
+      "step": 2788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9722482389018096e-05,
+      "loss": 4.8606,
+      "step": 2789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9654127096453806e-05,
+      "loss": 4.239,
+      "step": 2790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.958579431310216e-05,
+      "loss": 4.898,
+      "step": 2791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951748407708651e-05,
+      "loss": 4.3248,
+      "step": 2792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.944919642651754e-05,
+      "loss": 4.4395,
+      "step": 2793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938093139949327e-05,
+      "loss": 4.2997,
+      "step": 2794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.931268903409931e-05,
+      "loss": 4.4263,
+      "step": 2795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924446936840838e-05,
+      "loss": 4.5648,
+      "step": 2796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.917627244048078e-05,
+      "loss": 4.6199,
+      "step": 2797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9108098288363855e-05,
+      "loss": 4.851,
+      "step": 2798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903994695009254e-05,
+      "loss": 4.4498,
+      "step": 2799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.897181846368875e-05,
+      "loss": 4.4139,
+      "step": 2800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.890371286716192e-05,
+      "loss": 4.1816,
+      "step": 2801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883563019850845e-05,
+      "loss": 4.6891,
+      "step": 2802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876757049571221e-05,
+      "loss": 4.6654,
+      "step": 2803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.869953379674405e-05,
+      "loss": 4.1099,
+      "step": 2804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.863152013956208e-05,
+      "loss": 4.6648,
+      "step": 2805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.856352956211155e-05,
+      "loss": 4.3609,
+      "step": 2806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849556210232482e-05,
+      "loss": 4.3126,
+      "step": 2807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.842761779812138e-05,
+      "loss": 4.7351,
+      "step": 2808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.83596966874077e-05,
+      "loss": 4.6476,
+      "step": 2809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8291798808077513e-05,
+      "loss": 4.1624,
+      "step": 2810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8223924198011396e-05,
+      "loss": 4.3305,
+      "step": 2811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.815607289507701e-05,
+      "loss": 4.7168,
+      "step": 2812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8088244937129e-05,
+      "loss": 5.0462,
+      "step": 2813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.802044036200907e-05,
+      "loss": 4.165,
+      "step": 2814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795265920754581e-05,
+      "loss": 4.5615,
+      "step": 2815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7884901511554633e-05,
+      "loss": 4.7102,
+      "step": 2816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.78171673118381e-05,
+      "loss": 4.3615,
+      "step": 2817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774945664618546e-05,
+      "loss": 4.2682,
+      "step": 2818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768176955237303e-05,
+      "loss": 4.5009,
+      "step": 2819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.761410606816371e-05,
+      "loss": 4.193,
+      "step": 2820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7546466231307475e-05,
+      "loss": 4.2599,
+      "step": 2821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7478850079540994e-05,
+      "loss": 4.2615,
+      "step": 2822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7411257650587703e-05,
+      "loss": 4.4478,
+      "step": 2823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.734368898215785e-05,
+      "loss": 4.2127,
+      "step": 2824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727614411194844e-05,
+      "loss": 4.8663,
+      "step": 2825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.720862307764316e-05,
+      "loss": 4.6175,
+      "step": 2826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714112591691243e-05,
+      "loss": 4.3788,
+      "step": 2827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.707365266741331e-05,
+      "loss": 4.4363,
+      "step": 2828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700620336678952e-05,
+      "loss": 4.1782,
+      "step": 2829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.69387780526715e-05,
+      "loss": 4.846,
+      "step": 2830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.687137676267623e-05,
+      "loss": 3.8176,
+      "step": 2831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6803999534407314e-05,
+      "loss": 4.113,
+      "step": 2832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6736646405454866e-05,
+      "loss": 4.4402,
+      "step": 2833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666931741339569e-05,
+      "loss": 4.4743,
+      "step": 2834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.660201259579304e-05,
+      "loss": 4.5538,
+      "step": 2835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653473199019668e-05,
+      "loss": 4.5769,
+      "step": 2836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646747563414281e-05,
+      "loss": 4.2593,
+      "step": 2837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6400243565154274e-05,
+      "loss": 4.2015,
+      "step": 2838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.633303582074021e-05,
+      "loss": 4.6625,
+      "step": 2839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6265852438396235e-05,
+      "loss": 4.6181,
+      "step": 2840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.619869345560439e-05,
+      "loss": 4.2946,
+      "step": 2841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6131558909833035e-05,
+      "loss": 4.5264,
+      "step": 2842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6064448838537033e-05,
+      "loss": 4.5936,
+      "step": 2843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5997363279157454e-05,
+      "loss": 4.2785,
+      "step": 2844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593030226912177e-05,
+      "loss": 4.2311,
+      "step": 2845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5863265845843695e-05,
+      "loss": 4.4043,
+      "step": 2846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579625404672335e-05,
+      "loss": 4.4898,
+      "step": 2847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5729266909146926e-05,
+      "loss": 4.2725,
+      "step": 2848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5662304470487034e-05,
+      "loss": 4.7166,
+      "step": 2849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.559536676810242e-05,
+      "loss": 4.5537,
+      "step": 2850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5528453839338046e-05,
+      "loss": 4.3106,
+      "step": 2851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5461565721525035e-05,
+      "loss": 4.4212,
+      "step": 2852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539470245198064e-05,
+      "loss": 4.7796,
+      "step": 2853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532786406800841e-05,
+      "loss": 4.5383,
+      "step": 2854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5261050606897794e-05,
+      "loss": 4.3671,
+      "step": 2855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.519426210592449e-05,
+      "loss": 4.4237,
+      "step": 2856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5127498602350145e-05,
+      "loss": 4.3084,
+      "step": 2857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506076013342263e-05,
+      "loss": 4.2721,
+      "step": 2858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4994046736375704e-05,
+      "loss": 4.0473,
+      "step": 2859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.492735844842918e-05,
+      "loss": 4.2366,
+      "step": 2860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4860695306788836e-05,
+      "loss": 4.4937,
+      "step": 2861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4794057348646534e-05,
+      "loss": 4.2506,
+      "step": 2862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.472744461117994e-05,
+      "loss": 4.4229,
+      "step": 2863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.466085713155275e-05,
+      "loss": 4.2504,
+      "step": 2864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459429494691451e-05,
+      "loss": 4.3107,
+      "step": 2865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452775809440062e-05,
+      "loss": 3.7714,
+      "step": 2866
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.446124661113251e-05,
+      "loss": 4.3122,
+      "step": 2867
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4394760534217304e-05,
+      "loss": 4.6499,
+      "step": 2868
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432829990074799e-05,
+      "loss": 4.9666,
+      "step": 2869
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426186474780333e-05,
+      "loss": 4.4237,
+      "step": 2870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.419545511244799e-05,
+      "loss": 4.3837,
+      "step": 2871
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4129071031732284e-05,
+      "loss": 4.5572,
+      "step": 2872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406271254269229e-05,
+      "loss": 4.4573,
+      "step": 2873
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3996379682349806e-05,
+      "loss": 4.0182,
+      "step": 2874
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393007248771233e-05,
+      "loss": 4.3157,
+      "step": 2875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.386379099577319e-05,
+      "loss": 4.005,
+      "step": 2876
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3797535243511046e-05,
+      "loss": 4.6683,
+      "step": 2877
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373130526789054e-05,
+      "loss": 4.2811,
+      "step": 2878
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.366510110586168e-05,
+      "loss": 4.5138,
+      "step": 2879
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.359892279436033e-05,
+      "loss": 4.3207,
+      "step": 2880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.353277037030759e-05,
+      "loss": 4.362,
+      "step": 2881
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.346664387061045e-05,
+      "loss": 4.8447,
+      "step": 2882
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340054333216122e-05,
+      "loss": 4.5484,
+      "step": 2883
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333446879183789e-05,
+      "loss": 4.9001,
+      "step": 2884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.326842028650376e-05,
+      "loss": 4.7231,
+      "step": 2885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.320239785300769e-05,
+      "loss": 4.2657,
+      "step": 2886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313640152818409e-05,
+      "loss": 4.4361,
+      "step": 2887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3070431348852665e-05,
+      "loss": 4.9508,
+      "step": 2888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300448735181862e-05,
+      "loss": 4.4386,
+      "step": 2889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.293856957387244e-05,
+      "loss": 4.247,
+      "step": 2890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.287267805179017e-05,
+      "loss": 4.6392,
+      "step": 2891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280681282233305e-05,
+      "loss": 4.6409,
+      "step": 2892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2740973922247706e-05,
+      "loss": 4.4216,
+      "step": 2893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.267516138826601e-05,
+      "loss": 4.4015,
+      "step": 2894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2609375257105274e-05,
+      "loss": 4.4124,
+      "step": 2895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2543615565467965e-05,
+      "loss": 4.3027,
+      "step": 2896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2477882350041805e-05,
+      "loss": 4.4248,
+      "step": 2897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.241217564749976e-05,
+      "loss": 3.9936,
+      "step": 2898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2346495494499966e-05,
+      "loss": 4.2374,
+      "step": 2899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2280841927685874e-05,
+      "loss": 4.4091,
+      "step": 2900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221521498368599e-05,
+      "loss": 4.4485,
+      "step": 2901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.214961469911397e-05,
+      "loss": 4.4613,
+      "step": 2902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208404111056858e-05,
+      "loss": 4.1922,
+      "step": 2903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201849425463383e-05,
+      "loss": 4.4887,
+      "step": 2904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195297416787867e-05,
+      "loss": 4.1988,
+      "step": 2905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1887480886857156e-05,
+      "loss": 4.5064,
+      "step": 2906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1822014448108345e-05,
+      "loss": 4.5239,
+      "step": 2907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1756574888156486e-05,
+      "loss": 4.1677,
+      "step": 2908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.169116224351067e-05,
+      "loss": 4.4924,
+      "step": 2909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162577655066494e-05,
+      "loss": 4.2377,
+      "step": 2910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1560417846098475e-05,
+      "loss": 4.0818,
+      "step": 2911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149508616627522e-05,
+      "loss": 4.667,
+      "step": 2912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1429781547644254e-05,
+      "loss": 4.747,
+      "step": 2913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136450402663926e-05,
+      "loss": 4.3091,
+      "step": 2914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.129925363967911e-05,
+      "loss": 4.5486,
+      "step": 2915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12340304231673e-05,
+      "loss": 4.3603,
+      "step": 2916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1168834413492384e-05,
+      "loss": 4.428,
+      "step": 2917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1103665647027476e-05,
+      "loss": 4.3044,
+      "step": 2918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1038524160130754e-05,
+      "loss": 4.3124,
+      "step": 2919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.097340998914499e-05,
+      "loss": 4.4582,
+      "step": 2920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090832317039782e-05,
+      "loss": 4.4376,
+      "step": 2921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084326374020159e-05,
+      "loss": 4.483,
+      "step": 2922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0778231734853265e-05,
+      "loss": 4.4802,
+      "step": 2923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.071322719063474e-05,
+      "loss": 4.305,
+      "step": 2924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064825014381238e-05,
+      "loss": 3.8301,
+      "step": 2925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.05833006306373e-05,
+      "loss": 4.6008,
+      "step": 2926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051837868734517e-05,
+      "loss": 4.0866,
+      "step": 2927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0453484350156465e-05,
+      "loss": 3.9923,
+      "step": 2928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038861765527607e-05,
+      "loss": 4.1909,
+      "step": 2929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032377863889353e-05,
+      "loss": 4.2568,
+      "step": 2930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.025896733718288e-05,
+      "loss": 4.9351,
+      "step": 2931
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019418378630284e-05,
+      "loss": 4.2854,
+      "step": 2932
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0129428022396505e-05,
+      "loss": 4.6616,
+      "step": 2933
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.006470008159153e-05,
+      "loss": 4.5051,
+      "step": 2934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 4.8397,
+      "step": 2935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.993532781371852e-05,
+      "loss": 4.3047,
+      "step": 2936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987068355882813e-05,
+      "loss": 4.3981,
+      "step": 2937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980606727139422e-05,
+      "loss": 4.4911,
+      "step": 2938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.974147898746664e-05,
+      "loss": 4.1446,
+      "step": 2939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9676918743079536e-05,
+      "loss": 4.1176,
+      "step": 2940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9612386574251565e-05,
+      "loss": 4.0157,
+      "step": 2941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9547882516985575e-05,
+      "loss": 4.1749,
+      "step": 2942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948340660726878e-05,
+      "loss": 4.2274,
+      "step": 2943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.941895888107269e-05,
+      "loss": 4.4934,
+      "step": 2944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935453937435305e-05,
+      "loss": 4.3429,
+      "step": 2945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929014812305002e-05,
+      "loss": 4.3652,
+      "step": 2946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9225785163087735e-05,
+      "loss": 4.3934,
+      "step": 2947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9161450530374786e-05,
+      "loss": 4.3258,
+      "step": 2948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.90971442608038e-05,
+      "loss": 4.0376,
+      "step": 2949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903286639025175e-05,
+      "loss": 4.3676,
+      "step": 2950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8968616954579516e-05,
+      "loss": 4.2492,
+      "step": 2951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890439598963236e-05,
+      "loss": 4.3413,
+      "step": 2952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.884020353123948e-05,
+      "loss": 4.6358,
+      "step": 2953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.877603961521438e-05,
+      "loss": 4.3996,
+      "step": 2954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871190427735437e-05,
+      "loss": 4.2139,
+      "step": 2955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8647797553440964e-05,
+      "loss": 4.4539,
+      "step": 2956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8583719479239775e-05,
+      "loss": 4.14,
+      "step": 2957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851967009050033e-05,
+      "loss": 4.4214,
+      "step": 2958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.845564942295615e-05,
+      "loss": 4.3056,
+      "step": 2959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.839165751232475e-05,
+      "loss": 4.443,
+      "step": 2960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8327694394307686e-05,
+      "loss": 4.0152,
+      "step": 2961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826376010459034e-05,
+      "loss": 4.5219,
+      "step": 2962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.819985467884204e-05,
+      "loss": 4.3555,
+      "step": 2963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.813597815271599e-05,
+      "loss": 3.8482,
+      "step": 2964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807213056184938e-05,
+      "loss": 4.1926,
+      "step": 2965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8008311941863127e-05,
+      "loss": 4.3756,
+      "step": 2966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794452232836204e-05,
+      "loss": 4.3097,
+      "step": 2967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788076175693474e-05,
+      "loss": 4.368,
+      "step": 2968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.781703026315361e-05,
+      "loss": 3.8668,
+      "step": 2969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7753327882574913e-05,
+      "loss": 4.6664,
+      "step": 2970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.768965465073858e-05,
+      "loss": 4.3731,
+      "step": 2971
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7626010603168294e-05,
+      "loss": 4.854,
+      "step": 2972
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7562395775371416e-05,
+      "loss": 4.5309,
+      "step": 2973
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7498810202839136e-05,
+      "loss": 4.3446,
+      "step": 2974
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7435253921046206e-05,
+      "loss": 4.619,
+      "step": 2975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737172696545106e-05,
+      "loss": 4.3249,
+      "step": 2976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.730822937149576e-05,
+      "loss": 4.974,
+      "step": 2977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7244761174606056e-05,
+      "loss": 4.2841,
+      "step": 2978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.718132241019128e-05,
+      "loss": 4.3953,
+      "step": 2979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7117913113644175e-05,
+      "loss": 4.3191,
+      "step": 2980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.70545333203413e-05,
+      "loss": 4.6154,
+      "step": 2981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69911830656426e-05,
+      "loss": 4.3603,
+      "step": 2982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.692786238489157e-05,
+      "loss": 4.081,
+      "step": 2983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6864571313415187e-05,
+      "loss": 4.1199,
+      "step": 2984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680130988652398e-05,
+      "loss": 4.2079,
+      "step": 2985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673807813951189e-05,
+      "loss": 4.4311,
+      "step": 2986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667487610765627e-05,
+      "loss": 4.3879,
+      "step": 2987
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66117038262179e-05,
+      "loss": 4.3154,
+      "step": 2988
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654856133044107e-05,
+      "loss": 5.0319,
+      "step": 2989
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648544865555331e-05,
+      "loss": 4.527,
+      "step": 2990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64223658367656e-05,
+      "loss": 4.5416,
+      "step": 2991
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.635931290927221e-05,
+      "loss": 4.527,
+      "step": 2992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6296289908250745e-05,
+      "loss": 4.3399,
+      "step": 2993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.62332968688622e-05,
+      "loss": 4.5723,
+      "step": 2994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617033382625072e-05,
+      "loss": 4.279,
+      "step": 2995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610740081554381e-05,
+      "loss": 4.265,
+      "step": 2996
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.604449787185212e-05,
+      "loss": 4.4358,
+      "step": 2997
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5981625030269693e-05,
+      "loss": 4.298,
+      "step": 2998
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591878232587362e-05,
+      "loss": 4.6241,
+      "step": 2999
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585596979372424e-05,
+      "loss": 4.3317,
+      "step": 3000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579318746886507e-05,
+      "loss": 4.186,
+      "step": 3001
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57304353863227e-05,
+      "loss": 4.2976,
+      "step": 3002
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566771358110703e-05,
+      "loss": 4.3579,
+      "step": 3003
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560502208821079e-05,
+      "loss": 4.6162,
+      "step": 3004
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554236094261006e-05,
+      "loss": 4.5381,
+      "step": 3005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547973017926382e-05,
+      "loss": 4.0612,
+      "step": 3006
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.541712983311427e-05,
+      "loss": 4.3005,
+      "step": 3007
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.535455993908637e-05,
+      "loss": 4.262,
+      "step": 3008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529202053208838e-05,
+      "loss": 5.0075,
+      "step": 3009
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522951164701132e-05,
+      "loss": 4.165,
+      "step": 3010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516703331872943e-05,
+      "loss": 4.4241,
+      "step": 3011
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.51045855820996e-05,
+      "loss": 4.2514,
+      "step": 3012
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504216847196191e-05,
+      "loss": 4.0814,
+      "step": 3013
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497978202313923e-05,
+      "loss": 4.0844,
+      "step": 3014
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491742627043732e-05,
+      "loss": 4.9712,
+      "step": 3015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485510124864487e-05,
+      "loss": 4.2163,
+      "step": 3016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.479280699253335e-05,
+      "loss": 4.1708,
+      "step": 3017
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4730543536857186e-05,
+      "loss": 4.292,
+      "step": 3018
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4668310916353495e-05,
+      "loss": 4.055,
+      "step": 3019
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460610916574227e-05,
+      "loss": 4.1682,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454393831972619e-05,
+      "loss": 4.2894,
+      "step": 3021
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4481798412990835e-05,
+      "loss": 4.3427,
+      "step": 3022
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441968948020442e-05,
+      "loss": 4.2365,
+      "step": 3023
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.43576115560179e-05,
+      "loss": 4.1324,
+      "step": 3024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4295564675064915e-05,
+      "loss": 4.5365,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423354887196179e-05,
+      "loss": 4.1351,
+      "step": 3026
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.41715641813076e-05,
+      "loss": 4.3347,
+      "step": 3027
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.410961063768393e-05,
+      "loss": 4.1528,
+      "step": 3028
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.404768827565508e-05,
+      "loss": 3.4416,
+      "step": 3029
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398579712976783e-05,
+      "loss": 4.4314,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3923937234551774e-05,
+      "loss": 4.8113,
+      "step": 3031
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386210862451884e-05,
+      "loss": 4.1929,
+      "step": 3032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3800311334163623e-05,
+      "loss": 4.0247,
+      "step": 3033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373854539796316e-05,
+      "loss": 4.2097,
+      "step": 3034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367681085037714e-05,
+      "loss": 4.4099,
+      "step": 3035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361510772584764e-05,
+      "loss": 4.2936,
+      "step": 3036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35534360587991e-05,
+      "loss": 4.0713,
+      "step": 3037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349179588363865e-05,
+      "loss": 4.1598,
+      "step": 3038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.343018723475565e-05,
+      "loss": 4.4294,
+      "step": 3039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.336861014652204e-05,
+      "loss": 4.2805,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.330706465329193e-05,
+      "loss": 4.3434,
+      "step": 3041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324555078940205e-05,
+      "loss": 4.097,
+      "step": 3042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318406858917127e-05,
+      "loss": 4.1885,
+      "step": 3043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3122618086901e-05,
+      "loss": 4.1092,
+      "step": 3044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306119931687473e-05,
+      "loss": 4.7246,
+      "step": 3045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.299981231335848e-05,
+      "loss": 4.358,
+      "step": 3046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.293845711060036e-05,
+      "loss": 4.9592,
+      "step": 3047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287713374283092e-05,
+      "loss": 4.3665,
+      "step": 3048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.281584224426275e-05,
+      "loss": 4.2418,
+      "step": 3049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275458264909075e-05,
+      "loss": 4.0318,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.269335499149211e-05,
+      "loss": 4.4828,
+      "step": 3051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.263215930562607e-05,
+      "loss": 4.1458,
+      "step": 3052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257099562563409e-05,
+      "loss": 4.4944,
+      "step": 3053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250986398563973e-05,
+      "loss": 4.2102,
+      "step": 3054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244876441974879e-05,
+      "loss": 3.996,
+      "step": 3055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238769696204904e-05,
+      "loss": 4.252,
+      "step": 3056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232666164661042e-05,
+      "loss": 4.2062,
+      "step": 3057
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.226565850748485e-05,
+      "loss": 4.6686,
+      "step": 3058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220468757870646e-05,
+      "loss": 4.1752,
+      "step": 3059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2143748894291245e-05,
+      "loss": 4.2053,
+      "step": 3060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208284248823728e-05,
+      "loss": 4.0283,
+      "step": 3061
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.202196839452465e-05,
+      "loss": 4.2378,
+      "step": 3062
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196112664711532e-05,
+      "loss": 4.1455,
+      "step": 3063
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1900317279953384e-05,
+      "loss": 4.2408,
+      "step": 3064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18395403269647e-05,
+      "loss": 3.9232,
+      "step": 3065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1778795822057124e-05,
+      "loss": 4.9017,
+      "step": 3066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171808379912034e-05,
+      "loss": 4.7554,
+      "step": 3067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165740429202606e-05,
+      "loss": 4.5443,
+      "step": 3068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.15967573346277e-05,
+      "loss": 4.4234,
+      "step": 3069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.153614296076058e-05,
+      "loss": 4.0932,
+      "step": 3070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147556120424183e-05,
+      "loss": 4.5197,
+      "step": 3071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141501209887036e-05,
+      "loss": 4.1886,
+      "step": 3072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1354495678427e-05,
+      "loss": 3.9207,
+      "step": 3073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129401197667408e-05,
+      "loss": 3.9132,
+      "step": 3074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.123356102735595e-05,
+      "loss": 4.5732,
+      "step": 3075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.11731428641985e-05,
+      "loss": 4.3925,
+      "step": 3076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1112757520909494e-05,
+      "loss": 4.145,
+      "step": 3077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1052405031178165e-05,
+      "loss": 5.2009,
+      "step": 3078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.099208542867563e-05,
+      "loss": 4.1424,
+      "step": 3079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0931798747054506e-05,
+      "loss": 4.3749,
+      "step": 3080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087154501994921e-05,
+      "loss": 4.5372,
+      "step": 3081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0811324280975525e-05,
+      "loss": 4.0636,
+      "step": 3082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.07511365637311e-05,
+      "loss": 4.5852,
+      "step": 3083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069098190179498e-05,
+      "loss": 4.0807,
+      "step": 3084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.063086032872782e-05,
+      "loss": 4.3943,
+      "step": 3085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.057077187807185e-05,
+      "loss": 4.086,
+      "step": 3086
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051071658335072e-05,
+      "loss": 4.1518,
+      "step": 3087
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.045069447806973e-05,
+      "loss": 4.0554,
+      "step": 3088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039070559571556e-05,
+      "loss": 4.6918,
+      "step": 3089
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0330749969756363e-05,
+      "loss": 4.3856,
+      "step": 3090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.027082763364175e-05,
+      "loss": 4.2891,
+      "step": 3091
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0210938620802796e-05,
+      "loss": 4.3093,
+      "step": 3092
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015108296465194e-05,
+      "loss": 5.256,
+      "step": 3093
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0091260698583035e-05,
+      "loss": 3.966,
+      "step": 3094
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003147185597126e-05,
+      "loss": 3.9968,
+      "step": 3095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997171647017317e-05,
+      "loss": 4.1513,
+      "step": 3096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.991199457452674e-05,
+      "loss": 4.228,
+      "step": 3097
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985230620235112e-05,
+      "loss": 4.3517,
+      "step": 3098
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979265138694686e-05,
+      "loss": 4.3112,
+      "step": 3099
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973303016159569e-05,
+      "loss": 4.6379,
+      "step": 3100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9673442559560746e-05,
+      "loss": 4.2651,
+      "step": 3101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.961388861408628e-05,
+      "loss": 4.4235,
+      "step": 3102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.955436835839781e-05,
+      "loss": 3.9339,
+      "step": 3103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949488182570201e-05,
+      "loss": 4.3219,
+      "step": 3104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9435429049186876e-05,
+      "loss": 4.6283,
+      "step": 3105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937601006202147e-05,
+      "loss": 3.8998,
+      "step": 3106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93166248973559e-05,
+      "loss": 4.4697,
+      "step": 3107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925727358832164e-05,
+      "loss": 4.1705,
+      "step": 3108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.919795616803107e-05,
+      "loss": 4.0612,
+      "step": 3109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913867266957786e-05,
+      "loss": 4.3077,
+      "step": 3110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9079423126036483e-05,
+      "loss": 4.2899,
+      "step": 3111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902020757046275e-05,
+      "loss": 4.3407,
+      "step": 3112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.896102603589332e-05,
+      "loss": 4.6522,
+      "step": 3113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890187855534604e-05,
+      "loss": 4.4238,
+      "step": 3114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8842765161819496e-05,
+      "loss": 4.3852,
+      "step": 3115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878368588829354e-05,
+      "loss": 4.0023,
+      "step": 3116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872464076772884e-05,
+      "loss": 4.1066,
+      "step": 3117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8665629833067004e-05,
+      "loss": 4.1314,
+      "step": 3118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860665311723063e-05,
+      "loss": 4.3502,
+      "step": 3119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854771065312315e-05,
+      "loss": 4.4314,
+      "step": 3120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8488802473628984e-05,
+      "loss": 4.1959,
+      "step": 3121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842992861161334e-05,
+      "loss": 4.1037,
+      "step": 3122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.837108909992233e-05,
+      "loss": 4.0347,
+      "step": 3123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8312283971382835e-05,
+      "loss": 4.1433,
+      "step": 3124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825351325880266e-05,
+      "loss": 4.8595,
+      "step": 3125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819477699497034e-05,
+      "loss": 4.1261,
+      "step": 3126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.813607521265518e-05,
+      "loss": 4.1208,
+      "step": 3127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807740794460725e-05,
+      "loss": 4.1854,
+      "step": 3128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801877522355742e-05,
+      "loss": 4.2329,
+      "step": 3129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796017708221725e-05,
+      "loss": 4.1981,
+      "step": 3130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7901613553279006e-05,
+      "loss": 4.158,
+      "step": 3131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7843084669415606e-05,
+      "loss": 4.7754,
+      "step": 3132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7784590463280676e-05,
+      "loss": 4.5148,
+      "step": 3133
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7726130967508564e-05,
+      "loss": 4.2339,
+      "step": 3134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.766770621471414e-05,
+      "loss": 4.011,
+      "step": 3135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7609316237492944e-05,
+      "loss": 4.3277,
+      "step": 3136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.755096106842105e-05,
+      "loss": 4.0993,
+      "step": 3137
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749264074005527e-05,
+      "loss": 4.3256,
+      "step": 3138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.743435528493283e-05,
+      "loss": 4.6426,
+      "step": 3139
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7376104735571537e-05,
+      "loss": 4.2152,
+      "step": 3140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.731788912446974e-05,
+      "loss": 4.1959,
+      "step": 3141
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.725970848410629e-05,
+      "loss": 4.0881,
+      "step": 3142
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.720156284694053e-05,
+      "loss": 5.2017,
+      "step": 3143
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7143452245412235e-05,
+      "loss": 3.8858,
+      "step": 3144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.708537671194173e-05,
+      "loss": 4.2537,
+      "step": 3145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.70273362789297e-05,
+      "loss": 4.069,
+      "step": 3146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.696933097875726e-05,
+      "loss": 4.2932,
+      "step": 3147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.691136084378586e-05,
+      "loss": 3.9862,
+      "step": 3148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.68534259063575e-05,
+      "loss": 4.1927,
+      "step": 3149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.679552619879439e-05,
+      "loss": 4.4028,
+      "step": 3150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.673766175339915e-05,
+      "loss": 4.4412,
+      "step": 3151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.667983260245471e-05,
+      "loss": 4.1025,
+      "step": 3152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.662203877822428e-05,
+      "loss": 4.0411,
+      "step": 3153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.656428031295144e-05,
+      "loss": 4.4512,
+      "step": 3154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.650655723885999e-05,
+      "loss": 4.2534,
+      "step": 3155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6448869588153954e-05,
+      "loss": 4.0335,
+      "step": 3156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.639121739301764e-05,
+      "loss": 4.1265,
+      "step": 3157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63336006856156e-05,
+      "loss": 4.3986,
+      "step": 3158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.627601949809254e-05,
+      "loss": 4.5415,
+      "step": 3159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6218473862573344e-05,
+      "loss": 4.9513,
+      "step": 3160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.616096381116306e-05,
+      "loss": 4.1752,
+      "step": 3161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.610348937594696e-05,
+      "loss": 4.4397,
+      "step": 3162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6046050588990354e-05,
+      "loss": 4.2111,
+      "step": 3163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.598864748233871e-05,
+      "loss": 4.2739,
+      "step": 3164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.593128008801756e-05,
+      "loss": 4.6449,
+      "step": 3165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.587394843803251e-05,
+      "loss": 4.1135,
+      "step": 3166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.581665256436933e-05,
+      "loss": 4.3108,
+      "step": 3167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.575939249899363e-05,
+      "loss": 4.1821,
+      "step": 3168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5702168273851225e-05,
+      "loss": 4.5233,
+      "step": 3169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5644979920867824e-05,
+      "loss": 4.9178,
+      "step": 3170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5587827471949254e-05,
+      "loss": 4.3614,
+      "step": 3171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5530710958981086e-05,
+      "loss": 4.1308,
+      "step": 3172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.547363041382909e-05,
+      "loss": 4.3106,
+      "step": 3173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.541658586833877e-05,
+      "loss": 4.2776,
+      "step": 3174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.535957735433574e-05,
+      "loss": 4.19,
+      "step": 3175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.53026049036253e-05,
+      "loss": 4.279,
+      "step": 3176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5245668547992735e-05,
+      "loss": 4.3458,
+      "step": 3177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.518876831920328e-05,
+      "loss": 4.3761,
+      "step": 3178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5131904249001854e-05,
+      "loss": 4.0833,
+      "step": 3179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.50750763691133e-05,
+      "loss": 3.9614,
+      "step": 3180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.50182847112422e-05,
+      "loss": 3.5371,
+      "step": 3181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.496152930707305e-05,
+      "loss": 4.091,
+      "step": 3182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4904810188269985e-05,
+      "loss": 4.2429,
+      "step": 3183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484812738647698e-05,
+      "loss": 4.1697,
+      "step": 3184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479148093331768e-05,
+      "loss": 4.327,
+      "step": 3185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.473487086039555e-05,
+      "loss": 3.9223,
+      "step": 3186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467829719929369e-05,
+      "loss": 4.4018,
+      "step": 3187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.462175998157485e-05,
+      "loss": 4.3643,
+      "step": 3188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4565259238781556e-05,
+      "loss": 3.9833,
+      "step": 3189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.450879500243583e-05,
+      "loss": 4.851,
+      "step": 3190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4452367304039534e-05,
+      "loss": 4.2845,
+      "step": 3191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4395976175073973e-05,
+      "loss": 4.2301,
+      "step": 3192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.433962164700012e-05,
+      "loss": 4.2129,
+      "step": 3193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4283303751258463e-05,
+      "loss": 4.3386,
+      "step": 3194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422702251926918e-05,
+      "loss": 4.19,
+      "step": 3195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.41707779824319e-05,
+      "loss": 4.484,
+      "step": 3196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4114570172125794e-05,
+      "loss": 4.0918,
+      "step": 3197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4058399119709496e-05,
+      "loss": 4.0019,
+      "step": 3198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.400226485652127e-05,
+      "loss": 4.7569,
+      "step": 3199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.394616741387876e-05,
+      "loss": 4.1806,
+      "step": 3200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.389010682307898e-05,
+      "loss": 4.2713,
+      "step": 3201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3834083115398585e-05,
+      "loss": 4.5225,
+      "step": 3202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.377809632209348e-05,
+      "loss": 4.3443,
+      "step": 3203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.372214647439914e-05,
+      "loss": 4.1136,
+      "step": 3204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.366623360353021e-05,
+      "loss": 4.1268,
+      "step": 3205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.361035774068092e-05,
+      "loss": 4.1296,
+      "step": 3206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355451891702468e-05,
+      "loss": 4.7184,
+      "step": 3207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3498717163714445e-05,
+      "loss": 3.9636,
+      "step": 3208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.34429525118822e-05,
+      "loss": 4.0813,
+      "step": 3209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.33872249926395e-05,
+      "loss": 4.7013,
+      "step": 3210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3331534637077045e-05,
+      "loss": 3.9697,
+      "step": 3211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3275881476264826e-05,
+      "loss": 4.5769,
+      "step": 3212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3220265541252083e-05,
+      "loss": 4.0305,
+      "step": 3213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.316468686306725e-05,
+      "loss": 4.5942,
+      "step": 3214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3109145472718104e-05,
+      "loss": 4.0367,
+      "step": 3215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3053641401191494e-05,
+      "loss": 4.4892,
+      "step": 3216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.299817467945346e-05,
+      "loss": 4.5652,
+      "step": 3217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2942745338449235e-05,
+      "loss": 3.7937,
+      "step": 3218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288735340910323e-05,
+      "loss": 4.5812,
+      "step": 3219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.283199892231891e-05,
+      "loss": 4.2829,
+      "step": 3220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2776681908978915e-05,
+      "loss": 3.9512,
+      "step": 3221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.272140239994492e-05,
+      "loss": 4.1639,
+      "step": 3222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.266616042605769e-05,
+      "loss": 4.4611,
+      "step": 3223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2610956018137126e-05,
+      "loss": 3.9743,
+      "step": 3224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2555789206982076e-05,
+      "loss": 4.0153,
+      "step": 3225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2500660023370455e-05,
+      "loss": 4.4846,
+      "step": 3226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244556849805913e-05,
+      "loss": 4.0575,
+      "step": 3227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2390514661784076e-05,
+      "loss": 4.41,
+      "step": 3228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2335498545260136e-05,
+      "loss": 4.2383,
+      "step": 3229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228052017918114e-05,
+      "loss": 4.0507,
+      "step": 3230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.222557959421984e-05,
+      "loss": 4.3638,
+      "step": 3231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2170676821027966e-05,
+      "loss": 4.5708,
+      "step": 3232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.211581189023611e-05,
+      "loss": 4.2848,
+      "step": 3233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2060984832453755e-05,
+      "loss": 4.0871,
+      "step": 3234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.200619567826925e-05,
+      "loss": 3.9217,
+      "step": 3235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.195144445824976e-05,
+      "loss": 4.227,
+      "step": 3236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1896731202941444e-05,
+      "loss": 4.3386,
+      "step": 3237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.184205594286903e-05,
+      "loss": 4.5941,
+      "step": 3238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.178741870853629e-05,
+      "loss": 4.0806,
+      "step": 3239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.173281953042559e-05,
+      "loss": 4.0798,
+      "step": 3240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1678258438998274e-05,
+      "loss": 4.1028,
+      "step": 3241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.162373546469416e-05,
+      "loss": 4.5938,
+      "step": 3242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.156925063793207e-05,
+      "loss": 3.98,
+      "step": 3243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1514803989109355e-05,
+      "loss": 4.4931,
+      "step": 3244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.146039554860223e-05,
+      "loss": 4.2763,
+      "step": 3245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.140602534676541e-05,
+      "loss": 4.2996,
+      "step": 3246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135169341393236e-05,
+      "loss": 4.4837,
+      "step": 3247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.129739978041528e-05,
+      "loss": 4.4713,
+      "step": 3248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.124314447650488e-05,
+      "loss": 4.1862,
+      "step": 3249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1188927532470534e-05,
+      "loss": 4.0547,
+      "step": 3250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.113474897856019e-05,
+      "loss": 4.0823,
+      "step": 3251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1080608845000445e-05,
+      "loss": 4.1752,
+      "step": 3252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10265071619964e-05,
+      "loss": 3.9998,
+      "step": 3253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.097244395973171e-05,
+      "loss": 4.2616,
+      "step": 3254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0918419268368536e-05,
+      "loss": 4.0378,
+      "step": 3255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.086443311804767e-05,
+      "loss": 4.4755,
+      "step": 3256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0810485538888255e-05,
+      "loss": 4.3623,
+      "step": 3257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.075657656098801e-05,
+      "loss": 4.2606,
+      "step": 3258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.070270621442307e-05,
+      "loss": 4.1833,
+      "step": 3259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.064887452924802e-05,
+      "loss": 3.9379,
+      "step": 3260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.059508153549595e-05,
+      "loss": 4.4436,
+      "step": 3261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0541327263178265e-05,
+      "loss": 4.252,
+      "step": 3262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0487611742284804e-05,
+      "loss": 3.9424,
+      "step": 3263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0433935002783763e-05,
+      "loss": 4.4936,
+      "step": 3264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.038029707462181e-05,
+      "loss": 4.0442,
+      "step": 3265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032669798772382e-05,
+      "loss": 4.1535,
+      "step": 3266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0273137771993053e-05,
+      "loss": 4.3051,
+      "step": 3267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0219616457311074e-05,
+      "loss": 4.2399,
+      "step": 3268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.016613407353781e-05,
+      "loss": 4.104,
+      "step": 3269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.011269065051142e-05,
+      "loss": 4.1226,
+      "step": 3270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0059286218048223e-05,
+      "loss": 4.284,
+      "step": 3271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0005920805942977e-05,
+      "loss": 3.876,
+      "step": 3272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9952594443968506e-05,
+      "loss": 4.1336,
+      "step": 3273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9899307161876022e-05,
+      "loss": 3.975,
+      "step": 3274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9846058989394697e-05,
+      "loss": 4.0969,
+      "step": 3275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9792849956232128e-05,
+      "loss": 4.2167,
+      "step": 3276
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9739680092073906e-05,
+      "loss": 4.3401,
+      "step": 3277
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9686549426583853e-05,
+      "loss": 4.0129,
+      "step": 3278
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9633457989403857e-05,
+      "loss": 4.2882,
+      "step": 3279
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.958040581015401e-05,
+      "loss": 4.3796,
+      "step": 3280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.952739291843244e-05,
+      "loss": 4.093,
+      "step": 3281
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.947441934381535e-05,
+      "loss": 4.4665,
+      "step": 3282
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.942148511585704e-05,
+      "loss": 4.0242,
+      "step": 3283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9368590264089814e-05,
+      "loss": 3.9706,
+      "step": 3284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.931573481802409e-05,
+      "loss": 4.287,
+      "step": 3285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9262918807148222e-05,
+      "loss": 4.0963,
+      "step": 3286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9210142260928586e-05,
+      "loss": 4.1468,
+      "step": 3287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.915740520880952e-05,
+      "loss": 4.1403,
+      "step": 3288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9104707680213405e-05,
+      "loss": 4.2807,
+      "step": 3289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9052049704540508e-05,
+      "loss": 4.6095,
+      "step": 3290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8999431311169012e-05,
+      "loss": 3.8676,
+      "step": 3291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.894685252945507e-05,
+      "loss": 4.2407,
+      "step": 3292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8894313388732664e-05,
+      "loss": 4.0769,
+      "step": 3293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8841813918313776e-05,
+      "loss": 4.7156,
+      "step": 3294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8789354147488155e-05,
+      "loss": 4.2486,
+      "step": 3295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8736934105523428e-05,
+      "loss": 4.2172,
+      "step": 3296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.868455382166504e-05,
+      "loss": 4.3299,
+      "step": 3297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8632213325136347e-05,
+      "loss": 4.0308,
+      "step": 3298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.857991264513834e-05,
+      "loss": 4.1967,
+      "step": 3299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8527651810849964e-05,
+      "loss": 4.1081,
+      "step": 3300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.847543085142781e-05,
+      "loss": 3.9886,
+      "step": 3301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8423249796006368e-05,
+      "loss": 4.1327,
+      "step": 3302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.837110867369769e-05,
+      "loss": 4.5065,
+      "step": 3303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.831900751359161e-05,
+      "loss": 3.5818,
+      "step": 3304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8266946344755763e-05,
+      "loss": 4.157,
+      "step": 3305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8214925196235364e-05,
+      "loss": 4.4117,
+      "step": 3306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8162944097053335e-05,
+      "loss": 4.1408,
+      "step": 3307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8111003076210218e-05,
+      "loss": 4.219,
+      "step": 3308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8059102162684305e-05,
+      "loss": 4.4919,
+      "step": 3309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8007241385431393e-05,
+      "loss": 4.0523,
+      "step": 3310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7955420773384945e-05,
+      "loss": 3.9567,
+      "step": 3311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.790364035545595e-05,
+      "loss": 4.3889,
+      "step": 3312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7851900160533106e-05,
+      "loss": 4.1801,
+      "step": 3313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7800200217482552e-05,
+      "loss": 4.1009,
+      "step": 3314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7748540555148006e-05,
+      "loss": 3.8548,
+      "step": 3315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7696921202350713e-05,
+      "loss": 3.9016,
+      "step": 3316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76453421878894e-05,
+      "loss": 3.9836,
+      "step": 3317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7593803540540375e-05,
+      "loss": 4.1744,
+      "step": 3318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7542305289057345e-05,
+      "loss": 3.988,
+      "step": 3319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.74908474621715e-05,
+      "loss": 4.6102,
+      "step": 3320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7439430088591456e-05,
+      "loss": 4.1804,
+      "step": 3321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7388053197003337e-05,
+      "loss": 4.4903,
+      "step": 3322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.733671681607062e-05,
+      "loss": 4.1405,
+      "step": 3323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.728542097443415e-05,
+      "loss": 4.1373,
+      "step": 3324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.72341657007122e-05,
+      "loss": 4.0444,
+      "step": 3325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7182951023500458e-05,
+      "loss": 4.0144,
+      "step": 3326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7131776971371903e-05,
+      "loss": 4.6018,
+      "step": 3327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7080643572876775e-05,
+      "loss": 4.1818,
+      "step": 3328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7029550856542785e-05,
+      "loss": 4.0924,
+      "step": 3329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6978498850874845e-05,
+      "loss": 4.1806,
+      "step": 3330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6927487584355247e-05,
+      "loss": 4.6095,
+      "step": 3331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6876517085443397e-05,
+      "loss": 3.9989,
+      "step": 3332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6825587382576123e-05,
+      "loss": 4.1575,
+      "step": 3333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6774698504167374e-05,
+      "loss": 4.0382,
+      "step": 3334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.672385047860846e-05,
+      "loss": 4.2996,
+      "step": 3335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6673043334267676e-05,
+      "loss": 4.4877,
+      "step": 3336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6622277099490757e-05,
+      "loss": 4.2093,
+      "step": 3337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6571551802600446e-05,
+      "loss": 4.1677,
+      "step": 3338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6520867471896714e-05,
+      "loss": 4.2452,
+      "step": 3339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6470224135656675e-05,
+      "loss": 4.3769,
+      "step": 3340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.64196218221345e-05,
+      "loss": 3.9443,
+      "step": 3341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.636906055956162e-05,
+      "loss": 4.0986,
+      "step": 3342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.631854037614643e-05,
+      "loss": 4.0973,
+      "step": 3343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6268061300074466e-05,
+      "loss": 4.2114,
+      "step": 3344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.621762335950828e-05,
+      "loss": 4.0971,
+      "step": 3345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6167226582587567e-05,
+      "loss": 4.1857,
+      "step": 3346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.611687099742899e-05,
+      "loss": 4.1443,
+      "step": 3347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6066556632126226e-05,
+      "loss": 4.1624,
+      "step": 3348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.601628351474994e-05,
+      "loss": 3.9142,
+      "step": 3349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.596605167334789e-05,
+      "loss": 4.1796,
+      "step": 3350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5915861135944695e-05,
+      "loss": 4.0698,
+      "step": 3351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5865711930541958e-05,
+      "loss": 3.8041,
+      "step": 3352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5815604085118237e-05,
+      "loss": 4.4194,
+      "step": 3353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576553762762899e-05,
+      "loss": 3.8501,
+      "step": 3354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5715512586006664e-05,
+      "loss": 4.4486,
+      "step": 3355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.566552898816048e-05,
+      "loss": 4.3392,
+      "step": 3356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5615586861976635e-05,
+      "loss": 4.2911,
+      "step": 3357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55656862353181e-05,
+      "loss": 4.189,
+      "step": 3358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5515827136024815e-05,
+      "loss": 4.306,
+      "step": 3359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.546600959191343e-05,
+      "loss": 3.6244,
+      "step": 3360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5416233630777497e-05,
+      "loss": 4.255,
+      "step": 3361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.53664992803873e-05,
+      "loss": 4.3406,
+      "step": 3362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5316806568489924e-05,
+      "loss": 4.0983,
+      "step": 3363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.526715552280934e-05,
+      "loss": 4.5307,
+      "step": 3364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5217546171046035e-05,
+      "loss": 4.2382,
+      "step": 3365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.516797854087748e-05,
+      "loss": 4.1765,
+      "step": 3366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5118452659957703e-05,
+      "loss": 4.0734,
+      "step": 3367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5068968555917572e-05,
+      "loss": 4.4871,
+      "step": 3368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5019526256364477e-05,
+      "loss": 4.0413,
+      "step": 3369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.497012578888267e-05,
+      "loss": 4.0191,
+      "step": 3370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4920767181032913e-05,
+      "loss": 4.1321,
+      "step": 3371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.487145046035275e-05,
+      "loss": 4.4509,
+      "step": 3372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4822175654356238e-05,
+      "loss": 4.4642,
+      "step": 3373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.477294279053406e-05,
+      "loss": 3.8419,
+      "step": 3374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.472375189635362e-05,
+      "loss": 3.6924,
+      "step": 3375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.467460299925878e-05,
+      "loss": 4.0309,
+      "step": 3376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4625496126670012e-05,
+      "loss": 4.343,
+      "step": 3377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4576431305984325e-05,
+      "loss": 3.927,
+      "step": 3378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4527408564575337e-05,
+      "loss": 4.2071,
+      "step": 3379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4478427929793114e-05,
+      "loss": 4.0713,
+      "step": 3380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4429489428964258e-05,
+      "loss": 4.0213,
+      "step": 3381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4380593089391835e-05,
+      "loss": 4.042,
+      "step": 3382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4331738938355462e-05,
+      "loss": 4.2239,
+      "step": 3383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4282927003111157e-05,
+      "loss": 4.0387,
+      "step": 3384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4234157310891393e-05,
+      "loss": 4.1596,
+      "step": 3385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4185429888905086e-05,
+      "loss": 3.8732,
+      "step": 3386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4136744764337537e-05,
+      "loss": 4.3327,
+      "step": 3387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.408810196435055e-05,
+      "loss": 4.589,
+      "step": 3388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4039501516082198e-05,
+      "loss": 4.1139,
+      "step": 3389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3990943446646973e-05,
+      "loss": 3.9164,
+      "step": 3390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3942427783135702e-05,
+      "loss": 4.1356,
+      "step": 3391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3893954552615638e-05,
+      "loss": 3.912,
+      "step": 3392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3845523782130254e-05,
+      "loss": 3.9308,
+      "step": 3393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3797135498699384e-05,
+      "loss": 4.1484,
+      "step": 3394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.374878972931912e-05,
+      "loss": 4.2664,
+      "step": 3395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3700486500961925e-05,
+      "loss": 4.4574,
+      "step": 3396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.365222584057648e-05,
+      "loss": 4.3067,
+      "step": 3397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360400777508761e-05,
+      "loss": 4.5782,
+      "step": 3398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3555832331396555e-05,
+      "loss": 4.5458,
+      "step": 3399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3507699536380645e-05,
+      "loss": 4.6448,
+      "step": 3400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3459609416893536e-05,
+      "loss": 4.6322,
+      "step": 3401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.341156199976491e-05,
+      "loss": 4.4978,
+      "step": 3402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33635573118008e-05,
+      "loss": 3.9282,
+      "step": 3403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.331559537978325e-05,
+      "loss": 4.3936,
+      "step": 3404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3267676230470613e-05,
+      "loss": 4.1368,
+      "step": 3405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.321979989059716e-05,
+      "loss": 4.0943,
+      "step": 3406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31719663868735e-05,
+      "loss": 4.5754,
+      "step": 3407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31241757459862e-05,
+      "loss": 4.3728,
+      "step": 3408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.307642799459795e-05,
+      "loss": 4.1502,
+      "step": 3409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.302872315934752e-05,
+      "loss": 4.6975,
+      "step": 3410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2981061266849713e-05,
+      "loss": 4.0191,
+      "step": 3411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2933442343695434e-05,
+      "loss": 4.2831,
+      "step": 3412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2885866416451573e-05,
+      "loss": 4.0685,
+      "step": 3413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2838333511661014e-05,
+      "loss": 4.1801,
+      "step": 3414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2790843655842642e-05,
+      "loss": 4.5536,
+      "step": 3415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2743396875491386e-05,
+      "loss": 4.9096,
+      "step": 3416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.269599319707809e-05,
+      "loss": 4.2243,
+      "step": 3417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.264863264704954e-05,
+      "loss": 3.9566,
+      "step": 3418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.260131525182846e-05,
+      "loss": 3.7692,
+      "step": 3419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.255404103781359e-05,
+      "loss": 3.924,
+      "step": 3420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2506810031379444e-05,
+      "loss": 3.8736,
+      "step": 3421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2459622258876533e-05,
+      "loss": 4.5457,
+      "step": 3422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2412477746631177e-05,
+      "loss": 3.9841,
+      "step": 3423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.236537652094558e-05,
+      "loss": 4.3475,
+      "step": 3424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2318318608097867e-05,
+      "loss": 4.47,
+      "step": 3425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.22713040343419e-05,
+      "loss": 4.0984,
+      "step": 3426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.22243328259074e-05,
+      "loss": 4.1992,
+      "step": 3427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2177405008999873e-05,
+      "loss": 4.1903,
+      "step": 3428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.21305206098007e-05,
+      "loss": 4.3011,
+      "step": 3429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2083679654466938e-05,
+      "loss": 4.1843,
+      "step": 3430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2036882169131456e-05,
+      "loss": 4.1937,
+      "step": 3431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.199012817990287e-05,
+      "loss": 4.0333,
+      "step": 3432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.19434177128655e-05,
+      "loss": 4.1497,
+      "step": 3433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1896750794079412e-05,
+      "loss": 4.0324,
+      "step": 3434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1850127449580348e-05,
+      "loss": 3.9202,
+      "step": 3435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1803547705379813e-05,
+      "loss": 4.5051,
+      "step": 3436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1757011587464903e-05,
+      "loss": 4.0645,
+      "step": 3437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1710519121798412e-05,
+      "loss": 4.1574,
+      "step": 3438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1664070334318744e-05,
+      "loss": 4.3793,
+      "step": 3439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.161766525094002e-05,
+      "loss": 4.053,
+      "step": 3440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.157130389755191e-05,
+      "loss": 4.2839,
+      "step": 3441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.152498630001968e-05,
+      "loss": 4.4275,
+      "step": 3442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1478712484184217e-05,
+      "loss": 4.1963,
+      "step": 3443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1432482475861937e-05,
+      "loss": 4.4321,
+      "step": 3444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.138629630084491e-05,
+      "loss": 4.3397,
+      "step": 3445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1340153984900655e-05,
+      "loss": 4.1266,
+      "step": 3446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1294055553772262e-05,
+      "loss": 4.0246,
+      "step": 3447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.12480010331783e-05,
+      "loss": 4.0403,
+      "step": 3448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1201990448812923e-05,
+      "loss": 4.2994,
+      "step": 3449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1156023826345704e-05,
+      "loss": 4.225,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1110101191421707e-05,
+      "loss": 4.0074,
+      "step": 3451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1064222569661418e-05,
+      "loss": 4.0541,
+      "step": 3452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.101838798666086e-05,
+      "loss": 4.0418,
+      "step": 3453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0972597467991407e-05,
+      "loss": 4.2399,
+      "step": 3454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.092685103919987e-05,
+      "loss": 4.3543,
+      "step": 3455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0881148725808475e-05,
+      "loss": 3.7632,
+      "step": 3456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0835490553314786e-05,
+      "loss": 4.3307,
+      "step": 3457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0789876547191866e-05,
+      "loss": 4.1125,
+      "step": 3458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0744306732887953e-05,
+      "loss": 4.1917,
+      "step": 3459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0698781135826795e-05,
+      "loss": 3.7736,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0653299781407364e-05,
+      "loss": 4.0979,
+      "step": 3461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0607862695004064e-05,
+      "loss": 4.2145,
+      "step": 3462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.056246990196642e-05,
+      "loss": 4.0719,
+      "step": 3463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0517121427619435e-05,
+      "loss": 3.8843,
+      "step": 3464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047181729726325e-05,
+      "loss": 4.4117,
+      "step": 3465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.042655753617341e-05,
+      "loss": 3.8607,
+      "step": 3466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0381342169600515e-05,
+      "loss": 4.1986,
+      "step": 3467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0336171222770506e-05,
+      "loss": 4.3593,
+      "step": 3468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.02910447208846e-05,
+      "loss": 4.016,
+      "step": 3469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0245962689119113e-05,
+      "loss": 3.7182,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0200925152625593e-05,
+      "loss": 3.9595,
+      "step": 3471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0155932136530732e-05,
+      "loss": 4.2635,
+      "step": 3472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0110983665936478e-05,
+      "loss": 4.1608,
+      "step": 3473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0066079765919822e-05,
+      "loss": 3.7075,
+      "step": 3474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.002122046153293e-05,
+      "loss": 4.1085,
+      "step": 3475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9976405777803065e-05,
+      "loss": 3.7411,
+      "step": 3476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9931635739732667e-05,
+      "loss": 4.1537,
+      "step": 3477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9886910372299185e-05,
+      "loss": 4.1154,
+      "step": 3478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9842229700455194e-05,
+      "loss": 4.3856,
+      "step": 3479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9797593749128308e-05,
+      "loss": 4.5977,
+      "step": 3480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9753002543221166e-05,
+      "loss": 4.2404,
+      "step": 3481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970845610761155e-05,
+      "loss": 4.3576,
+      "step": 3482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9663954467152156e-05,
+      "loss": 4.2804,
+      "step": 3483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9619497646670715e-05,
+      "loss": 3.968,
+      "step": 3484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.957508567096996e-05,
+      "loss": 4.3448,
+      "step": 3485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9530718564827643e-05,
+      "loss": 4.4588,
+      "step": 3486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9486396352996426e-05,
+      "loss": 4.2106,
+      "step": 3487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9442119060203944e-05,
+      "loss": 3.9906,
+      "step": 3488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.939788671115278e-05,
+      "loss": 3.8621,
+      "step": 3489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9353699330520402e-05,
+      "loss": 3.8194,
+      "step": 3490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.930955694295932e-05,
+      "loss": 3.9118,
+      "step": 3491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.92654595730967e-05,
+      "loss": 3.9504,
+      "step": 3492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9221407245534863e-05,
+      "loss": 4.3371,
+      "step": 3493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9177399984850786e-05,
+      "loss": 4.2681,
+      "step": 3494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.91334378155965e-05,
+      "loss": 4.0881,
+      "step": 3495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908952076229865e-05,
+      "loss": 4.0475,
+      "step": 3496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9045648849458918e-05,
+      "loss": 4.4483,
+      "step": 3497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.900182210155367e-05,
+      "loss": 3.9798,
+      "step": 3498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8958040543034184e-05,
+      "loss": 4.259,
+      "step": 3499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.891430419832636e-05,
+      "loss": 4.0255,
+      "step": 3500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8870613091831068e-05,
+      "loss": 4.2752,
+      "step": 3501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.882696724792381e-05,
+      "loss": 4.098,
+      "step": 3502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8783366690954873e-05,
+      "loss": 3.8857,
+      "step": 3503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.873981144524928e-05,
+      "loss": 4.3656,
+      "step": 3504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8696301535106754e-05,
+      "loss": 4.4411,
+      "step": 3505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.865283698480178e-05,
+      "loss": 4.0542,
+      "step": 3506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8609417818583497e-05,
+      "loss": 4.1028,
+      "step": 3507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8566044060675703e-05,
+      "loss": 4.225,
+      "step": 3508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8522715735276873e-05,
+      "loss": 4.1746,
+      "step": 3509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8479432866560196e-05,
+      "loss": 4.1349,
+      "step": 3510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8436195478673434e-05,
+      "loss": 4.2353,
+      "step": 3511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.839300359573899e-05,
+      "loss": 4.3241,
+      "step": 3512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8349857241853874e-05,
+      "loss": 3.895,
+      "step": 3513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.830675644108969e-05,
+      "loss": 4.2016,
+      "step": 3514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8263701217492703e-05,
+      "loss": 4.0585,
+      "step": 3515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.822069159508364e-05,
+      "loss": 4.049,
+      "step": 3516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.817772759785784e-05,
+      "loss": 3.8211,
+      "step": 3517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8134809249785177e-05,
+      "loss": 4.3419,
+      "step": 3518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8091936574810098e-05,
+      "loss": 3.8744,
+      "step": 3519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.804910959685151e-05,
+      "loss": 4.0521,
+      "step": 3520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.800632833980286e-05,
+      "loss": 3.8138,
+      "step": 3521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.796359282753204e-05,
+      "loss": 3.7587,
+      "step": 3522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.792090308388151e-05,
+      "loss": 3.8653,
+      "step": 3523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.787825913266814e-05,
+      "loss": 4.4643,
+      "step": 3524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.783566099768319e-05,
+      "loss": 4.0427,
+      "step": 3525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7793108702692495e-05,
+      "loss": 4.1514,
+      "step": 3526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.775060227143619e-05,
+      "loss": 4.4678,
+      "step": 3527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.770814172762896e-05,
+      "loss": 3.8998,
+      "step": 3528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7665727094959695e-05,
+      "loss": 3.9853,
+      "step": 3529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.762335839709186e-05,
+      "loss": 4.053,
+      "step": 3530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.758103565766316e-05,
+      "loss": 3.6212,
+      "step": 3531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.75387589002858e-05,
+      "loss": 4.2326,
+      "step": 3532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.749652814854613e-05,
+      "loss": 4.7159,
+      "step": 3533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7454343426005047e-05,
+      "loss": 4.2176,
+      "step": 3534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.741220475619759e-05,
+      "loss": 4.2697,
+      "step": 3535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7370112162633288e-05,
+      "loss": 5.0415,
+      "step": 3536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7328065668795767e-05,
+      "loss": 3.973,
+      "step": 3537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.728606529814303e-05,
+      "loss": 3.9913,
+      "step": 3538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7244111074107383e-05,
+      "loss": 4.2134,
+      "step": 3539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7202203020095343e-05,
+      "loss": 4.4925,
+      "step": 3540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.716034115948767e-05,
+      "loss": 4.3039,
+      "step": 3541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7118525515639317e-05,
+      "loss": 3.7717,
+      "step": 3542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7076756111879556e-05,
+      "loss": 4.3567,
+      "step": 3543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7035032971511757e-05,
+      "loss": 3.9572,
+      "step": 3544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.699335611781353e-05,
+      "loss": 4.2267,
+      "step": 3545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6951725574036625e-05,
+      "loss": 4.3471,
+      "step": 3546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6910141363407016e-05,
+      "loss": 4.3268,
+      "step": 3547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.686860350912479e-05,
+      "loss": 3.916,
+      "step": 3548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6827112034364157e-05,
+      "loss": 3.6936,
+      "step": 3549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.678566696227347e-05,
+      "loss": 4.2258,
+      "step": 3550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6744268315975176e-05,
+      "loss": 3.8737,
+      "step": 3551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.670291611856587e-05,
+      "loss": 4.4255,
+      "step": 3552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.666161039311619e-05,
+      "loss": 4.0666,
+      "step": 3553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6620351162670855e-05,
+      "loss": 4.1314,
+      "step": 3554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.657913845024861e-05,
+      "loss": 4.1943,
+      "step": 3555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6537972278842338e-05,
+      "loss": 4.3278,
+      "step": 3556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6496852671418884e-05,
+      "loss": 4.0853,
+      "step": 3557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6455779650919113e-05,
+      "loss": 3.8878,
+      "step": 3558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.641475324025794e-05,
+      "loss": 3.8046,
+      "step": 3559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6373773462324215e-05,
+      "loss": 4.4819,
+      "step": 3560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.633284033998088e-05,
+      "loss": 3.8481,
+      "step": 3561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6291953896064704e-05,
+      "loss": 3.9723,
+      "step": 3562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6251114153386536e-05,
+      "loss": 4.0515,
+      "step": 3563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6210321134731077e-05,
+      "loss": 3.824,
+      "step": 3564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6169574862857096e-05,
+      "loss": 4.6103,
+      "step": 3565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6128875360497065e-05,
+      "loss": 4.0135,
+      "step": 3566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.608822265035759e-05,
+      "loss": 4.5197,
+      "step": 3567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6047616755119e-05,
+      "loss": 4.2764,
+      "step": 3568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6007057697435657e-05,
+      "loss": 4.0072,
+      "step": 3569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5966545499935613e-05,
+      "loss": 4.0714,
+      "step": 3570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.592608018522096e-05,
+      "loss": 4.2764,
+      "step": 3571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5885661775867498e-05,
+      "loss": 4.3533,
+      "step": 3572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.584529029442492e-05,
+      "loss": 3.9643,
+      "step": 3573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5804965763416734e-05,
+      "loss": 4.1234,
+      "step": 3574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5764688205340216e-05,
+      "loss": 3.9216,
+      "step": 3575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5724457642666514e-05,
+      "loss": 4.1029,
+      "step": 3576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5684274097840478e-05,
+      "loss": 4.2507,
+      "step": 3577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.564413759328076e-05,
+      "loss": 4.2336,
+      "step": 3578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5604048151379746e-05,
+      "loss": 4.198,
+      "step": 3579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5564005794503632e-05,
+      "loss": 5.0108,
+      "step": 3580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5524010544992274e-05,
+      "loss": 4.094,
+      "step": 3581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5484062425159252e-05,
+      "loss": 3.9447,
+      "step": 3582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5444161457291896e-05,
+      "loss": 4.0918,
+      "step": 3583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5404307663651165e-05,
+      "loss": 4.6907,
+      "step": 3584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5364501066471782e-05,
+      "loss": 3.9708,
+      "step": 3585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5324741687962075e-05,
+      "loss": 4.1627,
+      "step": 3586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5285029550304054e-05,
+      "loss": 4.1853,
+      "step": 3587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.524536467565334e-05,
+      "loss": 3.9226,
+      "step": 3588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5205747086139255e-05,
+      "loss": 3.9883,
+      "step": 3589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5166176803864674e-05,
+      "loss": 4.0598,
+      "step": 3590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5126653850906114e-05,
+      "loss": 4.0594,
+      "step": 3591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5087178249313638e-05,
+      "loss": 4.1225,
+      "step": 3592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5047750021111007e-05,
+      "loss": 4.6058,
+      "step": 3593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.50083691882954e-05,
+      "loss": 4.7207,
+      "step": 3594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4969035772837626e-05,
+      "loss": 4.1344,
+      "step": 3595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4929749796682101e-05,
+      "loss": 4.3381,
+      "step": 3596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4890511281746678e-05,
+      "loss": 3.8845,
+      "step": 3597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4851320249922773e-05,
+      "loss": 4.0981,
+      "step": 3598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4812176723075277e-05,
+      "loss": 4.2436,
+      "step": 3599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4773080723042665e-05,
+      "loss": 3.9031,
+      "step": 3600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4734032271636799e-05,
+      "loss": 4.0636,
+      "step": 3601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4695031390643066e-05,
+      "loss": 4.0079,
+      "step": 3602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.465607810182027e-05,
+      "loss": 3.967,
+      "step": 3603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4617172426900738e-05,
+      "loss": 4.4494,
+      "step": 3604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4578314387590165e-05,
+      "loss": 3.9734,
+      "step": 3605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4539504005567695e-05,
+      "loss": 4.0355,
+      "step": 3606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4500741302485876e-05,
+      "loss": 4.2024,
+      "step": 3607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4462026299970632e-05,
+      "loss": 4.136,
+      "step": 3608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4423359019621363e-05,
+      "loss": 4.0991,
+      "step": 3609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4384739483010756e-05,
+      "loss": 3.9809,
+      "step": 3610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4346167711684888e-05,
+      "loss": 4.1224,
+      "step": 3611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4307643727163166e-05,
+      "loss": 4.5013,
+      "step": 3612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4269167550938412e-05,
+      "loss": 4.4791,
+      "step": 3613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4230739204476695e-05,
+      "loss": 4.1678,
+      "step": 3614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4192358709217424e-05,
+      "loss": 4.4827,
+      "step": 3615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4154026086573303e-05,
+      "loss": 4.0766,
+      "step": 3616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4115741357930379e-05,
+      "loss": 4.3451,
+      "step": 3617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4077504544647934e-05,
+      "loss": 4.1172,
+      "step": 3618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.403931566805846e-05,
+      "loss": 3.9654,
+      "step": 3619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4001174749467838e-05,
+      "loss": 4.2368,
+      "step": 3620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3963081810155076e-05,
+      "loss": 4.3066,
+      "step": 3621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3925036871372532e-05,
+      "loss": 4.0329,
+      "step": 3622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3887039954345616e-05,
+      "loss": 4.7851,
+      "step": 3623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3849091080273124e-05,
+      "loss": 4.232,
+      "step": 3624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3811190270326913e-05,
+      "loss": 3.7853,
+      "step": 3625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3773337545652143e-05,
+      "loss": 4.0518,
+      "step": 3626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3735532927367e-05,
+      "loss": 3.9898,
+      "step": 3627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3697776436562992e-05,
+      "loss": 3.7098,
+      "step": 3628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3660068094304656e-05,
+      "loss": 4.6098,
+      "step": 3629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3622407921629732e-05,
+      "loss": 4.4233,
+      "step": 3630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3584795939549044e-05,
+      "loss": 4.0687,
+      "step": 3631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3547232169046542e-05,
+      "loss": 3.9427,
+      "step": 3632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3509716631079328e-05,
+      "loss": 4.1506,
+      "step": 3633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3472249346577525e-05,
+      "loss": 3.9523,
+      "step": 3634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3434830336444371e-05,
+      "loss": 4.153,
+      "step": 3635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339745962155613e-05,
+      "loss": 4.4308,
+      "step": 3636
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3360137222762215e-05,
+      "loss": 3.8599,
+      "step": 3637
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.332286316088499e-05,
+      "loss": 4.0204,
+      "step": 3638
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.32856374567199e-05,
+      "loss": 3.6062,
+      "step": 3639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3248460131035378e-05,
+      "loss": 4.3125,
+      "step": 3640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3211331204572863e-05,
+      "loss": 3.9436,
+      "step": 3641
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3174250698046876e-05,
+      "loss": 4.5805,
+      "step": 3642
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3137218632144832e-05,
+      "loss": 3.9321,
+      "step": 3643
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.310023502752713e-05,
+      "loss": 3.8943,
+      "step": 3644
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3063299904827164e-05,
+      "loss": 4.1872,
+      "step": 3645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3026413284651283e-05,
+      "loss": 3.8775,
+      "step": 3646
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2989575187578751e-05,
+      "loss": 3.5543,
+      "step": 3647
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.295278563416178e-05,
+      "loss": 4.0518,
+      "step": 3648
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2916044644925452e-05,
+      "loss": 4.0582,
+      "step": 3649
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2879352240367848e-05,
+      "loss": 3.9199,
+      "step": 3650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.284270844095985e-05,
+      "loss": 3.928,
+      "step": 3651
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.280611326714528e-05,
+      "loss": 5.0838,
+      "step": 3652
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.276956673934081e-05,
+      "loss": 4.7061,
+      "step": 3653
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2733068877935939e-05,
+      "loss": 4.0445,
+      "step": 3654
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2696619703293133e-05,
+      "loss": 4.1762,
+      "step": 3655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2660219235747506e-05,
+      "loss": 4.0464,
+      "step": 3656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.262386749560719e-05,
+      "loss": 3.8224,
+      "step": 3657
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2587564503152993e-05,
+      "loss": 3.8477,
+      "step": 3658
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2551310278638639e-05,
+      "loss": 3.7745,
+      "step": 3659
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2515104842290504e-05,
+      "loss": 3.8793,
+      "step": 3660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2478948214307884e-05,
+      "loss": 4.1437,
+      "step": 3661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2442840414862767e-05,
+      "loss": 4.0461,
+      "step": 3662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2406781464099947e-05,
+      "loss": 4.0623,
+      "step": 3663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2370771382136902e-05,
+      "loss": 4.1434,
+      "step": 3664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.233481018906386e-05,
+      "loss": 3.7517,
+      "step": 3665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.229889790494384e-05,
+      "loss": 3.9331,
+      "step": 3666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2263034549812524e-05,
+      "loss": 4.1637,
+      "step": 3667
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2227220143678276e-05,
+      "loss": 4.0566,
+      "step": 3668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2191454706522164e-05,
+      "loss": 3.7948,
+      "step": 3669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2155738258298e-05,
+      "loss": 3.9771,
+      "step": 3670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2120070818932172e-05,
+      "loss": 3.9822,
+      "step": 3671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2084452408323777e-05,
+      "loss": 3.8649,
+      "step": 3672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2048883046344528e-05,
+      "loss": 4.2545,
+      "step": 3673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.201336275283883e-05,
+      "loss": 4.2398,
+      "step": 3674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1977891547623654e-05,
+      "loss": 4.0941,
+      "step": 3675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1942469450488603e-05,
+      "loss": 4.1041,
+      "step": 3676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.190709648119589e-05,
+      "loss": 4.0507,
+      "step": 3677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1871772659480285e-05,
+      "loss": 3.9727,
+      "step": 3678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1836498005049223e-05,
+      "loss": 3.7787,
+      "step": 3679
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.180127253758262e-05,
+      "loss": 4.2642,
+      "step": 3680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1766096276732985e-05,
+      "loss": 3.897,
+      "step": 3681
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1730969242125344e-05,
+      "loss": 3.7823,
+      "step": 3682
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1695891453357344e-05,
+      "loss": 4.2777,
+      "step": 3683
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1660862929999072e-05,
+      "loss": 4.0956,
+      "step": 3684
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1625883691593164e-05,
+      "loss": 3.8586,
+      "step": 3685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1590953757654731e-05,
+      "loss": 4.4296,
+      "step": 3686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1556073147671431e-05,
+      "loss": 4.1494,
+      "step": 3687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1521241881103383e-05,
+      "loss": 4.1337,
+      "step": 3688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1486459977383102e-05,
+      "loss": 3.8139,
+      "step": 3689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1451727455915695e-05,
+      "loss": 3.6964,
+      "step": 3690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1417044336078597e-05,
+      "loss": 4.0166,
+      "step": 3691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1382410637221807e-05,
+      "loss": 3.4975,
+      "step": 3692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1347826378667603e-05,
+      "loss": 3.852,
+      "step": 3693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1313291579710805e-05,
+      "loss": 4.1214,
+      "step": 3694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1278806259618557e-05,
+      "loss": 3.7634,
+      "step": 3695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1244370437630502e-05,
+      "loss": 4.1417,
+      "step": 3696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1209984132958507e-05,
+      "loss": 4.5175,
+      "step": 3697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1175647364786967e-05,
+      "loss": 4.8019,
+      "step": 3698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1141360152272562e-05,
+      "loss": 3.7517,
+      "step": 3699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1107122514544332e-05,
+      "loss": 3.9473,
+      "step": 3700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1072934470703678e-05,
+      "loss": 3.8519,
+      "step": 3701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1038796039824295e-05,
+      "loss": 4.307,
+      "step": 3702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1004707240952261e-05,
+      "loss": 3.9298,
+      "step": 3703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0970668093105918e-05,
+      "loss": 4.1036,
+      "step": 3704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.093667861527592e-05,
+      "loss": 4.317,
+      "step": 3705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0902738826425175e-05,
+      "loss": 4.2584,
+      "step": 3706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0868848745488947e-05,
+      "loss": 4.0348,
+      "step": 3707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0835008391374713e-05,
+      "loss": 3.9027,
+      "step": 3708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0801217782962204e-05,
+      "loss": 4.1443,
+      "step": 3709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0767476939103405e-05,
+      "loss": 4.63,
+      "step": 3710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0733785878622537e-05,
+      "loss": 4.2321,
+      "step": 3711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0700144620316078e-05,
+      "loss": 3.8659,
+      "step": 3712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0666553182952676e-05,
+      "loss": 3.9344,
+      "step": 3713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0633011585273222e-05,
+      "loss": 3.9574,
+      "step": 3714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0599519845990735e-05,
+      "loss": 3.8971,
+      "step": 3715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0566077983790524e-05,
+      "loss": 4.1254,
+      "step": 3716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0532686017329974e-05,
+      "loss": 3.9621,
+      "step": 3717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0499343965238684e-05,
+      "loss": 3.9606,
+      "step": 3718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0466051846118364e-05,
+      "loss": 3.8696,
+      "step": 3719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0432809678542943e-05,
+      "loss": 3.7918,
+      "step": 3720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.03996174810584e-05,
+      "loss": 3.9741,
+      "step": 3721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0366475272182873e-05,
+      "loss": 4.2224,
+      "step": 3722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.033338307040661e-05,
+      "loss": 3.6382,
+      "step": 3723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0300340894191928e-05,
+      "loss": 4.4599,
+      "step": 3724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0267348761973328e-05,
+      "loss": 3.925,
+      "step": 3725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.023440669215724e-05,
+      "loss": 4.0391,
+      "step": 3726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0201514703122318e-05,
+      "loss": 3.8427,
+      "step": 3727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0168672813219172e-05,
+      "loss": 3.9117,
+      "step": 3728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0135881040770512e-05,
+      "loss": 3.9556,
+      "step": 3729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0103139404071038e-05,
+      "loss": 4.8918,
+      "step": 3730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.007044792138756e-05,
+      "loss": 3.8606,
+      "step": 3731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0037806610958845e-05,
+      "loss": 4.2982,
+      "step": 3732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0005215490995668e-05,
+      "loss": 3.887,
+      "step": 3733
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.972674579680818e-06,
+      "loss": 3.9689,
+      "step": 3734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.940183895169064e-06,
+      "loss": 4.0139,
+      "step": 3735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.907743455587194e-06,
+      "loss": 3.8065,
+      "step": 3736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.875353279033905e-06,
+      "loss": 4.1427,
+      "step": 3737
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.84301338357988e-06,
+      "loss": 4.2442,
+      "step": 3738
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.810723787267728e-06,
+      "loss": 4.1702,
+      "step": 3739
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.778484508112062e-06,
+      "loss": 3.9922,
+      "step": 3740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.74629556409935e-06,
+      "loss": 4.2201,
+      "step": 3741
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.714156973188004e-06,
+      "loss": 4.2662,
+      "step": 3742
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.682068753308315e-06,
+      "loss": 4.3199,
+      "step": 3743
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.650030922362551e-06,
+      "loss": 4.0263,
+      "step": 3744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.618043498224794e-06,
+      "loss": 4.0912,
+      "step": 3745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.586106498741022e-06,
+      "loss": 3.9457,
+      "step": 3746
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.554219941729092e-06,
+      "loss": 4.0125,
+      "step": 3747
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.522383844978688e-06,
+      "loss": 4.1053,
+      "step": 3748
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.490598226251402e-06,
+      "loss": 3.967,
+      "step": 3749
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.458863103280613e-06,
+      "loss": 4.3928,
+      "step": 3750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.427178493771538e-06,
+      "loss": 3.534,
+      "step": 3751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.395544415401202e-06,
+      "loss": 3.958,
+      "step": 3752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.363960885818512e-06,
+      "loss": 3.9548,
+      "step": 3753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.332427922644037e-06,
+      "loss": 4.2157,
+      "step": 3754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.300945543470273e-06,
+      "loss": 4.1713,
+      "step": 3755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.269513765861392e-06,
+      "loss": 4.1311,
+      "step": 3756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.238132607353433e-06,
+      "loss": 4.0978,
+      "step": 3757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.206802085454081e-06,
+      "loss": 3.9744,
+      "step": 3758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.17552221764284e-06,
+      "loss": 4.1214,
+      "step": 3759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.144293021370964e-06,
+      "loss": 3.8173,
+      "step": 3760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.113114514061417e-06,
+      "loss": 4.3348,
+      "step": 3761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.081986713108858e-06,
+      "loss": 3.8036,
+      "step": 3762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.050909635879678e-06,
+      "loss": 3.9374,
+      "step": 3763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.019883299712007e-06,
+      "loss": 3.9906,
+      "step": 3764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.988907721915595e-06,
+      "loss": 3.9166,
+      "step": 3765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.957982919771934e-06,
+      "loss": 4.1542,
+      "step": 3766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.927108910534131e-06,
+      "loss": 4.2286,
+      "step": 3767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.896285711427022e-06,
+      "loss": 4.1873,
+      "step": 3768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.865513339647046e-06,
+      "loss": 4.2057,
+      "step": 3769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.834791812362297e-06,
+      "loss": 3.9244,
+      "step": 3770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.804121146712519e-06,
+      "loss": 3.9683,
+      "step": 3771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.773501359809033e-06,
+      "loss": 4.0322,
+      "step": 3772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.742932468734843e-06,
+      "loss": 4.5298,
+      "step": 3773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.712414490544518e-06,
+      "loss": 3.7183,
+      "step": 3774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.681947442264216e-06,
+      "loss": 4.2935,
+      "step": 3775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.651531340891672e-06,
+      "loss": 4.1138,
+      "step": 3776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.62116620339627e-06,
+      "loss": 4.3024,
+      "step": 3777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.590852046718878e-06,
+      "loss": 4.7633,
+      "step": 3778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.560588887771947e-06,
+      "loss": 4.2219,
+      "step": 3779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.530376743439505e-06,
+      "loss": 3.9936,
+      "step": 3780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.500215630577047e-06,
+      "loss": 4.555,
+      "step": 3781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.47010556601171e-06,
+      "loss": 4.7149,
+      "step": 3782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.440046566542025e-06,
+      "loss": 3.835,
+      "step": 3783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.410038648938134e-06,
+      "loss": 3.8149,
+      "step": 3784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.38008182994161e-06,
+      "loss": 4.0242,
+      "step": 3785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.350176126265597e-06,
+      "loss": 3.9555,
+      "step": 3786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.320321554594612e-06,
+      "loss": 3.8615,
+      "step": 3787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.290518131584746e-06,
+      "loss": 4.0921,
+      "step": 3788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.260765873863497e-06,
+      "loss": 4.0364,
+      "step": 3789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.231064798029863e-06,
+      "loss": 4.0802,
+      "step": 3790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.201414920654216e-06,
+      "loss": 4.0391,
+      "step": 3791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.171816258278408e-06,
+      "loss": 4.0153,
+      "step": 3792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.142268827415755e-06,
+      "loss": 3.8569,
+      "step": 3793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.112772644550925e-06,
+      "loss": 3.9914,
+      "step": 3794
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.083327726140033e-06,
+      "loss": 4.3733,
+      "step": 3795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.053934088610548e-06,
+      "loss": 4.6075,
+      "step": 3796
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.02459174836142e-06,
+      "loss": 4.1317,
+      "step": 3797
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.995300721762878e-06,
+      "loss": 4.421,
+      "step": 3798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.96606102515659e-06,
+      "loss": 3.5419,
+      "step": 3799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.936872674855534e-06,
+      "loss": 3.8754,
+      "step": 3800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.907735687144113e-06,
+      "loss": 3.9594,
+      "step": 3801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.878650078278016e-06,
+      "loss": 3.7431,
+      "step": 3802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.849615864484284e-06,
+      "loss": 3.8824,
+      "step": 3803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.82063306196128e-06,
+      "loss": 4.2954,
+      "step": 3804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.791701686878671e-06,
+      "loss": 4.3507,
+      "step": 3805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.762821755377491e-06,
+      "loss": 3.7759,
+      "step": 3806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.733993283570007e-06,
+      "loss": 4.1389,
+      "step": 3807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.705216287539819e-06,
+      "loss": 3.8293,
+      "step": 3808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.676490783341761e-06,
+      "loss": 3.8261,
+      "step": 3809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.647816787002005e-06,
+      "loss": 3.7794,
+      "step": 3810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.619194314517941e-06,
+      "loss": 4.8933,
+      "step": 3811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5906233818582085e-06,
+      "loss": 4.4938,
+      "step": 3812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.562104004962711e-06,
+      "loss": 3.7922,
+      "step": 3813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.533636199742611e-06,
+      "loss": 4.1373,
+      "step": 3814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5052199820802695e-06,
+      "loss": 3.9823,
+      "step": 3815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.476855367829227e-06,
+      "loss": 3.8971,
+      "step": 3816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.448542372814315e-06,
+      "loss": 4.3199,
+      "step": 3817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4202810128315205e-06,
+      "loss": 4.4392,
+      "step": 3818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.392071303648052e-06,
+      "loss": 4.3383,
+      "step": 3819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.363913261002242e-06,
+      "loss": 4.2991,
+      "step": 3820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3358069006036585e-06,
+      "loss": 3.8241,
+      "step": 3821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.307752238133014e-06,
+      "loss": 3.9909,
+      "step": 3822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.279749289242199e-06,
+      "loss": 3.8289,
+      "step": 3823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2517980695541834e-06,
+      "loss": 4.4542,
+      "step": 3824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.223898594663158e-06,
+      "loss": 3.5312,
+      "step": 3825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.196050880134419e-06,
+      "loss": 3.7521,
+      "step": 3826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.168254941504371e-06,
+      "loss": 4.0309,
+      "step": 3827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.14051079428053e-06,
+      "loss": 3.6656,
+      "step": 3828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.112818453941517e-06,
+      "loss": 3.5439,
+      "step": 3829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.085177935937104e-06,
+      "loss": 3.9374,
+      "step": 3830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.057589255688079e-06,
+      "loss": 3.9736,
+      "step": 3831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.030052428586342e-06,
+      "loss": 4.2858,
+      "step": 3832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.002567469994836e-06,
+      "loss": 3.8919,
+      "step": 3833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.97513439524764e-06,
+      "loss": 3.8402,
+      "step": 3834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.947753219649799e-06,
+      "loss": 3.8126,
+      "step": 3835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.920423958477462e-06,
+      "loss": 3.5348,
+      "step": 3836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8931466269777535e-06,
+      "loss": 3.8768,
+      "step": 3837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.865921240368922e-06,
+      "loss": 4.012,
+      "step": 3838
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.838747813840141e-06,
+      "loss": 4.1859,
+      "step": 3839
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.811626362551649e-06,
+      "loss": 3.8295,
+      "step": 3840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7845569016346755e-06,
+      "loss": 4.3225,
+      "step": 3841
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.757539446191408e-06,
+      "loss": 3.9474,
+      "step": 3842
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.730574011295088e-06,
+      "loss": 4.2038,
+      "step": 3843
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.703660611989904e-06,
+      "loss": 4.0968,
+      "step": 3844
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.676799263290978e-06,
+      "loss": 4.0254,
+      "step": 3845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.649989980184423e-06,
+      "loss": 4.1421,
+      "step": 3846
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6232327776273416e-06,
+      "loss": 3.8136,
+      "step": 3847
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5965276705477144e-06,
+      "loss": 4.2666,
+      "step": 3848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.56987467384449e-06,
+      "loss": 3.8154,
+      "step": 3849
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.543273802387551e-06,
+      "loss": 4.0332,
+      "step": 3850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5167250710176486e-06,
+      "loss": 3.8223,
+      "step": 3851
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.490228494546558e-06,
+      "loss": 4.0342,
+      "step": 3852
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.463784087756808e-06,
+      "loss": 4.3333,
+      "step": 3853
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4373918654019536e-06,
+      "loss": 4.4655,
+      "step": 3854
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.411051842206328e-06,
+      "loss": 3.8148,
+      "step": 3855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.384764032865265e-06,
+      "loss": 3.6607,
+      "step": 3856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.358528452044809e-06,
+      "loss": 3.6444,
+      "step": 3857
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3323451143820215e-06,
+      "loss": 4.2741,
+      "step": 3858
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.306214034484714e-06,
+      "loss": 3.8676,
+      "step": 3859
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.280135226931616e-06,
+      "loss": 4.0039,
+      "step": 3860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.25410870627221e-06,
+      "loss": 4.4519,
+      "step": 3861
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.228134487026848e-06,
+      "loss": 4.1841,
+      "step": 3862
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.202212583686762e-06,
+      "loss": 3.9863,
+      "step": 3863
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.176343010713903e-06,
+      "loss": 3.9214,
+      "step": 3864
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.150525782541083e-06,
+      "loss": 3.9961,
+      "step": 3865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.124760913571869e-06,
+      "loss": 4.2067,
+      "step": 3866
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.099048418180675e-06,
+      "loss": 4.0062,
+      "step": 3867
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.073388310712646e-06,
+      "loss": 3.9881,
+      "step": 3868
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.047780605483711e-06,
+      "loss": 3.9593,
+      "step": 3869
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.022225316780561e-06,
+      "loss": 3.9581,
+      "step": 3870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.996722458860671e-06,
+      "loss": 3.7369,
+      "step": 3871
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.971272045952225e-06,
+      "loss": 3.7168,
+      "step": 3872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.945874092254189e-06,
+      "loss": 3.8878,
+      "step": 3873
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.920528611936215e-06,
+      "loss": 4.1598,
+      "step": 3874
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8952356191386945e-06,
+      "loss": 4.2357,
+      "step": 3875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.869995127972783e-06,
+      "loss": 3.7097,
+      "step": 3876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8448071525202816e-06,
+      "loss": 3.9653,
+      "step": 3877
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8196717068337295e-06,
+      "loss": 4.1778,
+      "step": 3878
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.79458880493633e-06,
+      "loss": 3.8229,
+      "step": 3879
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.769558460822e-06,
+      "loss": 3.6469,
+      "step": 3880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.744580688455336e-06,
+      "loss": 4.1248,
+      "step": 3881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.719655501771582e-06,
+      "loss": 4.7493,
+      "step": 3882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6947829146766325e-06,
+      "loss": 3.9675,
+      "step": 3883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.669962941047102e-06,
+      "loss": 4.5522,
+      "step": 3884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.64519559473019e-06,
+      "loss": 3.789,
+      "step": 3885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6204808895437226e-06,
+      "loss": 4.2511,
+      "step": 3886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5958188392762235e-06,
+      "loss": 3.8715,
+      "step": 3887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.571209457686788e-06,
+      "loss": 4.1061,
+      "step": 3888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.546652758505144e-06,
+      "loss": 4.0937,
+      "step": 3889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.522148755431622e-06,
+      "loss": 3.9562,
+      "step": 3890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.497697462137175e-06,
+      "loss": 3.9697,
+      "step": 3891
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4732988922633235e-06,
+      "loss": 3.7105,
+      "step": 3892
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.448953059422168e-06,
+      "loss": 3.8898,
+      "step": 3893
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.424659977196401e-06,
+      "loss": 4.2609,
+      "step": 3894
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.400419659139311e-06,
+      "loss": 4.2622,
+      "step": 3895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.376232118774693e-06,
+      "loss": 3.9533,
+      "step": 3896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.352097369596931e-06,
+      "loss": 3.7622,
+      "step": 3897
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3280154250709644e-06,
+      "loss": 4.292,
+      "step": 3898
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.303986298632235e-06,
+      "loss": 3.9174,
+      "step": 3899
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.280010003686775e-06,
+      "loss": 3.7362,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.256086553611073e-06,
+      "loss": 4.0716,
+      "step": 3901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.232215961752207e-06,
+      "loss": 4.1416,
+      "step": 3902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.208398241427681e-06,
+      "loss": 3.8091,
+      "step": 3903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.184633405925587e-06,
+      "loss": 4.1374,
+      "step": 3904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.160921468504476e-06,
+      "loss": 3.9788,
+      "step": 3905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1372624423933645e-06,
+      "loss": 3.7954,
+      "step": 3906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.113656340791762e-06,
+      "loss": 3.9342,
+      "step": 3907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.090103176869687e-06,
+      "loss": 4.4716,
+      "step": 3908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.066602963767597e-06,
+      "loss": 3.9597,
+      "step": 3909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.04315571459637e-06,
+      "loss": 3.962,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.01976144243741e-06,
+      "loss": 3.9465,
+      "step": 3911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9964201603424855e-06,
+      "loss": 4.0447,
+      "step": 3912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9731318813339036e-06,
+      "loss": 3.7461,
+      "step": 3913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.94989661840427e-06,
+      "loss": 3.7728,
+      "step": 3914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9267143845167395e-06,
+      "loss": 3.9706,
+      "step": 3915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.903585192604776e-06,
+      "loss": 4.2896,
+      "step": 3916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8805090555723576e-06,
+      "loss": 4.3397,
+      "step": 3917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.857485986293742e-06,
+      "loss": 3.9611,
+      "step": 3918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.834515997613687e-06,
+      "loss": 4.4791,
+      "step": 3919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.811599102347264e-06,
+      "loss": 4.3891,
+      "step": 3920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.788735313279968e-06,
+      "loss": 3.8957,
+      "step": 3921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.765924643167641e-06,
+      "loss": 3.7338,
+      "step": 3922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.74316710473649e-06,
+      "loss": 3.4863,
+      "step": 3923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7204627106831045e-06,
+      "loss": 3.9493,
+      "step": 3924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.697811473674407e-06,
+      "loss": 3.9518,
+      "step": 3925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.675213406347645e-06,
+      "loss": 4.5595,
+      "step": 3926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6526685213104215e-06,
+      "loss": 3.7729,
+      "step": 3927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.630176831140687e-06,
+      "loss": 3.8562,
+      "step": 3928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.607738348386681e-06,
+      "loss": 4.1232,
+      "step": 3929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5853530855669794e-06,
+      "loss": 3.7229,
+      "step": 3930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.563021055170447e-06,
+      "loss": 3.7778,
+      "step": 3931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.540742269656262e-06,
+      "loss": 4.0409,
+      "step": 3932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.518516741453926e-06,
+      "loss": 3.9568,
+      "step": 3933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.496344482963166e-06,
+      "loss": 4.0841,
+      "step": 3934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.474225506554031e-06,
+      "loss": 4.0524,
+      "step": 3935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.452159824566837e-06,
+      "loss": 4.1284,
+      "step": 3936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.430147449312172e-06,
+      "loss": 3.8455,
+      "step": 3937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.408188393070867e-06,
+      "loss": 3.9754,
+      "step": 3938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.386282668094033e-06,
+      "loss": 4.3727,
+      "step": 3939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3644302866029715e-06,
+      "loss": 4.0925,
+      "step": 3940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.342631260789298e-06,
+      "loss": 4.4062,
+      "step": 3941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.320885602814828e-06,
+      "loss": 3.5674,
+      "step": 3942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.299193324811568e-06,
+      "loss": 4.3892,
+      "step": 3943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.277554438881792e-06,
+      "loss": 4.1251,
+      "step": 3944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2559689570979555e-06,
+      "loss": 3.7517,
+      "step": 3945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.234436891502769e-06,
+      "loss": 3.3487,
+      "step": 3946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.212958254109045e-06,
+      "loss": 4.0895,
+      "step": 3947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.19153305689991e-06,
+      "loss": 4.2385,
+      "step": 3948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.170161311828569e-06,
+      "loss": 3.7719,
+      "step": 3949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.148843030818494e-06,
+      "loss": 3.9675,
+      "step": 3950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.12757822576324e-06,
+      "loss": 3.7679,
+      "step": 3951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.106366908526616e-06,
+      "loss": 4.2068,
+      "step": 3952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.085209090942499e-06,
+      "loss": 3.9712,
+      "step": 3953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.064104784815026e-06,
+      "loss": 3.9196,
+      "step": 3954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.043054001918378e-06,
+      "loss": 4.0829,
+      "step": 3955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.022056753996916e-06,
+      "loss": 3.6409,
+      "step": 3956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.00111305276516e-06,
+      "loss": 3.8713,
+      "step": 3957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.980222909907716e-06,
+      "loss": 4.2007,
+      "step": 3958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.959386337079329e-06,
+      "loss": 3.7264,
+      "step": 3959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.938603345904834e-06,
+      "loss": 3.9036,
+      "step": 3960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9178739479792224e-06,
+      "loss": 4.3776,
+      "step": 3961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.897198154867532e-06,
+      "loss": 4.3707,
+      "step": 3962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.876575978104924e-06,
+      "loss": 4.1838,
+      "step": 3963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.856007429196617e-06,
+      "loss": 4.07,
+      "step": 3964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.835492519617978e-06,
+      "loss": 3.6882,
+      "step": 3965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.815031260814361e-06,
+      "loss": 3.69,
+      "step": 3966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.794623664201247e-06,
+      "loss": 4.1108,
+      "step": 3967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7742697411641603e-06,
+      "loss": 4.2148,
+      "step": 3968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.753969503058663e-06,
+      "loss": 3.8611,
+      "step": 3969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7337229612104175e-06,
+      "loss": 3.7197,
+      "step": 3970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7135301269150767e-06,
+      "loss": 3.9216,
+      "step": 3971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6933910114383406e-06,
+      "loss": 3.9277,
+      "step": 3972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6733056260159436e-06,
+      "loss": 3.7286,
+      "step": 3973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6532739818536667e-06,
+      "loss": 4.7223,
+      "step": 3974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6332960901272807e-06,
+      "loss": 3.9893,
+      "step": 3975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.61337196198257e-06,
+      "loss": 4.2033,
+      "step": 3976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5935016085353323e-06,
+      "loss": 4.2689,
+      "step": 3977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.573685040871344e-06,
+      "loss": 3.8569,
+      "step": 3978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.553922270046439e-06,
+      "loss": 4.0594,
+      "step": 3979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5342133070863424e-06,
+      "loss": 3.9191,
+      "step": 3980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.514558162986825e-06,
+      "loss": 4.0812,
+      "step": 3981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4949568487135932e-06,
+      "loss": 4.2104,
+      "step": 3982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4754093752023985e-06,
+      "loss": 3.9865,
+      "step": 3983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4559157533588295e-06,
+      "loss": 4.0826,
+      "step": 3984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.436475994058541e-06,
+      "loss": 3.5219,
+      "step": 3985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4170901081470695e-06,
+      "loss": 3.877,
+      "step": 3986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.397758106439952e-06,
+      "loss": 3.9775,
+      "step": 3987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.378479999722595e-06,
+      "loss": 4.2844,
+      "step": 3988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.359255798750405e-06,
+      "loss": 4.3097,
+      "step": 3989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.340085514248681e-06,
+      "loss": 3.7281,
+      "step": 3990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3209691569126233e-06,
+      "loss": 4.1165,
+      "step": 3991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3019067374073787e-06,
+      "loss": 3.9573,
+      "step": 3992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.282898266367973e-06,
+      "loss": 4.176,
+      "step": 3993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.263943754399379e-06,
+      "loss": 4.1166,
+      "step": 3994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.245043212076415e-06,
+      "loss": 4.0791,
+      "step": 3995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2261966499438244e-06,
+      "loss": 3.9784,
+      "step": 3996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2074040785162075e-06,
+      "loss": 3.9056,
+      "step": 3997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.188665508278066e-06,
+      "loss": 3.6564,
+      "step": 3998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1699809496837597e-06,
+      "loss": 4.0589,
+      "step": 3999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.151350413157528e-06,
+      "loss": 4.4795,
+      "step": 4000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.132773909093456e-06,
+      "loss": 3.9885,
+      "step": 4001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1142514478554763e-06,
+      "loss": 4.0569,
+      "step": 4002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.095783039777411e-06,
+      "loss": 3.873,
+      "step": 4003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0773686951628857e-06,
+      "loss": 3.5247,
+      "step": 4004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.059008424285381e-06,
+      "loss": 4.18,
+      "step": 4005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0407022373881934e-06,
+      "loss": 4.3153,
+      "step": 4006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0224501446844855e-06,
+      "loss": 4.0896,
+      "step": 4007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0042521563571903e-06,
+      "loss": 3.9349,
+      "step": 4008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9861082825590856e-06,
+      "loss": 4.0461,
+      "step": 4009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9680185334127418e-06,
+      "loss": 4.2607,
+      "step": 4010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.949982919010563e-06,
+      "loss": 3.9291,
+      "step": 4011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.932001449414723e-06,
+      "loss": 3.9347,
+      "step": 4012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.914074134657163e-06,
+      "loss": 3.9174,
+      "step": 4013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.896200984739694e-06,
+      "loss": 3.7986,
+      "step": 4014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8783820096338175e-06,
+      "loss": 3.9721,
+      "step": 4015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8606172192808813e-06,
+      "loss": 4.0356,
+      "step": 4016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8429066235919455e-06,
+      "loss": 3.9208,
+      "step": 4017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8252502324478735e-06,
+      "loss": 4.5281,
+      "step": 4018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8076480556992748e-06,
+      "loss": 3.5537,
+      "step": 4019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7901001031665265e-06,
+      "loss": 4.1475,
+      "step": 4020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7726063846396977e-06,
+      "loss": 3.748,
+      "step": 4021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7551669098786925e-06,
+      "loss": 3.8371,
+      "step": 4022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7377816886130726e-06,
+      "loss": 3.9883,
+      "step": 4023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.720450730542179e-06,
+      "loss": 4.1816,
+      "step": 4024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.703174045335044e-06,
+      "loss": 3.9908,
+      "step": 4025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6859516426304355e-06,
+      "loss": 3.8271,
+      "step": 4026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.668783532036856e-06,
+      "loss": 4.1914,
+      "step": 4027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.651669723132488e-06,
+      "loss": 3.8739,
+      "step": 4028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6346102254652395e-06,
+      "loss": 3.9135,
+      "step": 4029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.617605048552696e-06,
+      "loss": 3.9843,
+      "step": 4030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.600654201882169e-06,
+      "loss": 4.0013,
+      "step": 4031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5837576949106267e-06,
+      "loss": 3.7721,
+      "step": 4032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5669155370647513e-06,
+      "loss": 3.8763,
+      "step": 4033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5501277377408596e-06,
+      "loss": 3.9191,
+      "step": 4034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5333943063050035e-06,
+      "loss": 3.9225,
+      "step": 4035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5167152520928495e-06,
+      "loss": 4.2509,
+      "step": 4036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.500090584409753e-06,
+      "loss": 4.2045,
+      "step": 4037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4835203125307072e-06,
+      "loss": 4.2037,
+      "step": 4038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4670044457003605e-06,
+      "loss": 4.1366,
+      "step": 4039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4505429931330427e-06,
+      "loss": 3.9279,
+      "step": 4040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.434135964012696e-06,
+      "loss": 3.9803,
+      "step": 4041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4177833674928985e-06,
+      "loss": 3.9463,
+      "step": 4042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4014852126968522e-06,
+      "loss": 3.9559,
+      "step": 4043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3852415087174283e-06,
+      "loss": 3.7818,
+      "step": 4044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3690522646170554e-06,
+      "loss": 4.7043,
+      "step": 4045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3529174894278416e-06,
+      "loss": 4.0227,
+      "step": 4046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3368371921514754e-06,
+      "loss": 4.1132,
+      "step": 4047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.320811381759236e-06,
+      "loss": 4.2428,
+      "step": 4048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3048400671920267e-06,
+      "loss": 4.2316,
+      "step": 4049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2889232573603424e-06,
+      "loss": 4.1143,
+      "step": 4050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.273060961144291e-06,
+      "loss": 4.0831,
+      "step": 4051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2572531873935265e-06,
+      "loss": 3.7339,
+      "step": 4052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2414999449273167e-06,
+      "loss": 3.9708,
+      "step": 4053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2258012425344642e-06,
+      "loss": 3.8421,
+      "step": 4054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.210157088973408e-06,
+      "loss": 4.2401,
+      "step": 4055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1945674929720994e-06,
+      "loss": 4.4164,
+      "step": 4056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1790324632280813e-06,
+      "loss": 4.6064,
+      "step": 4057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1635520084084205e-06,
+      "loss": 5.0342,
+      "step": 4058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1481261371497974e-06,
+      "loss": 4.1237,
+      "step": 4059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.132754858058361e-06,
+      "loss": 3.8449,
+      "step": 4060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.117438179709874e-06,
+      "loss": 4.0994,
+      "step": 4061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1021761106495895e-06,
+      "loss": 4.0323,
+      "step": 4062
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0869686593923076e-06,
+      "loss": 4.0872,
+      "step": 4063
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0718158344223635e-06,
+      "loss": 3.8282,
+      "step": 4064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0567176441936285e-06,
+      "loss": 4.0243,
+      "step": 4065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0416740971294536e-06,
+      "loss": 4.1328,
+      "step": 4066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0266852016227357e-06,
+      "loss": 3.6935,
+      "step": 4067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0117509660358746e-06,
+      "loss": 3.9896,
+      "step": 4068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9968713987007725e-06,
+      "loss": 4.0245,
+      "step": 4069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9820465079188335e-06,
+      "loss": 3.8861,
+      "step": 4070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9672763019609632e-06,
+      "loss": 4.1721,
+      "step": 4071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9525607890675146e-06,
+      "loss": 3.5714,
+      "step": 4072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.937899977448421e-06,
+      "loss": 3.788,
+      "step": 4073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9232938752829943e-06,
+      "loss": 4.1199,
+      "step": 4074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9087424907200837e-06,
+      "loss": 4.0059,
+      "step": 4075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8942458318779943e-06,
+      "loss": 4.1363,
+      "step": 4076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.879803906844524e-06,
+      "loss": 4.1504,
+      "step": 4077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8654167236768826e-06,
+      "loss": 3.7841,
+      "step": 4078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8510842904017832e-06,
+      "loss": 4.0218,
+      "step": 4079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.836806615015363e-06,
+      "loss": 4.2457,
+      "step": 4080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8225837054832607e-06,
+      "loss": 3.88,
+      "step": 4081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8084155697404959e-06,
+      "loss": 3.909,
+      "step": 4082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7943022156915567e-06,
+      "loss": 3.8829,
+      "step": 4083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7802436512104004e-06,
+      "loss": 3.8925,
+      "step": 4084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7662398841403527e-06,
+      "loss": 3.9383,
+      "step": 4085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7522909222942307e-06,
+      "loss": 4.1709,
+      "step": 4086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7383967734542206e-06,
+      "loss": 4.2182,
+      "step": 4087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7245574453719771e-06,
+      "loss": 4.0824,
+      "step": 4088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.710772945768535e-06,
+      "loss": 3.809,
+      "step": 4089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.697043282334354e-06,
+      "loss": 4.28,
+      "step": 4090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6833684627292956e-06,
+      "loss": 5.2331,
+      "step": 4091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6697484945826458e-06,
+      "loss": 3.9236,
+      "step": 4092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6561833854930374e-06,
+      "loss": 3.8193,
+      "step": 4093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6426731430285501e-06,
+      "loss": 3.7708,
+      "step": 4094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6292177747266324e-06,
+      "loss": 4.0313,
+      "step": 4095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6158172880941014e-06,
+      "loss": 3.872,
+      "step": 4096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6024716906071879e-06,
+      "loss": 4.1316,
+      "step": 4097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5891809897114807e-06,
+      "loss": 3.8255,
+      "step": 4098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5759451928219481e-06,
+      "loss": 3.9716,
+      "step": 4099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5627643073228948e-06,
+      "loss": 4.3464,
+      "step": 4100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5496383405680604e-06,
+      "loss": 3.4192,
+      "step": 4101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5365672998804758e-06,
+      "loss": 4.2021,
+      "step": 4102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5235511925525747e-06,
+      "loss": 4.699,
+      "step": 4103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5105900258460926e-06,
+      "loss": 4.2515,
+      "step": 4104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4976838069921784e-06,
+      "loss": 4.1041,
+      "step": 4105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4848325431912835e-06,
+      "loss": 4.2,
+      "step": 4106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4720362416131838e-06,
+      "loss": 3.8409,
+      "step": 4107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4592949093970465e-06,
+      "loss": 4.0073,
+      "step": 4108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4466085536513186e-06,
+      "loss": 4.1262,
+      "step": 4109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4339771814538162e-06,
+      "loss": 3.6594,
+      "step": 4110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4214007998516354e-06,
+      "loss": 4.1738,
+      "step": 4111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4088794158612417e-06,
+      "loss": 4.2119,
+      "step": 4112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3964130364683804e-06,
+      "loss": 3.5537,
+      "step": 4113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.384001668628121e-06,
+      "loss": 3.7048,
+      "step": 4114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.371645319264847e-06,
+      "loss": 4.3528,
+      "step": 4115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3593439952722443e-06,
+      "loss": 3.9579,
+      "step": 4116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3470977035132893e-06,
+      "loss": 3.9203,
+      "step": 4117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3349064508202724e-06,
+      "loss": 3.9715,
+      "step": 4118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3227702439947642e-06,
+      "loss": 4.4836,
+      "step": 4119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3106890898076264e-06,
+      "loss": 4.0788,
+      "step": 4120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2986629949990226e-06,
+      "loss": 4.1241,
+      "step": 4121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.286691966278375e-06,
+      "loss": 4.5059,
+      "step": 4122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2747760103243967e-06,
+      "loss": 3.9549,
+      "step": 4123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2629151337850697e-06,
+      "loss": 4.1802,
+      "step": 4124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2511093432776566e-06,
+      "loss": 3.8792,
+      "step": 4125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2393586453886886e-06,
+      "loss": 3.8607,
+      "step": 4126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2276630466739326e-06,
+      "loss": 4.0811,
+      "step": 4127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.216022553658458e-06,
+      "loss": 3.8332,
+      "step": 4128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2044371728365478e-06,
+      "loss": 3.848,
+      "step": 4129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.192906910671765e-06,
+      "loss": 4.2221,
+      "step": 4130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1814317735969193e-06,
+      "loss": 4.125,
+      "step": 4131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1700117680140677e-06,
+      "loss": 3.974,
+      "step": 4132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.15864690029448e-06,
+      "loss": 3.831,
+      "step": 4133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1473371767787066e-06,
+      "loss": 4.3387,
+      "step": 4134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1360826037765004e-06,
+      "loss": 3.6967,
+      "step": 4135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1248831875668608e-06,
+      "loss": 3.9196,
+      "step": 4136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1137389343980121e-06,
+      "loss": 3.8577,
+      "step": 4137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1026498504874027e-06,
+      "loss": 4.3796,
+      "step": 4138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0916159420217066e-06,
+      "loss": 4.1116,
+      "step": 4139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.080637215156799e-06,
+      "loss": 3.5946,
+      "step": 4140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0697136760177917e-06,
+      "loss": 3.9191,
+      "step": 4141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0588453306989765e-06,
+      "loss": 3.6968,
+      "step": 4142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.048032185263892e-06,
+      "loss": 4.1599,
+      "step": 4143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0372742457452455e-06,
+      "loss": 4.0556,
+      "step": 4144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0265715181449587e-06,
+      "loss": 3.8728,
+      "step": 4145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0159240084341547e-06,
+      "loss": 3.6186,
+      "step": 4146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.005331722553149e-06,
+      "loss": 4.2619,
+      "step": 4147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.94794666411436e-07,
+      "loss": 5.3816,
+      "step": 4148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.843128458877028e-07,
+      "loss": 4.3637,
+      "step": 4149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.738862668298266e-07,
+      "loss": 4.131,
+      "step": 4150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.635149350548655e-07,
+      "loss": 3.8221,
+      "step": 4151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.531988563490468e-07,
+      "loss": 3.9845,
+      "step": 4152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.429380364677553e-07,
+      "loss": 4.5498,
+      "step": 4153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.327324811356008e-07,
+      "loss": 3.8708,
+      "step": 4154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.225821960462955e-07,
+      "loss": 4.0071,
+      "step": 4155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.124871868627649e-07,
+      "loss": 3.9247,
+      "step": 4156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.024474592170706e-07,
+      "loss": 4.1048,
+      "step": 4157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.924630187104322e-07,
+      "loss": 4.1767,
+      "step": 4158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.825338709132492e-07,
+      "loss": 3.9678,
+      "step": 4159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.72660021365046e-07,
+      "loss": 4.3225,
+      "step": 4160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.62841475574494e-07,
+      "loss": 3.6839,
+      "step": 4161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.530782390194337e-07,
+      "loss": 3.9425,
+      "step": 4162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.433703171468299e-07,
+      "loss": 4.0268,
+      "step": 4163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.337177153727838e-07,
+      "loss": 4.1127,
+      "step": 4164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.241204390825652e-07,
+      "loss": 4.1487,
+      "step": 4165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.145784936305246e-07,
+      "loss": 4.1458,
+      "step": 4166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.050918843402033e-07,
+      "loss": 3.5938,
+      "step": 4167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.956606165042235e-07,
+      "loss": 4.0346,
+      "step": 4168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.862846953843427e-07,
+      "loss": 3.8709,
+      "step": 4169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.769641262114436e-07,
+      "loss": 3.6368,
+      "step": 4170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.67698914185555e-07,
+      "loss": 3.677,
+      "step": 4171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.584890644757647e-07,
+      "loss": 3.7856,
+      "step": 4172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.49334582220329e-07,
+      "loss": 4.4278,
+      "step": 4173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.402354725265737e-07,
+      "loss": 3.9954,
+      "step": 4174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.3119174047096e-07,
+      "loss": 4.1816,
+      "step": 4175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.2220339109903e-07,
+      "loss": 4.0051,
+      "step": 4176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.132704294254499e-07,
+      "loss": 4.466,
+      "step": 4177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.043928604339778e-07,
+      "loss": 3.8495,
+      "step": 4178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.955706890774516e-07,
+      "loss": 3.9288,
+      "step": 4179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.86803920277812e-07,
+      "loss": 3.6505,
+      "step": 4180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.780925589261133e-07,
+      "loss": 3.9926,
+      "step": 4181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.694366098824679e-07,
+      "loss": 3.7113,
+      "step": 4182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.608360779760791e-07,
+      "loss": 3.6466,
+      "step": 4183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.522909680052536e-07,
+      "loss": 4.1435,
+      "step": 4184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.438012847373332e-07,
+      "loss": 3.9839,
+      "step": 4185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.353670329087847e-07,
+      "loss": 3.9343,
+      "step": 4186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.269882172251329e-07,
+      "loss": 3.7867,
+      "step": 4187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.186648423609609e-07,
+      "loss": 3.8957,
+      "step": 4188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.103969129599319e-07,
+      "loss": 3.9387,
+      "step": 4189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.021844336347781e-07,
+      "loss": 3.9348,
+      "step": 4190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.940274089672904e-07,
+      "loss": 4.1984,
+      "step": 4191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.859258435083281e-07,
+      "loss": 4.0337,
+      "step": 4192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.778797417777982e-07,
+      "loss": 3.6829,
+      "step": 4193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.698891082646651e-07,
+      "loss": 4.1737,
+      "step": 4194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.619539474269742e-07,
+      "loss": 4.213,
+      "step": 4195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.540742636917839e-07,
+      "loss": 3.8979,
+      "step": 4196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.462500614552224e-07,
+      "loss": 3.6688,
+      "step": 4197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.384813450824644e-07,
+      "loss": 3.8746,
+      "step": 4198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.307681189077097e-07,
+      "loss": 4.0593,
+      "step": 4199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.23110387234238e-07,
+      "loss": 3.719,
+      "step": 4200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.155081543343321e-07,
+      "loss": 4.3925,
+      "step": 4201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.079614244493324e-07,
+      "loss": 4.6505,
+      "step": 4202
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.004702017895934e-07,
+      "loss": 4.246,
+      "step": 4203
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.930344905345274e-07,
+      "loss": 3.8903,
+      "step": 4204
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.856542948325493e-07,
+      "loss": 3.8533,
+      "step": 4205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.783296188011321e-07,
+      "loss": 3.6736,
+      "step": 4206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.710604665267515e-07,
+      "loss": 3.6982,
+      "step": 4207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6384684206491893e-07,
+      "loss": 4.1425,
+      "step": 4208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.566887494401373e-07,
+      "loss": 4.1835,
+      "step": 4209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4958619264597877e-07,
+      "loss": 4.121,
+      "step": 4210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.425391756449959e-07,
+      "loss": 3.9316,
+      "step": 4211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.355477023687549e-07,
+      "loss": 4.0233,
+      "step": 4212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2861177671784656e-07,
+      "loss": 4.0349,
+      "step": 4213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.217314025618646e-07,
+      "loss": 4.0276,
+      "step": 4214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1490658373941613e-07,
+      "loss": 3.8984,
+      "step": 4215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.08137324058111e-07,
+      "loss": 3.881,
+      "step": 4216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.014236272945504e-07,
+      "loss": 3.9139,
+      "step": 4217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.947654971943604e-07,
+      "loss": 3.6345,
+      "step": 4218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.881629374721474e-07,
+      "loss": 3.8672,
+      "step": 4219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8161595181153144e-07,
+      "loss": 4.4337,
+      "step": 4220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7512454386510186e-07,
+      "loss": 3.9586,
+      "step": 4221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6868871725445066e-07,
+      "loss": 4.2405,
+      "step": 4222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6230847557018335e-07,
+      "loss": 3.9628,
+      "step": 4223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.559838223718637e-07,
+      "loss": 3.7284,
+      "step": 4224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4971476118805803e-07,
+      "loss": 3.9393,
+      "step": 4225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4350129551632414e-07,
+      "loss": 4.2659,
+      "step": 4226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.373434288231669e-07,
+      "loss": 4.1864,
+      "step": 4227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.312411645441271e-07,
+      "loss": 4.8755,
+      "step": 4228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.251945060836925e-07,
+      "loss": 4.2149,
+      "step": 4229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1920345681530906e-07,
+      "loss": 3.9948,
+      "step": 4230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.132680200814364e-07,
+      "loss": 3.9539,
+      "step": 4231
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.073881991935035e-07,
+      "loss": 4.0165,
+      "step": 4232
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0156399743187516e-07,
+      "loss": 3.9424,
+      "step": 4233
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.957954180459188e-07,
+      "loss": 3.7949,
+      "step": 4234
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.900824642539601e-07,
+      "loss": 4.1354,
+      "step": 4235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8442513924329397e-07,
+      "loss": 3.6415,
+      "step": 4236
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.788234461701733e-07,
+      "loss": 4.2871,
+      "step": 4237
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7327738815982053e-07,
+      "loss": 3.729,
+      "step": 4238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.67786968306416e-07,
+      "loss": 4.1122,
+      "step": 4239
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.623521896730985e-07,
+      "loss": 4.0779,
+      "step": 4240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5697305529197577e-07,
+      "loss": 3.9288,
+      "step": 4241
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5164956816406957e-07,
+      "loss": 4.2546,
+      "step": 4242
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.463817312594152e-07,
+      "loss": 4.2102,
+      "step": 4243
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4116954751696176e-07,
+      "loss": 4.177,
+      "step": 4244
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3601301984461644e-07,
+      "loss": 4.001,
+      "step": 4245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3091215111924468e-07,
+      "loss": 3.4927,
+      "step": 4246
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.258669441866479e-07,
+      "loss": 3.6639,
+      "step": 4247
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2087740186157446e-07,
+      "loss": 3.8811,
+      "step": 4248
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.15943526927731e-07,
+      "loss": 3.7198,
+      "step": 4249
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.11065322137749e-07,
+      "loss": 3.7703,
+      "step": 4250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0624279021321803e-07,
+      "loss": 3.7148,
+      "step": 4251
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0147593384464147e-07,
+      "loss": 3.9517,
+      "step": 4252
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967647556915031e-07,
+      "loss": 3.6795,
+      "step": 4253
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9210925838216708e-07,
+      "loss": 3.6879,
+      "step": 4254
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.875094445139891e-07,
+      "loss": 4.1861,
+      "step": 4255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8296531665322747e-07,
+      "loss": 3.9919,
+      "step": 4256
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7847687733507646e-07,
+      "loss": 3.8255,
+      "step": 4257
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.740441290636663e-07,
+      "loss": 4.0616,
+      "step": 4258
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.696670743120632e-07,
+      "loss": 4.0775,
+      "step": 4259
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6534571552222488e-07,
+      "loss": 4.6739,
+      "step": 4260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6108005510510061e-07,
+      "loss": 4.4516,
+      "step": 4261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568700954404978e-07,
+      "loss": 4.0136,
+      "step": 4262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.527158388772043e-07,
+      "loss": 3.923,
+      "step": 4263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.486172877328884e-07,
+      "loss": 4.1051,
+      "step": 4264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4457444429416544e-07,
+      "loss": 4.0961,
+      "step": 4265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4058731081655342e-07,
+      "loss": 3.6802,
+      "step": 4266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3665588952451737e-07,
+      "loss": 3.6721,
+      "step": 4267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3278018261140278e-07,
+      "loss": 4.3967,
+      "step": 4268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2896019223949117e-07,
+      "loss": 5.1238,
+      "step": 4269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2519592053999995e-07,
+      "loss": 3.7387,
+      "step": 4270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2148736961301587e-07,
+      "loss": 3.9526,
+      "step": 4271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1783454152757279e-07,
+      "loss": 4.0276,
+      "step": 4272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1423743832160715e-07,
+      "loss": 3.6236,
+      "step": 4273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1069606200195814e-07,
+      "loss": 3.9167,
+      "step": 4274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0721041454438974e-07,
+      "loss": 3.8494,
+      "step": 4275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0378049789356859e-07,
+      "loss": 3.9342,
+      "step": 4276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0040631396305289e-07,
+      "loss": 4.0253,
+      "step": 4277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.708786463533681e-08,
+      "loss": 3.795,
+      "step": 4278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.382515176179496e-08,
+      "loss": 3.65,
+      "step": 4279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.06181771627046e-08,
+      "loss": 4.4679,
+      "step": 4280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.746694262727894e-08,
+      "loss": 3.8505,
+      "step": 4281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.437144991360057e-08,
+      "loss": 3.9947,
+      "step": 4282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.13317007486547e-08,
+      "loss": 4.1149,
+      "step": 4283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.834769682835141e-08,
+      "loss": 3.8249,
+      "step": 4284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.541943981748122e-08,
+      "loss": 4.1297,
+      "step": 4285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.254693134972623e-08,
+      "loss": 3.7661,
+      "step": 4286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.973017302769335e-08,
+      "loss": 3.7804,
+      "step": 4287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.696916642285889e-08,
+      "loss": 3.8065,
+      "step": 4288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.426391307561286e-08,
+      "loss": 3.5878,
+      "step": 4289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.161441449522576e-08,
+      "loss": 4.0024,
+      "step": 4290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.902067215987073e-08,
+      "loss": 4.0726,
+      "step": 4291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6482687516612455e-08,
+      "loss": 4.1054,
+      "step": 4292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.400046198140718e-08,
+      "loss": 4.4061,
+      "step": 4293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.15739969391027e-08,
+      "loss": 4.1832,
+      "step": 4294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.920329374344945e-08,
+      "loss": 4.0786,
+      "step": 4295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6888353717067235e-08,
+      "loss": 3.9602,
+      "step": 4296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.462917815147849e-08,
+      "loss": 3.8271,
+      "step": 4297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.242576830709721e-08,
+      "loss": 4.0339,
+      "step": 4298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.027812541320675e-08,
+      "loss": 4.4231,
+      "step": 4299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8186250668004185e-08,
+      "loss": 4.1084,
+      "step": 4300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.615014523854487e-08,
+      "loss": 3.5772,
+      "step": 4301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.416981026079791e-08,
+      "loss": 4.8922,
+      "step": 4302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2245246839590625e-08,
+      "loss": 4.2483,
+      "step": 4303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.037645604866413e-08,
+      "loss": 3.8204,
+      "step": 4304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8563438930617748e-08,
+      "loss": 4.4837,
+      "step": 4305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6806196496953484e-08,
+      "loss": 3.9401,
+      "step": 4306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5104729728042674e-08,
+      "loss": 3.9361,
+      "step": 4307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.345903957313711e-08,
+      "loss": 4.1964,
+      "step": 4308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.186912695039123e-08,
+      "loss": 4.3097,
+      "step": 4309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0334992746817717e-08,
+      "loss": 4.1687,
+      "step": 4310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.885663781832081e-08,
+      "loss": 4.1234,
+      "step": 4311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7434062989685196e-08,
+      "loss": 3.6765,
+      "step": 4312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6067269054576008e-08,
+      "loss": 4.1824,
+      "step": 4313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4756256775538824e-08,
+      "loss": 3.9652,
+      "step": 4314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3501026883988577e-08,
+      "loss": 3.6689,
+      "step": 4315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2301580080220643e-08,
+      "loss": 4.6788,
+      "step": 4316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1157917033433052e-08,
+      "loss": 4.1787,
+      "step": 4317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0070038381670977e-08,
+      "loss": 4.3405,
+      "step": 4318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.037944731860038e-09,
+      "loss": 4.4309,
+      "step": 4319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.06163665982851e-09,
+      "loss": 4.086,
+      "step": 4320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.141114710251806e-09,
+      "loss": 4.1978,
+      "step": 4321
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.276379396696896e-09,
+      "loss": 3.9263,
+      "step": 4322
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.467431201611195e-09,
+      "loss": 3.9264,
+      "step": 4323
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.714270576311463e-09,
+      "loss": 4.0075,
+      "step": 4324
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0168979409949126e-09,
+      "loss": 3.979,
+      "step": 4325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.375313684705894e-09,
+      "loss": 4.5507,
+      "step": 4326
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.789518165413618e-09,
+      "loss": 3.9266,
+      "step": 4327
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2595117099344366e-09,
+      "loss": 4.203,
+      "step": 4328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.785294613942945e-09,
+      "loss": 3.7031,
+      "step": 4329
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3668671420274947e-09,
+      "loss": 3.9402,
+      "step": 4330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0042295276346813e-09,
+      "loss": 3.7111,
+      "step": 4331
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.97381973058242e-10,
+      "loss": 4.1454,
+      "step": 4332
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.463246495167717e-10,
+      "loss": 4.0625,
+      "step": 4333
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5105769705380253e-10,
+      "loss": 4.3776,
+      "step": 4334
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1158122462662235e-10,
+      "loss": 4.5477,
+      "step": 4335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7895310050762845e-11,
+      "loss": 4.2698,
+      "step": 4336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 3.6413,
+      "step": 4337
+    },
+    {
+      "epoch": 1.0,
+      "step": 4337,
+      "total_flos": 179355569651712.0,
+      "train_loss": 3.331768640565905,
+      "train_runtime": 43638.2071,
+      "train_samples_per_second": 1.59,
+      "train_steps_per_second": 0.099
+    }
+  ],
+  "max_steps": 4337,
+  "num_train_epochs": 1,
+  "total_flos": 179355569651712.0,
+  "trial_name": null,
+  "trial_params": null
+}